system

Abstract

Provide a system. A means for receiving conference audio data; a speech recognition means for converting received speech data into text data; a natural language processing means for creating minutes based on the converted text data; A database management means for storing, retrieving and referencing the minutes prepared; A means to compare the new text data with the minutes and materials of the archived related meetings and to examine the content. means for generating and transmitting the final minutes to the user; A system including:

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] When preparing meeting minutes, there are many cases where minutes are required without sufficient prior information sharing about the background and related information of the meeting. In such cases, it takes a lot of time to research and confirm the necessary information, making the process extremely inefficient. Furthermore, there is a risk of information omissions or misunderstandings when preparing minutes manually, so accurate and timely minutes are required. [Means for solving the problem]

[0005] The present invention includes a means for receiving conference audio data and converting the audio data into text data using speech recognition technology. It also uses natural language processing means for creating minutes based on the converted text data, and includes database management means for saving, retrieving, and referencing the minutes and materials of related conferences. This database management means stores the minutes and materials of past related conferences, allowing them to be quickly referenced when creating minutes for new conferences. Furthermore, natural language processing technology is used to compare old and new text data and carefully examine the content, automatically generating accurate and easy-to-understand minutes. The final minutes are sent to the user, who can easily review and edit them. This enables more efficient and accurate minutes creation than conventional manual minutes creation.

[0006] "Conference audio data" refers to audio files or digital data in streaming format recorded during a conference.

[0007] "Speech recognition means" refers to software or devices for converting conference voice data into text data.

[0008] "Natural language processing means" refers to technologies and algorithms for extracting necessary information from text data and analyzing its meaning.

[0009] "Database management means" refers to systems and software for storing, retrieving, and referencing minutes and materials from related meetings.

[0010] "Text data" is character information obtained from voice data by voice recognition means.

[0011] A "minutes" is a document that records in writing the contents of a meeting and the matters decided upon.

[0012] "User" means a person or organization that uploads conference audio data or reviews and edits the generated minutes.

[0013] A "terminal" is a device that a user uses to upload audio data and check the minutes.

[0014] "Means for comparison and scrutiny of content" refers to a technique for comparing new text data with minutes and materials from related meetings to confirm the content and extract the necessary information.

[0015] The "means for generating the final minutes and sending them to the user" refers to the system or process for creating minutes based on the scrutinized text data and delivering them to the user. [Brief explanation of the drawings]

[0016] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11]FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0017] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0018] First, the terms used in the following description will be explained.

[0019] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0020] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0021] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0022] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0023] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0024] [First embodiment]

[0025] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0026] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0027] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0028] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0029] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0030] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0031] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0032] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0033] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0034] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0035] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0036] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0037] The present invention is a system that automatically converts conference audio data into minutes and quickly provides them. This system is operated in cooperation between a server, terminals, and users.

[0038] Server

[0039] 1. Receiving audio data

[0040] The server receives the conference audio data sent from the terminals. The audio data is usually sent via an HTTP request, and the server receives the request and stores the audio file in a temporary storage location.

[0041] 2. Calling the voice recognition engine

[0042] The received voice data is passed to the speech recognition engine, which converts the data into text data and returns the converted text to the server, which stores the text data in its memory.

[0043] 3. Obtaining relevant data

[0044] The server uses database management means to retrieve minutes and materials from past relevant meetings using SQL queries, and stores the retrieved relevant data in memory.

[0045] 4. Use of natural language processing engines

[0046] The server passes the text data received from the speech recognition engine and related data retrieved from the database to the natural language processing engine, which then analyzes the meaning of the data and extracts the necessary information.

[0047] 5. Generate final transcripts

[0048] Based on the scrutinized data, the server generates the final transcript, which is again encoded in JSON format and sent back to the device.

[0049] Terminal (Client)

[0050] 1. Providing an upload interface

[0051] The terminal provides a user with an interface for uploading conference audio data. This interface is implemented as a GUI including a file selection and upload button.

[0052] 2. Sending audio data

[0053] When a user selects an audio file and presses the upload button, the device sends this audio data to the server as an HTTP POST request.

[0054] 3. Receiving and displaying minutes

[0055] Once the final minutes are returned from the server, the device receives this data and displays it to the user. The minutes data is encoded in JSON format, so the device parses it and displays it on the screen.

[0056] User

[0057] 1. Uploading audio data

[0058] Users upload conference audio data through the interface provided by their devices. They select the appropriate audio file in the file selection dialog and click the upload button.

[0059] 2. Review and edit the minutes

[0060] Once the final minutes are displayed on the device, the user can review them and make edits as needed. Editing is done directly in the text area, and once edits are complete, the changes are saved by pressing the save button.

[0061] Specific examples

[0062] For example, if a user uploads an audio file for a "Project Status Meeting on October 5, 2023," the audio data is sent to the server. The server then converts the audio into text using a speech recognition engine and retrieves minutes of related past meetings from the database. It then scrutinizes the old and new data using natural language processing, generates the final minutes, and sends them back to the device. The user can then review the received minutes and make any necessary corrections, completing the minutes quickly and accurately.

[0063] This system will greatly improve the efficiency of creating meeting minutes, increasing accuracy and convenience.

[0064] The processing flow will be explained below.

[0065] Server

[0066] Step 1:

[0067] The server receives the conference audio data sent from the terminal via an HTTP POST request, and the received audio file is saved in a temporary directory.

[0068] Step 2:

[0069] The server passes the saved audio file to a speech recognition engine and converts the audio data into text data. At this time, it calls the speech recognition API and passes the file path as an argument.

[0070] Step 3:

[0071] The server receives the text data returned by the speech recognition engine and stores it in memory. The returned data is usually provided in JSON format.

[0072] Step 4:

[0073] The server accesses the database to retrieve relevant meeting minutes and materials using SQL queries, and stores the retrieved data in memory.

[0074] Step 5:

[0075] The server calls a natural language processing engine to compare and examine the text data obtained through speech recognition with related data retrieved from the database, thereby extracting important information.

[0076] Step 6:

[0077] The server generates the final minutes based on the scrutinized data, which are generated in text format and then encoded again in JSON format.

[0078] Step 7:

[0079] The server sends the generated minutes to the terminal as an HTTP response.

[0080] Terminal (Client)

[0081] Step 1:

[0082] The terminal provides the user with an interface for uploading audio data, which is designed as a GUI containing a file selection dialog and an upload button.

[0083] Step 2:

[0084] When the user selects an audio file and presses the upload button, the device sends the audio data to the server via an HTTP POST request.

[0085] Step 3:

[0086] When the final minutes data is returned from the server, the terminal receives and parses this JSON format data.

[0087] Step 4:

[0088] The parsed minutes are displayed on the device screen as nicely formatted text so that users can intuitively check the contents.

[0089] User

[0090] Step 1:

[0091] After the conference, the user uploads the audio data through the interface provided by the device by selecting the correct audio file in the file selection dialog and clicking the upload button.

[0092] Step 2:

[0093] Once the upload is complete, the user can check the minutes sent back from the server on their device. The displayed minutes reflect the contents of past meetings, making it easy to confirm important points.

[0094] Step 3:

[0095] The user can check the displayed minutes, make corrections or add comments as necessary, and when the corrections are complete, press the save button to confirm the changes.

[0096] The above processing flow makes it possible to efficiently create meeting minutes and provide accurate and prompt minutes.

[0097] Example 1

[0098] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0099] The process of converting meeting audio data into minutes quickly and accurately is time-consuming and labor-intensive when done manually, making it inefficient. Furthermore, manual minutes are prone to errors and often lack sufficient linkage to past meeting information. This creates a need for improved accuracy in minutes and a more efficient creation process. It is also necessary to provide an interface that makes it easy to upload audio data and edit and review minutes.

[0100] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0101] In this invention, the server includes: means for receiving conference audio data; speech recognition means for converting the received audio data into text data; natural language processing means for creating minutes based on the converted text data; database management means for saving, retrieving, and referencing the created minutes; means for comparing the new text data with stored minutes and materials from related meetings and examining the content; means for generating and transmitting final minutes to users; and means for using a client-server architecture that smoothly transmits and receives audio data, generates, and displays minutes. This improves the efficiency of creating meeting minutes and enables the creation of highly accurate minutes. Users can also easily upload audio files and review and edit the minutes.

[0102] "Conference audio data" refers to digital audio files that record what is said during a conference or meeting.

[0103] "Speech recognition means" refers to technology or equipment for converting voice data into text data. Specifically, it uses a speech recognition engine or API.

[0104] "Natural language processing means" refers to technology or equipment that analyzes text data, understands its meaning and content, and extracts appropriate information. Specifically, it uses natural language processing engines and APIs.

[0105] "Database management means" refers to the technology or device used to store, retrieve, and reference data. Specifically, a database management system (DBMS) is used.

[0106] "Client-server architecture" is a system design model that allows communication between a server that provides data and resources and a client that uses those data and resources.

[0107] A "user" is a person who operates the system and uploads audio data and checks and edits minutes.

[0108] A "terminal" is a device that allows a user to access and operate the system. Specifically, it includes computers, smartphones, etc.

[0109] A "minutes" is a document that records what was said and what was decided at a conference or meeting.

[0110] An "audio file upload interface" is a screen or function that a user uses to submit audio data to the system.

[0111] "Text data" is character information converted by a voice recognition means.

[0112] The "JSON format" is a lightweight data interchange format for storing and exchanging data.

[0113] MODE FOR CARRYING OUT THE INVENTION

[0114] The present invention is a system for automatically converting conference audio data into minutes and providing them promptly. A specific embodiment of this system will now be described in detail.

[0115] server

[0116] The server uses the following main hardware and software to perform the following series of data processing and calculations:

[0117] 1. Receiving audio data

[0118] The server receives the conference audio data sent from the device via an HTTP request, processes the request using a Python framework such as Flask, and temporarily stores the audio file on the server's local disk.

[0119] 2. Calling the voice recognition engine

[0120] The server passes the saved voice data to a speech recognition engine, which converts the voice data into text using the Google® Cloud Speech-to-Text API. The converted text is stored in the server's memory.

[0121] 3. Obtaining relevant data

[0122] The server retrieves minutes and materials from past relevant meetings from a MySQL® database using SQL queries, including filter conditions such as meeting dates and project names, and stores the retrieved data in memory.

[0123] 4. Use of natural language processing engines

[0124] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to the IBM Watson® Natural Language Understanding API, which analyzes the meaning of the data, executes a process to extract the necessary information, and stores the extracted results in memory.

[0125] 5. Generate final transcripts

[0126] The server generates the final minutes based on the scrutinized data. It uses the Python Jinja2 template engine to generate the minutes in JSON format and sends it back to the terminal as an HTTP response.

[0127] Terminal

[0128] The terminal provides an interface for users to upload audio data and review / edit the generated minutes. The following hardware and software are used to perform the operation:

[0129] 1. Providing an upload interface

[0130] The terminal provides the user with a GUI for uploading conference audio data. An HTML form is used to implement a file selection dialog and an upload button, and JavaScript (registered trademark) is used to automatically start uploading after file selection.

[0131] 2. Sending audio data

[0132] When a user selects an audio file and presses the upload button, the device uses JavaScript's Fetch API to send this audio data to the server as an HTTP POST request, which includes the binary data of the selected audio file.

[0133] 3. Receiving and displaying minutes

[0134] Once the final minutes are returned from the server, the device receives this data, retrieves the Fetch API response, parses the response data as JSON, and displays it in the user interface using HTML and JavaScript.

[0135] User

[0136] Users use their devices to upload the meeting audio data and check and edit the generated minutes.

[0137] 1. Uploading audio data

[0138] The user uploads the conference audio data using the interface provided by the terminal: open the file selection dialog, select the audio file to upload, and click the upload button.

[0139] 2. Review and edit the minutes

[0140] The user checks the final minutes displayed on the device and edits them as necessary. They edit the content directly in the displayed text area and press the save button when they are done. The edits are captured by a JavaScript event listener and saved in local storage.

[0141] Specific examples

[0142] For example, if a user uploads an audio file for the "Project Status Meeting on October 5, 2023," the following process takes place: The user selects the audio file from the device's GUI and presses the upload button. The server receives an HTTP POST request using the Flask framework and saves the audio file to its local disk. It then sends a request to a speech recognition engine (Google Cloud Speech-to-Text API) and receives the converted text. Next, it retrieves past meeting minutes data from a MySQL database using filter criteria such as "project name" and "meeting date." This data is passed to the IBM Watson Natural Language Understanding API, where the content is examined and analyzed. Finally, the Jinja2 template engine is used to generate the final meeting minutes in JSON format and return them as an HTTP response. The device receives this response data and displays it in the user interface using HTML and JavaScript. The user reviews the displayed minutes and makes any necessary edits in the text area.

[0143] Prompt Sentence Examples

[0144] An example of a prompt is, "Convert the audio file of the project status meeting on October 5, 2023 into text, and display the minutes that have been reviewed and generated in conjunction with past related meeting minutes. The APIs used are speech recognition API and natural language processing API."

[0145] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0146] Step 1:

[0147] The server receives the conference audio data sent from the terminal. Specifically, the terminal creates an HTTP POST request and sends the binary data of the audio file to the server. The server processes this request using a Python framework such as Flask and temporarily saves the received audio file on the local disk. The input is the audio file sent from the terminal, and the output is the audio file saved on the server.

[0148] Step 2:

[0149] The server passes the saved audio file to the speech recognition engine. Specifically, it creates an API request to use the Google Cloud Speech-to-Text API to convert the audio data to text. The server uses the path to the audio file and the API key as input and receives the converted text data as a response. The output is the text data converted from the audio file.

[0150] Step 3:

[0151] The server retrieves minutes and materials from past relevant meetings from a MySQL database. Specifically, it executes an SQL query that includes filter conditions such as the meeting date and project name. The input is the filter conditions for searching the database, and the output is the retrieved meeting minutes and materials data. This data is stored in memory.

[0152] Step 4:

[0153] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to the natural language processing engine. Specifically, it uses the IBM Watson Natural Language Understanding API to analyze the meaning of the data and extract the necessary information. The input is the speech-recognized text data and related data, and the output is the scrutinized text data.

[0154] Step 5:

[0155] The server generates the final minutes based on the parsed data. Specifically, it uses the Python Jinja2 template engine to generate minutes in JSON format. The input is the parsed text data and template information, and the output is the generated minutes data. This minutes data is returned to the terminal as an HTTP response.

[0156] Step 6:

[0157] The terminal receives the final minutes data returned from the server. Specifically, it receives the HTTP response using the Fetch API and parses the response data as JSON. The input is the JSON minutes data returned from the server, and the output is the parsed minutes data.

[0158] Step 7:

[0159] The terminal displays the parsed minutes data on the user interface. Specifically, it uses HTML and JavaScript to insert the parsed data into a text display area. The input is the parsed minutes data, and the output is the minutes displayed on the screen.

[0160] Step 8:

[0161] The user checks the minutes displayed on the terminal and makes corrections as necessary. Specifically, the user edits the content directly in the displayed text area and clicks the save button when finished. The input is the user's edited content, and the output is the corrected minutes.

[0162] Step 9:

[0163] The terminal sends the edited or revised content by the user to the server and updates the final minutes. Specifically, JavaScript is used to collect the revisions and send them to the server as an HTTP POST request. The input is the minutes data edited by the user, and the output is the minutes data updated on the server. This series of processes allows meeting minutes to be created quickly and accurately.

[0164] (Application example 1)

[0165] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0166] Traditionally, work instructions and reports within factories have often been written by hand or orally, which has led to the problem of time-consuming communication and sharing of information. There is also a high risk of missed records and inaccurate reports. This can lead to reduced work efficiency within the factory and affect productivity. Furthermore, in order to provide quick and accurate work reports, workers and engineers have to spend time and effort to keep records, which increases the workload.

[0167] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0168] In this invention, the server includes means for receiving conference voice data, speech recognition means for converting the received voice data into text data, natural language processing means for creating minutes based on the converted text data, database management means for saving, retrieving, and referencing the created minutes, means for comparing the saved minutes and materials of related meetings with the new text data and examining the content, means for generating final minutes and sending them to the user, means for converting voice recordings of work instructions and reports within the factory into text, and means for uploading the generated minutes to a central system within the factory, thereby enabling quick and accurate recording of work instructions and reports within the factory.

[0169] "Conference voice data" refers to digital data of voice generated during a conference or meeting.

[0170] "Means for receiving audio data" refers to an interface or module for receiving and processing audio data from another device or system.

[0171] "Speech recognition means" refers to technology or equipment that analyzes voice data and converts it into text data.

[0172] "Natural language processing means" is a technology that analyzes the meaning of text data and organizes and extracts data.

[0173] A "database management tool" is a system or software used to efficiently store, retrieve, and access data.

[0174] "Means for examining the content" refers to techniques and methods for analyzing the acquired data in detail and extracting the necessary information.

[0175] "Means for transmitting to users" refers to methods or systems for delivering the final generated data or information to users.

[0176] "Means for converting audio recordings of work instructions and reports within a factory into text" refers to technology that records the audio instructions and reports of factory workers and engineers and converts them into text data.

[0177] "Means for uploading the generated minutes to a central system within the factory" refers to an interface or system for transferring the generated minutes to a central system installed within the factory for storage and management.

[0178] The present invention aims to improve the efficiency of labor in factories and the accuracy of reports. This system is operated through collaboration between a server, a terminal, and a user.

[0179] Server

[0180] The server plays a central role in receiving voice data of work instructions and reports within the factory, and processing and storing that data. Specifically, it performs the following processes:

[0181] Receiving audio data

[0182] The server receives the audio data sent from the device via an HTTP request and temporarily stores the received audio file.

[0183] Calling the speech recognition engine

[0184] The voice data is passed to a speech recognition engine and converted into text data. The speech recognition engine used is something like the Google Cloud Speech-to-Text API.

[0185] Retrieving related data

[0186] The server retrieves reports and documents of past related work from the database, searches using SQL queries, and stores the required data in memory.

[0187] Use of natural language processing engines

[0188] The text data received from the speech recognition engine and related data retrieved from the database are passed to a natural language processing engine for detailed analysis. Natural language processing engines such as "TENSORFLOW (registered trademark)" and "spaCy" are used.

[0189] Generate and send meeting minutes

[0190] Finally, based on the scrutinized data, the server generates a transcript, which is encoded in JSON format and sent back to the device.

[0191] Terminal (Client)

[0192] The terminal is the interface for the user and is responsible for uploading audio data and displaying received minutes.

[0193] Providing an upload interface

[0194] The terminal provides an audio file upload interface, which is implemented as a GUI (Graphical User Interface) with a file selection and upload button.

[0195] Sending audio data

[0196] After selecting an audio file and clicking the upload button, the device will send the audio file to the server as an HTTP POST request.

[0197] Receiving and viewing minutes

[0198] The minutes returned from the server are received, the JSON format data is parsed, and the data is displayed on the screen, allowing the user to check the minutes and edit them as necessary.

[0199] User

[0200] The user is the end of the system, providing the audio data and checking and editing the final minutes.

[0201] Uploading audio data

[0202] Users use the terminal interface to upload audio data of factory operations and reports.

[0203] Review and edit minutes

[0204] The generated minutes can be checked and edited as necessary. For example, by uploading audio data such as "There is a problem with the right-hand conveyor belt. The motor is overheating and needs to be replaced," minutes can be created quickly and accurately.

[0205] Specific prompt examples

[0206] Generate a transcript from the following audio: "There is a problem with the right-hand conveyor belt. The motor is overheating and needs to be replaced."

[0207] In this way, this system enables efficient and accurate work instructions and reports within the factory. The server is a high-performance computer, the terminals are mobile devices with internet access, and the users are assumed to be factory workers and engineers.

[0208] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0209] Step 1: Recording audio data

[0210] Users record audio data at the factory work site. Specifically, engineers and workers use microphones to record work instructions and reports as audio. This audio data is saved on the device.

[0211] Input: User speech

[0212] Output: Audio file saved on device

[0213] Step 2: Upload your audio data

[0214] The user sends the recorded audio data to the server using the upload interface on the device, by selecting the audio file through the GUI and clicking the upload button.

[0215] Input: Audio file saved on the device

[0216] Output: Audio file sent to the server

[0217] Step 3: Receiving audio data

[0218] The server receives the audio data sent from the device via an HTTP request and temporarily stores the received audio data.

[0219] Input: Audio data sent from the device

[0220] Output: Temporarily saved audio file

[0221] Step 4: Calling the speech recognition engine

[0222] The server passes the temporarily saved audio file to a speech recognition engine and converts the audio data into text data, using a speech recognition service such as the Google Cloud Speech-to-Text API.

[0223] Input: Temporarily saved audio file

[0224] Output: Audio information converted to text data

[0225] Step 5: Retrieve related data

[0226] The server retrieves the minutes and documents of past related work from the database, specifically by using SQL queries to search and retrieve the necessary past data.

[0227] Input: SQL query

[0228] Output: Obtained past minutes and document data

[0229] Step 6: Use a natural language processing engine

[0230] The server then passes the text data obtained from the speech recognition engine and related data retrieved from the database to a natural language processing engine for further analysis, such as using TensorFlow or spaCy to analyze the text data and parse its meaning.

[0231] Input: Text data and historical data

[0232] Output: Scanned text data

[0233] Step 7: Generate and send the minutes

[0234] The server generates the final transcript based on the scrutinized data, which is then encoded in JSON format and sent to the device.

[0235] Input: vetted text data

[0236] Output: JSON formatted minutes

[0237] Step 8: Receive and view the minutes

[0238] The terminal receives the minutes in JSON format sent from the server, parses them, and displays them to the user. Specifically, it displays the contents of the minutes on the screen using a GUI.

[0239] Input: JSON formatted minutes

[0240] Output: Transcript displayed on screen

[0241] Step 9: Review and edit the minutes

[0242] The user checks the minutes displayed on the terminal and edits them as necessary. For example, if there are any clerical errors, the user corrects them and saves them.

[0243] Input: Minutes displayed on screen

[0244] Output: Edited transcript

[0245] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0246] The present invention provides a system that automatically converts conference audio data into minutes and further adjusts the content of the minutes by recognizing the emotions of users. This system is operated in cooperation between a server, terminals, and users.

[0247] Server

[0248] 1. Receiving audio data

[0249] The server receives the conference audio data sent from the terminal. The audio data is sent via an HTTP POST request, and the server receives this request and saves the audio file in a temporary directory.

[0250] 2. Calling the voice recognition engine

[0251] The server passes the saved audio file to the speech recognition engine, converts the audio data into text data, and calls the speech recognition API, passing the file path as an argument.

[0252] 3. Receiving converted data and emotion recognition

[0253] The server receives the text data returned by the speech recognition engine and stores it in memory. Based on this text data, it uses an emotion recognition engine to analyze the user's emotions. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words.

[0254] 4. Acquiring relevant data

[0255] The server uses database management means to retrieve minutes and materials from past relevant meetings using SQL queries, and stores the retrieved data in memory.

[0256] 5. Use of natural language processing engines

[0257] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to the natural language processing engine, which then analyzes the meaning of the data and extracts the necessary information.

[0258] 6. Adjusting meeting minutes based on emotional information

[0259] The server adjusts the content of the minutes it generates based on the user's emotional information obtained by the emotion recognition engine, for example by emphasizing parts of the minutes in which the user expressed strong emotions.

[0260] 7. Generate and send final minutes

[0261] Based on the scrutinized data and adjusted sentiment information, the server generates the final transcript, which is generated in text format, encoded again in JSON format, and sent to the device as an HTTP response.

[0262] Terminal (Client)

[0263] 1. Providing an upload interface

[0264] The terminal provides the user with an interface for uploading audio data, which is implemented as a GUI containing a file selection dialog and an upload button.

[0265] 2. Sending audio data

[0266] When the user selects an audio file and presses the upload button, the device sends the audio data to the server via an HTTP POST request.

[0267] 3. Receiving and displaying minutes

[0268] When the final minutes data is returned from the server, the device receives and parses the JSON format data. The parsed minutes are displayed on the device screen and formatted so that the user can intuitively check the contents.

[0269] User

[0270] 1. Uploading audio data

[0271] After the conference, the user uploads the audio data through the interface provided by the device by selecting the correct audio file in the file selection dialog and clicking the upload button.

[0272] 2. Review and edit the minutes

[0273] Once the final minutes are displayed on the device, the user can review them and make edits or add comments as needed. Edits are made directly in the text area, and once corrections are complete, the changes can be confirmed by pressing the save button.

[0274] Specific examples

[0275] For example, if a user uploads an audio file for a "Project Progress Meeting on October 5, 2023," the audio data is sent to the server. The server then converts the audio into text using a speech recognition engine and retrieves minutes of related past meetings from a database. It then scrutinizes old and new data through natural language processing, and recognizes the user's emotions during the meeting using an emotion recognition engine. For example, if there is an emphasis on a particular topic, the system adjusts the display to highlight that part based on the emotional information. The final minutes are then sent back to the user via their device, who can review them and make any necessary corrections.

[0276] This system makes the creation of meeting minutes significantly more efficient than conventional manual operations, and also makes it possible to generate more detailed minutes that reflect the user's emotions during the meeting.

[0277] The processing flow will be explained below.

[0278] Server

[0279] Step 1:

[0280] The server receives the conference audio data sent from the terminal via an HTTP POST request, and the received audio file is saved in a temporary directory.

[0281] Step 2:

[0282] The server passes the saved audio file to a speech recognition engine and converts the audio data into text data. At this time, it calls the speech recognition API and passes the file path as an argument.

[0283] Step 3:

[0284] The server receives the text data returned by the speech recognition engine and stores it in memory. The returned data is usually provided in JSON format.

[0285] Step 4:

[0286] The server passes this text data to an emotion recognition engine, which analyzes the user's emotions. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words, and stores the results in memory.

[0287] Step 5:

[0288] The server accesses the database to retrieve relevant meeting minutes and materials using SQL queries, and stores the retrieved data in memory.

[0289] Step 6:

[0290] The server invokes a natural language processing engine to compare and examine the text data obtained through speech recognition with related data retrieved from the database, thereby extracting important information.

[0291] Step 7:

[0292] The server adjusts the content of the minutes it generates based on the user's emotional information obtained by the emotion recognition engine, for example by emphasizing parts of the minutes in which the user expressed strong emotions.

[0293] Step 8:

[0294] The server generates the final transcript based on the scrutinized data and adjusted sentiment information. The transcript is generated in text format, then encoded again in JSON format and sent to the device as an HTTP response.

[0295] Terminal (Client)

[0296] Step 1:

[0297] The terminal provides the user with an interface for uploading audio data, which is designed as a GUI containing a file selection dialog and an upload button.

[0298] Step 2:

[0299] When the user selects an audio file and presses the upload button, the device sends the audio data to the server via an HTTP POST request.

[0300] Step 3:

[0301] When the final minutes data is returned from the server, the terminal receives and parses this JSON format data.

[0302] Step 4:

[0303] The parsed minutes are displayed on the device screen as nicely formatted text so that users can intuitively check the contents.

[0304] User

[0305] Step 1:

[0306] After the conference, the user uploads the audio data through the interface provided by the device by selecting the correct audio file in the file selection dialog and clicking the upload button.

[0307] Step 2:

[0308] Once the upload is complete, the user can check the minutes sent back from the server on their device. The displayed minutes reflect the contents of past meetings, making it easy to confirm important points.

[0309] Step 3:

[0310] The user can check the displayed minutes, make corrections or add comments as necessary, and when the corrections are complete, press the save button to confirm the changes.

[0311] Examples:

[0312] When a user uploads an audio file for the "Project Progress Meeting on October 5, 2023," the audio data is sent to the server. The server then converts the audio into text using a speech recognition engine and retrieves minutes of past related meetings from a database. It then scrutinizes old and new data through natural language processing, and recognizes the user's emotions during the meeting using an emotion recognition engine. For example, if there is an emphasis on a particular topic, the system adjusts the display to highlight that part based on the emotional information. The final minutes are then sent back to the user via their device, who can review them and make any necessary corrections.

[0313] Example 2

[0314] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0315] Conventional systems that simply transcribe meeting audio data have the problem of not being able to thoroughly examine the content or reflect emotions, resulting in a decline in the quality of the minutes. Creating minutes that reflect the user's emotions is also a time-consuming and difficult process to do efficiently.

[0316] The specification process by the specification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for receiving voice data, voice recognition means for converting the received voice data into text data, emotion recognition means for analyzing the user's emotion based on the converted text data, database management means for acquiring past related data, natural language processing means for analyzing the acquired related data and the text data obtained from the voice recognition means and examining the content, means for adjusting the content of the minutes based on the emotion information obtained by the emotion recognition means, and means for generating the final minutes and sending them to the user. This makes it possible to automatically and efficiently generate highly accurate minutes that reflect the user's emotions.

[0317] "Audio data" refers to a recording file of a conference that is uploaded by a user, and is data that is converted into text data by a voice recognition means.

[0318] "Speech recognition means" refers to software or an engine for analyzing received voice data and converting it into text data.

[0319] The "emotion recognition means" is software or an engine for analyzing the user's emotions based on the converted text data.

[0320] "Database management means" refers to the means for storing, retrieving, and referencing minutes and materials from past related meetings.

[0321] "Natural language processing means" refers to software or an engine for analyzing and examining the acquired association data and the text data obtained from the speech recognition means.

[0322] The "minutes adjustment means" is a means for adjusting the contents of the minutes to be generated based on the emotion information obtained by the emotion recognition means.

[0323] The "minutes generation means" is a means for generating the final minutes and transmitting them to the user.

[0324] A "terminal" is a device or software that allows a user to upload audio files and view and edit the generated minutes.

[0325] "User" refers to a person who uses this system to upload conference audio data and to review and edit the generated minutes.

[0326] The present invention provides a system that automatically converts conference audio data into minutes and adjusts the content of the minutes by recognizing the emotions of users. This system is operated in cooperation between a server, terminals, and users.

[0327] First, let's explain about terminals. A terminal is a device or software that provides a user with an interface for uploading audio data. Specifically, it is implemented as a GUI (Graphical User Interface) that includes a file selection dialog and an upload button. When a user selects an audio file and presses the upload button, the terminal sends the audio data to the server via an HTTP POST request.

[0328] Next, we will explain the server processing. First, the server receives the audio data sent from the terminal. The audio data is sent via an HTTP POST request, and the server receives this request and saves the audio file in a temporary directory. For example, it may be saved in a path such as " / tmp / meeting_audio.wav".

[0329] The server passes the saved audio file to a speech recognition engine (generally a speech recognition API is used) and converts the audio data into text data. Specifically, it calls the Google Cloud Speech-to-Text API and passes the file path as an argument. The returned text data is stored in memory.

[0330] The server then analyzes the user's emotions using an emotion recognition engine, such as IBM Watson Natural Language Understanding, to determine the user's emotions from the content and word usage of the text data. The analysis results are stored in memory.

[0331] The server then issues SQL queries to a database management system (e.g., MySQL) to retrieve minutes and documents from past related meetings. The retrieved data is also stored in memory. This provides information for comparing past minutes with new text data and examining their content.

[0332] The server then uses a natural language processing engine (such as OpenAI® GPT-3®) to analyze the text data from the speech recognition engine and related data from the database to extract meaning from the data, a process that makes the content of the meeting clearer.

[0333] The server adjusts the content of the meeting minutes it generates based on the user's emotional information obtained from the emotion recognition engine. For example, it highlights parts of the meeting minutes in which the user expressed strong emotions about a particular topic. Once the adjustments are complete, the server generates the final meeting minutes in text format, encodes them again in JSON format, and sends them to the terminal as an HTTP response.

[0334] Finally, the terminal parses the JSON data of the final minutes received from the server, formats it so that the user can intuitively check the contents, and displays it on the screen. The user can review it and make edits or add comments as needed. This series of operations allows the user to obtain efficient and detailed minutes.

[0335] As a specific example, if a user uploads an audio file for the "Project Status Meeting on October 5, 2023," the audio data is sent to the server. The server converts the audio into text using the Google Cloud Speech-to-Text API and retrieves minutes of past related meetings from a database (MySQL). It then uses OpenAI GPT-3 to scrutinize old and new data, and IBM Watson Natural Language Understanding to recognize the user's emotions during the meeting. For example, if there are emphasized opinions on a particular topic, the system adjusts the display to highlight those parts based on the emotional information. The final minutes are then sent back to the user via their device, who can review them and make any necessary corrections.

[0336] An example of a prompt sentence could be "Please upload the audio data of next week's important meeting and automatically generate the minutes."

[0337] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0338] Step 1: Upload your audio data

[0339] The user accesses the interface provided by the device and selects an audio file. When the user clicks the "Upload" button, the audio file is sent from the device to the server. The audio data is sent to the server using the HTTP POST method, and the audio file is included as part of the request.

[0340] Input: An audio file selected by the user

[0341] Output: Audio data sent from the device to the server

[0342] Specific behavior: The user selects "Project Status Meeting on October 5, 2023.wav" and presses the upload button. This file is sent to the server as an HTTP POST request.

[0343] Step 2: Receiving and storing audio data

[0344] The server receives the HTTP POST request sent from the device and temporarily stores the audio data. When temporarily storing the data, an appropriate path (e.g., " / tmp / 20231005_meeting.wav") is specified.

[0345] Input: Audio data sent from the device

[0346] Output: Audio file saved on the server

[0347] Specific operation: The server saves the received audio file to " / tmp / 20231005_meeting.wav" and records a log that the upload is complete.

[0348] Step 3: Performing speech recognition

[0349] The server passes the path of the saved audio file to a speech recognition engine (for example, Google Cloud Speech-to-Text API) to convert the audio data into text data. A request is sent to the API, and when the request is completed, the converted text data is returned as a response.

[0350] Input: Path to the audio file

[0351] Output: Text data

[0352] Specific operation: The server passes " / tmp / 20231005_meeting.wav" to the Google Cloud Speech-to-Text API and records the log "Speech recognition completed, text data received."

[0353] Step 4: Performing Emotion Recognition

[0354] The server passes the acquired text data to an emotion recognition engine (e.g., IBM Watson Natural Language Understanding) to analyze the user's emotions. Emotional information is returned as the analysis result, and the server stores this in its memory.

[0355] Input: Text data

[0356] Output: Emotional information

[0357] Specific operation: The server passes the text data to IBM Watson and records a log indicating "emotion recognition completed, analysis results received."

[0358] Step 5: Retrieve related data

[0359] The server issues SQL queries to a database management system (e.g., MySQL) to retrieve minutes and materials from past related meetings, and the retrieved data is stored in memory.

[0360] Input: SQL query

[0361] Output: Relevant data retrieved from the database

[0362] What happens: The server executes the SQL query "SELECT FROM meeting_minutes WHERE date < '2023-10-05'" and stores the relevant meeting minutes data in memory.

[0363] Step 6: Performing Natural Language Processing

[0364] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to a natural language processing engine (e.g., OpenAI GPT-3) for further analysis. The natural language processing engine then analyzes the data and extracts meaning.

[0365] Input: Text data, related data

[0366] Output: Refined data

[0367] Specific operation: The server passes the text data and related data to GPT-3, records a log of "natural language analysis started", and stores the refined data in memory.

[0368] Step 7: Adjust the minutes

[0369] The server uses the results of a natural language processing engine to adjust the content of the minutes based on the emotional information obtained from the emotion recognition engine, for example by highlighting parts of the minutes where the user expressed strong emotions about a particular topic.

[0370] Input: Emotional information, refined data

[0371] Output: Adjusted minutes data

[0372] Specific operation: Based on the emotion information, the server highlights specific parts of the minutes and records a log indicating that the minutes have been adjusted.

[0373] Step 8: Generate and send the final transcript

[0374] The server generates the adjusted minutes data in text format, encodes it in JSON format, and then sends it to the terminal via an HTTP response.

[0375] Input: Adjusted minutes data

[0376] Output: Final minutes data in JSON format

[0377] Specific operation: The server saves the final minutes to " / tmp / 20231005_minutes.json" and logs "Minutes generated, ready to send" to the terminal. It then sends it to the terminal as an HTTP response.

[0378] Step 9: Receive and view the minutes

[0379] The terminal analyzes the JSON data of the final minutes received from the server, formats it so that the user can intuitively check the contents, and displays it on the screen.

[0380] Input: JSON format meeting minutes data

[0381] Output: On-screen transcript

[0382] Specific operation: The device parses the JSON data and logs "Minutes received, display started." The minutes are then formatted in an easy-to-understand format and displayed on the screen.

[0383] Step 10: Review and edit the minutes

[0384] The user can check the minutes displayed on the device screen, edit them in the text area, or add comments as needed. Once editing is complete, the user presses the save button to confirm the changes.

[0385] Input: User edits to the minutes

[0386] Output: Final edited transcript

[0387] Specific operation: The user clicks on a specific comment to edit it and presses the "Save" button. A message saying "Editing completed, minutes saved" is displayed on the terminal.

[0388] (Application example 2)

[0389] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0390] Conventional meeting minutes creation systems require manual input and editing, resulting in inefficient work. They also lack the ability to create minutes that reflect the user's emotions, and are unable to properly reflect the nuances and importance of conversations. Furthermore, they lack advanced functionality, such as operating infotainment systems based on conversation content and emotions in autonomous vehicles, making it difficult to improve the comfort of drivers and passengers.

[0391] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for receiving conversation voice data; speech recognition means for converting the received voice data into text data; natural language processing means for creating minutes based on the converted text data; database management means for saving, retrieving, and referencing the created minutes; means for comparing the saved minutes and materials of related conversations with the new text data and examining the content; means for generating final minutes and sending them to the user; emotion recognition means for analyzing the user's emotion information; and means for operating the vehicle's infotainment system based on the emotion information. This improves the efficiency of automatic minutes creation of conversations in an autonomous vehicle, generates minutes that reflect the user's emotion, and enables optimal operation of the infotainment system based on emotion.

[0392] "Conversational voice data" refers to data that includes the conversational voices of the driver and passengers.

[0393] "Means for receiving" refers to a device or function that receives audio data using a microphone or digital communication means.

[0394] "Speech recognition means" refers to a technology or system for converting voice data into text data.

[0395] "Natural language processing means" refers to technology or systems that analyze text data, understand the context and meaning, and create minutes.

[0396] "Database management means" refers to the systems and technologies used to store, retrieve, and query minutes and related materials.

[0397] A "comparison means" is a technology or system that compares new text data with stored minutes or materials of related conversations and examines their content.

[0398] The "means for generating final minutes and sending them to the user" refers to a technology or system that generates final minutes based on text data and analysis information and sends them to the user's terminal.

[0399] "Emotion recognition means" refers to technology or a system that analyzes a user's emotions based on text data.

[0400] The "means for operating a vehicle infotainment system" refers to a technology or system that appropriately operates the infotainment system based on the user's emotional information.

[0401] This invention describes a system that automatically converts conversational voice data collected in an autonomous vehicle into minutes, and further adjusts the contents of the minutes by recognizing the user's emotions. This system is operated by a server, an on-board computer, and a user terminal in cooperation with each other.

[0402] Server

[0403] 1. Receiving audio data:

[0404] The server receives the conversational audio data sent from the vehicle's onboard computer via an HTTP POST request, and the server receives the request and saves the audio file in a temporary directory.

[0405] 2. Call the speech recognition engine:

[0406] The server passes the saved audio file to a speech recognition engine (e.g., Google Speech-to-Text API) and converts the audio data into text data. It calls the speech recognition API and passes the file path as an argument.

[0407] 3. Receiving converted data and emotion recognition:

[0408] The server receives the text data returned by the speech recognition engine and stores it in memory. Based on this text data, it uses an emotion recognition engine (for example, IBM Watson Tone Analyzer) to analyze the user's emotions. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words.

[0409] 4. Obtaining relevant data:

[0410] The server uses a database management tool (e.g., MySQL) to retrieve minutes and documents of past relevant conversations using SQL queries, and stores the retrieved data in memory.

[0411] 5. Use of natural language processing engines:

[0412] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to a natural language processing engine (e.g., OpenAI GPT-4 (registered trademark)) for detailed analysis. The natural language processing engine analyzes the meaning of the data and extracts the necessary information.

[0413] 6. Adjusting meeting minutes based on emotional information:

[0414] The server adjusts the content of the minutes it generates based on the user's emotional information obtained by the emotion recognition engine, for example by emphasizing parts of the minutes in which the user expressed strong emotions.

[0415] 7. Generate and send the final minutes:

[0416] Based on the scrutinized data and adjusted emotional information, the server generates the final transcript, which is generated in text format, re-encoded in JSON format, and sent to the vehicle's computer as an HTTP response.

[0417] Vehicle Computer

[0418] 1. Collection and transmission of conversational audio data:

[0419] The on-board computer collects conversational voice data in real time using the vehicle's microphone and transmits the collected voice data to a server.

[0420] 2. Receiving and viewing minutes:

[0421] Once the final minutes data is returned from the server, the on-board computer receives and analyzes it, and displays the minutes on the vehicle's display in a format that allows the user to intuitively review the contents.

[0422] 3. Infotainment system operation:

[0423] Based on the emotional information provided by the server, the in-vehicle computer operates the infotainment system, for example, playing relaxing music if the user is feeling tense.

[0424] User

[0425] 1. Contribution to audio data collection:

[0426] The user simply has to naturally converse in the car, and the on-board computer automatically collects the voice data.

[0427] 2. Review and edit the minutes:

[0428] The final minutes are then displayed on the vehicle's display, where the user can review them, make edits or add additional comments if necessary, and confirm the changes when they are complete.

[0429] Specific examples

[0430] For example, if a driver and a passenger have a conversation like, "We talked about a new project at yesterday's meeting," the voice data is sent to the server via the in-vehicle computer. The server then converts the voice into text using a speech recognition engine and retrieves minutes of past related conversations from a database. It then scrutinizes the old and new data through natural language processing, and recognizes the user's emotions during the conversation using an emotion recognition engine. For example, if the passenger is nervous, the system will adjust the music played to be more relaxing based on the emotional information. The final minutes are then sent back to the user via the in-vehicle display, where they can review them and make any necessary corrections.

[0431] Example prompts for generative AI models

[0432] "Convert the conversation between the driver and passenger into text and analyze the sentiment."

[0433] "Suggest optimal music choices and rerouting based on conversational content and emotional information."

[0434] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0435] Step 1:

[0436] Audio data collection:

[0437] The terminal (on-board computer) uses the in-car microphone to collect real-time conversational voice data of the driver and passengers. The input is analog voice signals from the in-car microphone, which are converted into digital data and temporarily stored in internal memory. The collected voice data is formatted in batch format for transmission to the server.

[0438] Step 2:

[0439] Sending audio data:

[0440] The device sends the collected and formatted audio data to the server as an HTTP POST request. The input is the formatted audio data, and the output is the data transfer to the server. The audio data sent from the device is saved in a temporary directory on the server.

[0441] Step 3:

[0442] Saving audio data:

[0443] The server saves the audio data sent from the device to a temporary directory. The input is the audio data sent from the device, and the output is the saved audio file. The specified directory path is used to save the file.

[0444] Step 4:

[0445] Calling the speech recognition engine:

[0446] The server passes the saved audio file to a speech recognition engine (e.g., Google Speech-to-Text API) and converts the audio data into text data. The input is the path to the saved audio file, and the output is the text data returned by the speech recognition engine. This text data is stored in memory.

[0447] Step 5:

[0448] Calling the emotion recognition engine:

[0449] The server passes the text data obtained from the speech recognition engine to an emotion recognition engine (for example, IBM Watson Tone Analyzer) for analysis. The input is text data, and the output is analyzed emotional information. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words.

[0450] Step 6:

[0451] Retrieving related data:

[0452] The server uses a database management tool to retrieve minutes and related documents from past conversations. The input is an SQL query, and the output is the retrieved minutes and related documents. The retrieved data is stored in memory.

[0453] Step 7:

[0454] Using natural language processing engines:

[0455] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to a natural language processing engine (e.g., OpenAI GPT-4) for detailed analysis. The input is the text data and related data, and the output is the analyzed content information. The natural language processing engine is used to analyze the meaning of the data and extract the necessary information.

[0456] Step 8:

[0457] Adjusting meeting minutes based on emotional information:

[0458] The server adjusts the content of the meeting minutes it generates based on the emotional information obtained by the emotion recognition engine. The input is text data and emotional information, and the output is the adjusted minutes. For example, it may perform processing such as emphasizing parts of statements that express strong emotions.

[0459] Step 9:

[0460] Generate and send final transcripts:

[0461] The server finally generates the adjusted minutes and encodes them in JSON format again. The input is the adjusted minutes data, and the output is the encoded minutes data. This minutes data is sent to the terminal as an HTTP response.

[0462] Step 10:

[0463] Receive and view minutes:

[0464] The terminal receives the final minutes data from the server and analyzes it. The input is the minutes data from the server, and the output is the analyzed minutes content. The minutes data is displayed on the terminal display and formatted so that the user can intuitively check the content.

[0465] Step 11:

[0466] To operate the infotainment system:

[0467] The device operates the infotainment system based on the emotional information provided by the server. The input is the emotional information from the server, and the output is the system's operation command. For example, if the user is nervous, the device will automatically play relaxing music based on that emotional information.

[0468] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0469] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0470] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0471] [Second embodiment]

[0472] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0473] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0474] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0475] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0476] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0477] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0478] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0479] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0480] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0481] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0482] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0483] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0484] The present invention is a system that automatically converts conference audio data into minutes and quickly provides them. This system is operated in cooperation between a server, terminals, and users.

[0485] Server

[0486] 1. Receiving audio data

[0487] The server receives the conference audio data sent from the terminals. The audio data is usually sent via an HTTP request, and the server receives the request and stores the audio file in a temporary storage location.

[0488] 2. Calling the voice recognition engine

[0489] The received voice data is passed to the speech recognition engine, which converts the data into text data and returns the converted text to the server, which stores the text data in its memory.

[0490] 3. Obtaining relevant data

[0491] The server uses database management means to retrieve minutes and materials from past relevant meetings using SQL queries, and stores the retrieved relevant data in memory.

[0492] 4. Use of natural language processing engines

[0493] The server passes the text data received from the speech recognition engine and related data retrieved from the database to the natural language processing engine, which then analyzes the meaning of the data and extracts the necessary information.

[0494] 5. Generate final transcripts

[0495] Based on the scrutinized data, the server generates the final transcript, which is again encoded in JSON format and sent back to the device.

[0496] Terminal (Client)

[0497] 1. Providing an upload interface

[0498] The terminal provides a user with an interface for uploading conference audio data. This interface is implemented as a GUI including a file selection and upload button.

[0499] 2. Sending audio data

[0500] When a user selects an audio file and presses the upload button, the device sends this audio data to the server as an HTTP POST request.

[0501] 3. Receiving and displaying minutes

[0502] Once the final minutes are returned from the server, the device receives this data and displays it to the user. The minutes data is encoded in JSON format, so the device parses it and displays it on the screen.

[0503] User

[0504] 1. Uploading audio data

[0505] Users upload conference audio data through the interface provided by their devices. They select the appropriate audio file in the file selection dialog and click the upload button.

[0506] 2. Review and edit the minutes

[0507] Once the final minutes are displayed on the device, the user can review them and make edits as needed. Editing is done directly in the text area, and once edits are complete, the changes are saved by pressing the save button.

[0508] Specific examples

[0509] For example, if a user uploads an audio file for a "Project Status Meeting on October 5, 2023," the audio data is sent to the server. The server then converts the audio into text using a speech recognition engine and retrieves minutes of related past meetings from the database. It then scrutinizes the old and new data using natural language processing, generates the final minutes, and sends them back to the device. The user can then review the received minutes and make any necessary corrections, completing the minutes quickly and accurately.

[0510] This system will greatly improve the efficiency of creating meeting minutes, increasing accuracy and convenience.

[0511] The processing flow will be explained below.

[0512] Server

[0513] Step 1:

[0514] The server receives the conference audio data sent from the terminal via an HTTP POST request, and the received audio file is saved in a temporary directory.

[0515] Step 2:

[0516] The server passes the saved audio file to a speech recognition engine and converts the audio data into text data. At this time, it calls the speech recognition API and passes the file path as an argument.

[0517] Step 3:

[0518] The server receives the text data returned by the speech recognition engine and stores it in memory. The returned data is usually provided in JSON format.

[0519] Step 4:

[0520] The server accesses the database to retrieve relevant meeting minutes and materials using SQL queries, and stores the retrieved data in memory.

[0521] Step 5:

[0522] The server calls a natural language processing engine to compare and examine the text data obtained through speech recognition with related data retrieved from the database, thereby extracting important information.

[0523] Step 6:

[0524] The server generates the final minutes based on the scrutinized data, which are generated in text format and then encoded again in JSON format.

[0525] Step 7:

[0526] The server sends the generated minutes to the terminal as an HTTP response.

[0527] Terminal (Client)

[0528] Step 1:

[0529] The terminal provides the user with an interface for uploading audio data, which is designed as a GUI containing a file selection dialog and an upload button.

[0530] Step 2:

[0531] When the user selects an audio file and presses the upload button, the device sends the audio data to the server via an HTTP POST request.

[0532] Step 3:

[0533] When the final minutes data is returned from the server, the terminal receives and parses this JSON format data.

[0534] Step 4:

[0535] The parsed minutes are displayed on the device screen as nicely formatted text so that users can intuitively check the contents.

[0536] User

[0537] Step 1:

[0538] After the conference, the user uploads the audio data through the interface provided by the device by selecting the correct audio file in the file selection dialog and clicking the upload button.

[0539] Step 2:

[0540] Once the upload is complete, the user can check the minutes sent back from the server on their device. The displayed minutes reflect the contents of past meetings, making it easy to confirm important points.

[0541] Step 3:

[0542] The user can check the displayed minutes, make corrections or add comments as necessary, and when the corrections are complete, press the save button to confirm the changes.

[0543] The above processing flow makes it possible to efficiently create meeting minutes and provide accurate and prompt minutes.

[0544] Example 1

[0545] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0546] The process of converting meeting audio data into minutes quickly and accurately is time-consuming and labor-intensive when done manually, making it inefficient. Furthermore, manual minutes are prone to errors and often lack sufficient linkage to past meeting information. This creates a need for improved accuracy in minutes and a more efficient creation process. It is also necessary to provide an interface that makes it easy to upload audio data and edit and review minutes.

[0547] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0548] In this invention, the server includes: means for receiving conference audio data; speech recognition means for converting the received audio data into text data; natural language processing means for creating minutes based on the converted text data; database management means for saving, retrieving, and referencing the created minutes; means for comparing the new text data with stored minutes and materials from related meetings and examining the content; means for generating and transmitting final minutes to users; and means for using a client-server architecture that smoothly transmits and receives audio data, generates, and displays minutes. This improves the efficiency of creating meeting minutes and enables the creation of highly accurate minutes. Users can also easily upload audio files and review and edit the minutes.

[0549] "Conference audio data" refers to digital audio files that record what is said during a conference or meeting.

[0550] "Speech recognition means" refers to technology or equipment for converting voice data into text data. Specifically, it uses a speech recognition engine or API.

[0551] "Natural language processing means" refers to technology or equipment that analyzes text data, understands its meaning and content, and extracts appropriate information. Specifically, it uses natural language processing engines and APIs.

[0552] "Database management means" refers to the technology or device used to store, retrieve, and reference data. Specifically, a database management system (DBMS) is used.

[0553] "Client-server architecture" is a system design model that allows communication between a server that provides data and resources and a client that uses those data and resources.

[0554] A "user" is a person who operates the system and uploads audio data and checks and edits minutes.

[0555] A "terminal" is a device that allows a user to access and operate the system. Specifically, it includes computers, smartphones, etc.

[0556] A "minutes" is a document that records what was said and what was decided at a conference or meeting.

[0557] An "audio file upload interface" is a screen or function that a user uses to submit audio data to the system.

[0558] "Text data" is character information converted by a voice recognition means.

[0559] The "JSON format" is a lightweight data interchange format for storing and exchanging data.

[0560] MODE FOR CARRYING OUT THE INVENTION

[0561] The present invention is a system for automatically converting conference audio data into minutes and providing them promptly. A specific embodiment of this system will now be described in detail.

[0562] server

[0563] The server uses the following main hardware and software to perform the following series of data processing and calculations:

[0564] 1. Receiving audio data

[0565] The server receives the conference audio data sent from the device via an HTTP request, processes the request using a Python framework such as Flask, and temporarily stores the audio file on the server's local disk.

[0566] 2. Calling the voice recognition engine

[0567] The server passes the saved voice data to a speech recognition engine, which converts the voice data to text using the Google Cloud Speech-to-Text API, and stores the converted text in the server's memory.

[0568] 3. Obtaining relevant data

[0569] The server retrieves minutes and materials from past relevant meetings from the MySQL database using SQL queries, including filter conditions such as meeting dates and project names, and stores the retrieved data in memory.

[0570] 4. Use of natural language processing engines

[0571] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to the IBM Watson Natural Language Understanding API, which analyzes the meaning of the data, executes a process to extract the necessary information, and stores the extracted results in memory.

[0572] 5. Generate final transcripts

[0573] The server generates the final minutes based on the scrutinized data. It uses the Python Jinja2 template engine to generate the minutes in JSON format and sends it back to the terminal as an HTTP response.

[0574] Terminal

[0575] The terminal provides an interface for users to upload audio data and review / edit the generated minutes. The following hardware and software are used to perform the operation:

[0576] 1. Providing an upload interface

[0577] The terminal provides a GUI for users to upload conference audio data. An HTML form is used to implement a file selection dialog and an upload button, and JavaScript is used to automatically start uploading after file selection.

[0578] 2. Sending audio data

[0579] When a user selects an audio file and presses the upload button, the device uses JavaScript's Fetch API to send this audio data to the server as an HTTP POST request, which includes the binary data of the selected audio file.

[0580] 3. Receiving and displaying minutes

[0581] Once the final minutes are returned from the server, the device receives this data, retrieves the Fetch API response, parses the response data as JSON, and displays it in the user interface using HTML and JavaScript.

[0582] User

[0583] Users use their devices to upload the meeting audio data and check and edit the generated minutes.

[0584] 1. Uploading audio data

[0585] The user uploads the conference audio data using the interface provided by the terminal: open the file selection dialog, select the audio file to upload, and click the upload button.

[0586] 2. Review and edit the minutes

[0587] The user checks the final minutes displayed on the device and edits them as necessary. They edit the content directly in the displayed text area and press the save button when they are done. The edits are captured by a JavaScript event listener and saved in local storage.

[0588] Specific examples

[0589] For example, if a user uploads an audio file for the "Project Status Meeting on October 5, 2023," the following process takes place: The user selects the audio file from the device's GUI and presses the upload button. The server receives an HTTP POST request using the Flask framework and saves the audio file to its local disk. It then sends a request to a speech recognition engine (Google Cloud Speech-to-Text API) and receives the converted text. Next, it retrieves past meeting minutes data from a MySQL database using filter criteria such as "project name" and "meeting date." This data is passed to the IBM Watson Natural Language Understanding API, where the content is examined and analyzed. Finally, the Jinja2 template engine is used to generate the final meeting minutes in JSON format and return them as an HTTP response. The device receives this response data and displays it in the user interface using HTML and JavaScript. The user reviews the displayed minutes and makes any necessary edits in the text area.

[0590] Prompt Sentence Examples

[0591] An example of a prompt is, "Convert the audio file of the project status meeting on October 5, 2023 into text, and display the minutes that have been reviewed and generated in conjunction with past related meeting minutes. The APIs used are speech recognition API and natural language processing API."

[0592] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0593] Step 1:

[0594] The server receives the conference audio data sent from the terminal. Specifically, the terminal creates an HTTP POST request and sends the binary data of the audio file to the server. The server processes this request using a Python framework such as Flask and temporarily saves the received audio file on the local disk. The input is the audio file sent from the terminal, and the output is the audio file saved on the server.

[0595] Step 2:

[0596] The server passes the saved audio file to the speech recognition engine. Specifically, it creates an API request to use the Google Cloud Speech-to-Text API to convert the audio data to text. The server uses the path to the audio file and the API key as input and receives the converted text data as a response. The output is the text data converted from the audio file.

[0597] Step 3:

[0598] The server retrieves minutes and materials from past relevant meetings from a MySQL database. Specifically, it executes an SQL query that includes filter conditions such as the meeting date and project name. The input is the filter conditions for searching the database, and the output is the retrieved meeting minutes and materials data. This data is stored in memory.

[0599] Step 4:

[0600] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to the natural language processing engine. Specifically, it uses the IBM Watson Natural Language Understanding API to analyze the meaning of the data and extract the necessary information. The input is the speech-recognized text data and related data, and the output is the scrutinized text data.

[0601] Step 5:

[0602] The server generates the final minutes based on the parsed data. Specifically, it uses the Python Jinja2 template engine to generate minutes in JSON format. The input is the parsed text data and template information, and the output is the generated minutes data. This minutes data is returned to the terminal as an HTTP response.

[0603] Step 6:

[0604] The terminal receives the final minutes data returned from the server. Specifically, it receives the HTTP response using the Fetch API and parses the response data as JSON. The input is the JSON minutes data returned from the server, and the output is the parsed minutes data.

[0605] Step 7:

[0606] The terminal displays the parsed minutes data on the user interface. Specifically, it uses HTML and JavaScript to insert the parsed data into a text display area. The input is the parsed minutes data, and the output is the minutes displayed on the screen.

[0607] Step 8:

[0608] The user checks the minutes displayed on the terminal and makes corrections as necessary. Specifically, the user edits the content directly in the displayed text area and clicks the save button when finished. The input is the user's edited content, and the output is the corrected minutes.

[0609] Step 9:

[0610] The terminal sends the edited or revised content by the user to the server and updates the final minutes. Specifically, JavaScript is used to collect the revisions and send them to the server as an HTTP POST request. The input is the minutes data edited by the user, and the output is the minutes data updated on the server. This series of processes allows meeting minutes to be created quickly and accurately.

[0611] (Application example 1)

[0612] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0613] Traditionally, work instructions and reports within factories have often been written by hand or orally, which has led to the problem of time-consuming communication and sharing of information. There is also a high risk of missed records and inaccurate reports. This can lead to reduced work efficiency within the factory and affect productivity. Furthermore, in order to provide quick and accurate work reports, workers and engineers have to spend time and effort to keep records, which increases the workload.

[0614] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0615] In this invention, the server includes means for receiving conference voice data, speech recognition means for converting the received voice data into text data, natural language processing means for creating minutes based on the converted text data, database management means for saving, retrieving, and referencing the created minutes, means for comparing the saved minutes and materials of related meetings with the new text data and examining the content, means for generating final minutes and sending them to the user, means for converting voice recordings of work instructions and reports within the factory into text, and means for uploading the generated minutes to a central system within the factory, thereby enabling quick and accurate recording of work instructions and reports within the factory.

[0616] "Conference voice data" refers to digital data of voice generated during a conference or meeting.

[0617] "Means for receiving audio data" refers to an interface or module for receiving and processing audio data from another device or system.

[0618] "Speech recognition means" refers to technology or equipment that analyzes voice data and converts it into text data.

[0619] "Natural language processing means" is a technology that analyzes the meaning of text data and organizes and extracts data.

[0620] A "database management tool" is a system or software used to efficiently store, retrieve, and access data.

[0621] "Means for examining the content" refers to techniques and methods for analyzing the acquired data in detail and extracting the necessary information.

[0622] "Means for transmitting to users" refers to methods or systems for delivering the final generated data or information to users.

[0623] "Means for converting audio recordings of work instructions and reports within a factory into text" refers to technology that records the audio instructions and reports of factory workers and engineers and converts them into text data.

[0624] "Means for uploading the generated minutes to a central system within the factory" refers to an interface or system for transferring the generated minutes to a central system installed within the factory for storage and management.

[0625] The present invention aims to improve the efficiency of labor in factories and the accuracy of reports. This system is operated through collaboration between a server, a terminal, and a user.

[0626] Server

[0627] The server plays a central role in receiving voice data of work instructions and reports within the factory, and processing and storing that data. Specifically, it performs the following processes:

[0628] Receiving audio data

[0629] The server receives the audio data sent from the device via an HTTP request and temporarily stores the received audio file.

[0630] Calling the speech recognition engine

[0631] The voice data is passed to a speech recognition engine and converted into text data. The speech recognition engine used is something like the Google Cloud Speech-to-Text API.

[0632] Retrieving related data

[0633] The server retrieves reports and documents of past related work from the database, searches using SQL queries, and stores the required data in memory.

[0634] Use of natural language processing engines

[0635] The text data received from the speech recognition engine and related data retrieved from the database are passed to a natural language processing engine for detailed analysis, using engines such as TensorFlow and spaCy.

[0636] Generate and send meeting minutes

[0637] Finally, based on the scrutinized data, the server generates a transcript, which is encoded in JSON format and sent back to the device.

[0638] Terminal (Client)

[0639] The terminal is the interface for the user and is responsible for uploading audio data and displaying received minutes.

[0640] Providing an upload interface

[0641] The terminal provides an audio file upload interface, which is implemented as a GUI (Graphical User Interface) with a file selection and upload button.

[0642] Sending audio data

[0643] After selecting an audio file and clicking the upload button, the device will send the audio file to the server as an HTTP POST request.

[0644] Receiving and viewing minutes

[0645] The minutes returned from the server are received, the JSON format data is parsed, and the data is displayed on the screen, allowing the user to check the minutes and edit them as necessary.

[0646] User

[0647] The user is the end of the system, providing the audio data and checking and editing the final minutes.

[0648] Uploading audio data

[0649] Users use the terminal interface to upload audio data of factory operations and reports.

[0650] Review and edit minutes

[0651] The generated minutes can be checked and edited as necessary. For example, by uploading audio data such as "There is a problem with the right-hand conveyor belt. The motor is overheating and needs to be replaced," minutes can be created quickly and accurately.

[0652] Specific prompt examples

[0653] Generate a transcript from the following audio: "There is a problem with the right-hand conveyor belt. The motor is overheating and needs to be replaced."

[0654] In this way, this system enables efficient and accurate work instructions and reports within the factory. The server is a high-performance computer, the terminals are mobile devices with internet access, and the users are assumed to be factory workers and engineers.

[0655] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0656] Step 1: Recording audio data

[0657] Users record audio data at the factory work site. Specifically, engineers and workers use microphones to record work instructions and reports as audio. This audio data is saved on the device.

[0658] Input: User speech

[0659] Output: Audio file saved on device

[0660] Step 2: Upload your audio data

[0661] The user sends the recorded audio data to the server using the upload interface on the device, by selecting the audio file through the GUI and clicking the upload button.

[0662] Input: Audio file saved on the device

[0663] Output: Audio file sent to the server

[0664] Step 3: Receiving audio data

[0665] The server receives the audio data sent from the device via an HTTP request and temporarily stores the received audio data.

[0666] Input: Audio data sent from the device

[0667] Output: Temporarily saved audio file

[0668] Step 4: Calling the speech recognition engine

[0669] The server passes the temporarily saved audio file to a speech recognition engine and converts the audio data into text data, using a speech recognition service such as the Google Cloud Speech-to-Text API.

[0670] Input: Temporarily saved audio file

[0671] Output: Audio information converted to text data

[0672] Step 5: Retrieve related data

[0673] The server retrieves the minutes and documents of past related work from the database, specifically by using SQL queries to search and retrieve the necessary past data.

[0674] Input: SQL query

[0675] Output: Obtained past minutes and document data

[0676] Step 6: Use a natural language processing engine

[0677] The server then passes the text data obtained from the speech recognition engine and related data retrieved from the database to a natural language processing engine for further analysis, such as using TensorFlow or spaCy to analyze the text data and parse its meaning.

[0678] Input: Text data and historical data

[0679] Output: Scanned text data

[0680] Step 7: Generate and send the minutes

[0681] The server generates the final transcript based on the scrutinized data, which is then encoded in JSON format and sent to the device.

[0682] Input: vetted text data

[0683] Output: JSON formatted minutes

[0684] Step 8: Receive and view the minutes

[0685] The terminal receives the minutes in JSON format sent from the server, parses them, and displays them to the user. Specifically, it displays the contents of the minutes on the screen using a GUI.

[0686] Input: JSON formatted minutes

[0687] Output: Transcript displayed on screen

[0688] Step 9: Review and edit the minutes

[0689] The user checks the minutes displayed on the terminal and edits them as necessary. For example, if there are any clerical errors, the user corrects them and saves them.

[0690] Input: Minutes displayed on screen

[0691] Output: Edited transcript

[0692] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0693] The present invention provides a system that automatically converts conference audio data into minutes and further adjusts the content of the minutes by recognizing the emotions of users. This system is operated in cooperation between a server, terminals, and users.

[0694] Server

[0695] 1. Receiving audio data

[0696] The server receives the conference audio data sent from the terminal. The audio data is sent via an HTTP POST request, and the server receives this request and saves the audio file in a temporary directory.

[0697] 2. Calling the voice recognition engine

[0698] The server passes the saved audio file to the speech recognition engine, converts the audio data into text data, and calls the speech recognition API, passing the file path as an argument.

[0699] 3. Receiving converted data and emotion recognition

[0700] The server receives the text data returned by the speech recognition engine and stores it in memory. Based on this text data, it uses an emotion recognition engine to analyze the user's emotions. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words.

[0701] 4. Acquiring relevant data

[0702] The server uses database management means to retrieve minutes and materials from past relevant meetings using SQL queries, and stores the retrieved data in memory.

[0703] 5. Use of natural language processing engines

[0704] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to the natural language processing engine, which then analyzes the meaning of the data and extracts the necessary information.

[0705] 6. Adjusting meeting minutes based on emotional information

[0706] The server adjusts the content of the minutes it generates based on the user's emotional information obtained by the emotion recognition engine, for example by emphasizing parts of the minutes in which the user expressed strong emotions.

[0707] 7. Generate and send final minutes

[0708] Based on the scrutinized data and adjusted sentiment information, the server generates the final transcript, which is generated in text format, encoded again in JSON format, and sent to the device as an HTTP response.

[0709] Terminal (Client)

[0710] 1. Providing an upload interface

[0711] The terminal provides the user with an interface for uploading audio data, which is implemented as a GUI containing a file selection dialog and an upload button.

[0712] 2. Sending audio data

[0713] When the user selects an audio file and presses the upload button, the device sends the audio data to the server via an HTTP POST request.

[0714] 3. Receiving and displaying minutes

[0715] When the final minutes data is returned from the server, the device receives and parses the JSON format data. The parsed minutes are displayed on the device screen and formatted so that the user can intuitively check the contents.

[0716] User

[0717] 1. Uploading audio data

[0718] After the conference, the user uploads the audio data through the interface provided by the device by selecting the correct audio file in the file selection dialog and clicking the upload button.

[0719] 2. Review and edit the minutes

[0720] Once the final minutes are displayed on the device, the user can review them and make edits or add comments as needed. Edits are made directly in the text area, and once corrections are complete, the changes can be confirmed by pressing the save button.

[0721] Specific examples

[0722] For example, if a user uploads an audio file for a "Project Progress Meeting on October 5, 2023," the audio data is sent to the server. The server then converts the audio into text using a speech recognition engine and retrieves minutes of related past meetings from a database. It then scrutinizes old and new data through natural language processing, and recognizes the user's emotions during the meeting using an emotion recognition engine. For example, if there is an emphasis on a particular topic, the system adjusts the display to highlight that part based on the emotional information. The final minutes are then sent back to the user via their device, who can review them and make any necessary corrections.

[0723] This system makes the creation of meeting minutes significantly more efficient than conventional manual operations, and also makes it possible to generate more detailed minutes that reflect the user's emotions during the meeting.

[0724] The processing flow will be explained below.

[0725] Server

[0726] Step 1:

[0727] The server receives the conference audio data sent from the terminal via an HTTP POST request, and the received audio file is saved in a temporary directory.

[0728] Step 2:

[0729] The server passes the saved audio file to a speech recognition engine and converts the audio data into text data. At this time, it calls the speech recognition API and passes the file path as an argument.

[0730] Step 3:

[0731] The server receives the text data returned by the speech recognition engine and stores it in memory. The returned data is usually provided in JSON format.

[0732] Step 4:

[0733] The server passes this text data to an emotion recognition engine, which analyzes the user's emotions. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words, and stores the results in memory.

[0734] Step 5:

[0735] The server accesses the database to retrieve relevant meeting minutes and materials using SQL queries, and stores the retrieved data in memory.

[0736] Step 6:

[0737] The server invokes a natural language processing engine to compare and examine the text data obtained through speech recognition with related data retrieved from the database, thereby extracting important information.

[0738] Step 7:

[0739] The server adjusts the content of the minutes it generates based on the user's emotional information obtained by the emotion recognition engine, for example by emphasizing parts of the minutes in which the user expressed strong emotions.

[0740] Step 8:

[0741] The server generates the final transcript based on the scrutinized data and adjusted sentiment information. The transcript is generated in text format, then encoded again in JSON format and sent to the device as an HTTP response.

[0742] Terminal (Client)

[0743] Step 1:

[0744] The terminal provides the user with an interface for uploading audio data, which is designed as a GUI containing a file selection dialog and an upload button.

[0745] Step 2:

[0746] When the user selects an audio file and presses the upload button, the device sends the audio data to the server via an HTTP POST request.

[0747] Step 3:

[0748] When the final minutes data is returned from the server, the terminal receives and parses this JSON format data.

[0749] Step 4:

[0750] The parsed minutes are displayed on the device screen as nicely formatted text so that users can intuitively check the contents.

[0751] User

[0752] Step 1:

[0753] After the conference, the user uploads the audio data through the interface provided by the device by selecting the correct audio file in the file selection dialog and clicking the upload button.

[0754] Step 2:

[0755] Once the upload is complete, the user can check the minutes sent back from the server on their device. The displayed minutes reflect the contents of past meetings, making it easy to confirm important points.

[0756] Step 3:

[0757] The user can check the displayed minutes, make corrections or add comments as necessary, and when the corrections are complete, press the save button to confirm the changes.

[0758] Examples:

[0759] When a user uploads an audio file for the "Project Progress Meeting on October 5, 2023," the audio data is sent to the server. The server then converts the audio into text using a speech recognition engine and retrieves minutes of past related meetings from a database. It then scrutinizes old and new data through natural language processing, and recognizes the user's emotions during the meeting using an emotion recognition engine. For example, if there is an emphasis on a particular topic, the system adjusts the display to highlight that part based on the emotional information. The final minutes are then sent back to the user via their device, who can review them and make any necessary corrections.

[0760] Example 2

[0761] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0762] Conventional systems that simply transcribe meeting audio data have the problem of not being able to thoroughly examine the content or reflect emotions, resulting in a decline in the quality of the minutes. Creating minutes that reflect the user's emotions is also a time-consuming and difficult process to do efficiently.

[0763] The specification process by the specification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for receiving voice data, voice recognition means for converting the received voice data into text data, emotion recognition means for analyzing the user's emotion based on the converted text data, database management means for acquiring past related data, natural language processing means for analyzing the acquired related data and the text data obtained from the voice recognition means and examining the content, means for adjusting the content of the minutes based on the emotion information obtained by the emotion recognition means, and means for generating the final minutes and sending them to the user. This makes it possible to automatically and efficiently generate highly accurate minutes that reflect the user's emotions.

[0764] "Audio data" refers to a recording file of a conference that is uploaded by a user, and is data that is converted into text data by a voice recognition means.

[0765] "Speech recognition means" refers to software or an engine for analyzing received voice data and converting it into text data.

[0766] The "emotion recognition means" is software or an engine for analyzing the user's emotions based on the converted text data.

[0767] "Database management means" refers to the means for storing, retrieving, and referencing minutes and materials from past related meetings.

[0768] "Natural language processing means" refers to software or an engine for analyzing and examining the acquired association data and the text data obtained from the speech recognition means.

[0769] The "minutes adjustment means" is a means for adjusting the contents of the minutes to be generated based on the emotion information obtained by the emotion recognition means.

[0770] The "minutes generation means" is a means for generating the final minutes and transmitting them to the user.

[0771] A "terminal" is a device or software that allows a user to upload audio files and view and edit the generated minutes.

[0772] "User" refers to a person who uses this system to upload conference audio data and to review and edit the generated minutes.

[0773] The present invention provides a system that automatically converts conference audio data into minutes and adjusts the content of the minutes by recognizing the emotions of users. This system is operated in cooperation between a server, terminals, and users.

[0774] First, let's explain about terminals. A terminal is a device or software that provides a user with an interface for uploading audio data. Specifically, it is implemented as a GUI (Graphical User Interface) that includes a file selection dialog and an upload button. When a user selects an audio file and presses the upload button, the terminal sends the audio data to the server via an HTTP POST request.

[0775] Next, we will explain the server processing. First, the server receives the audio data sent from the terminal. The audio data is sent via an HTTP POST request, and the server receives this request and saves the audio file in a temporary directory. For example, it may be saved in a path such as " / tmp / meeting_audio.wav".

[0776] The server passes the saved audio file to a speech recognition engine (generally a speech recognition API is used) and converts the audio data into text data. Specifically, it calls the Google Cloud Speech-to-Text API and passes the file path as an argument. The returned text data is stored in memory.

[0777] The server then analyzes the user's emotions using an emotion recognition engine, such as IBM Watson Natural Language Understanding, to determine the user's emotions from the content and word usage of the text data. The analysis results are stored in memory.

[0778] The server then issues SQL queries to a database management system (e.g., MySQL) to retrieve minutes and documents from past related meetings. The retrieved data is also stored in memory. This provides information for comparing past minutes with new text data and examining their content.

[0779] The server then uses a natural language processing engine (such as OpenAI GPT-3) to analyze the text data from the speech recognition engine and related data from the database to extract meaning from the data, a process that makes the content of the meeting clearer.

[0780] The server adjusts the content of the meeting minutes it generates based on the user's emotional information obtained from the emotion recognition engine. For example, it highlights parts of the meeting minutes in which the user expressed strong emotions about a particular topic. Once the adjustments are complete, the server generates the final meeting minutes in text format, encodes them again in JSON format, and sends them to the terminal as an HTTP response.

[0781] Finally, the terminal parses the JSON data of the final minutes received from the server, formats it so that the user can intuitively check the contents, and displays it on the screen. The user can review it and make edits or add comments as needed. This series of operations allows the user to obtain efficient and detailed minutes.

[0782] As a specific example, if a user uploads an audio file for the "Project Status Meeting on October 5, 2023," the audio data is sent to the server. The server converts the audio into text using the Google Cloud Speech-to-Text API and retrieves minutes of past related meetings from a database (MySQL). It then uses OpenAI GPT-3 to scrutinize old and new data, and IBM Watson Natural Language Understanding to recognize the user's emotions during the meeting. For example, if there are emphasized opinions on a particular topic, the system adjusts the display to highlight those parts based on the emotional information. The final minutes are then sent back to the user via their device, who can review them and make any necessary corrections.

[0783] An example of a prompt sentence could be "Please upload the audio data of next week's important meeting and automatically generate the minutes."

[0784] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0785] Step 1: Upload your audio data

[0786] The user accesses the interface provided by the device and selects an audio file. When the user clicks the "Upload" button, the audio file is sent from the device to the server. The audio data is sent to the server using the HTTP POST method, and the audio file is included as part of the request.

[0787] Input: An audio file selected by the user

[0788] Output: Audio data sent from the device to the server

[0789] Specific behavior: The user selects "Project Status Meeting on October 5, 2023.wav" and presses the upload button. This file is sent to the server as an HTTP POST request.

[0790] Step 2: Receiving and storing audio data

[0791] The server receives the HTTP POST request sent from the device and temporarily stores the audio data. When temporarily storing the data, an appropriate path (e.g., " / tmp / 20231005_meeting.wav") is specified.

[0792] Input: Audio data sent from the device

[0793] Output: Audio file saved on the server

[0794] Specific operation: The server saves the received audio file to " / tmp / 20231005_meeting.wav" and records a log that the upload is complete.

[0795] Step 3: Performing speech recognition

[0796] The server passes the path of the saved audio file to a speech recognition engine (for example, Google Cloud Speech-to-Text API) to convert the audio data into text data. A request is sent to the API, and when the request is completed, the converted text data is returned as a response.

[0797] Input: Path to the audio file

[0798] Output: Text data

[0799] Specific operation: The server passes " / tmp / 20231005_meeting.wav" to the Google Cloud Speech-to-Text API and records the log "Speech recognition completed, text data received."

[0800] Step 4: Performing Emotion Recognition

[0801] The server passes the acquired text data to an emotion recognition engine (e.g., IBM Watson Natural Language Understanding) to analyze the user's emotions. Emotional information is returned as the analysis result, and the server stores this in its memory.

[0802] Input: Text data

[0803] Output: Emotional information

[0804] Specific operation: The server passes the text data to IBM Watson and records a log indicating "emotion recognition completed, analysis results received."

[0805] Step 5: Retrieve related data

[0806] The server issues SQL queries to a database management system (e.g., MySQL) to retrieve minutes and materials from past related meetings, and the retrieved data is stored in memory.

[0807] Input: SQL query

[0808] Output: Relevant data retrieved from the database

[0809] What happens: The server executes the SQL query "SELECT FROM meeting_minutes WHERE date < '2023-10-05'" and stores the relevant meeting minutes data in memory.

[0810] Step 6: Performing Natural Language Processing

[0811] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to a natural language processing engine (e.g., OpenAI GPT-3) for further analysis. The natural language processing engine then analyzes the data and extracts meaning.

[0812] Input: Text data, related data

[0813] Output: Refined data

[0814] Specific operation: The server passes the text data and related data to GPT-3, records a log of "natural language analysis started", and stores the refined data in memory.

[0815] Step 7: Adjust the minutes

[0816] The server uses the results of a natural language processing engine to adjust the content of the minutes based on the emotional information obtained from the emotion recognition engine, for example by highlighting parts of the minutes where the user expressed strong emotions about a particular topic.

[0817] Input: Emotional information, refined data

[0818] Output: Adjusted minutes data

[0819] Specific operation: Based on the emotion information, the server highlights specific parts of the minutes and records a log indicating that the minutes have been adjusted.

[0820] Step 8: Generate and send the final transcript

[0821] The server generates the adjusted minutes data in text format, encodes it in JSON format, and then sends it to the terminal via an HTTP response.

[0822] Input: Adjusted minutes data

[0823] Output: Final minutes data in JSON format

[0824] Specific operation: The server saves the final minutes to " / tmp / 20231005_minutes.json" and logs "Minutes generated, ready to send" to the terminal. It then sends it to the terminal as an HTTP response.

[0825] Step 9: Receive and view the minutes

[0826] The terminal analyzes the JSON data of the final minutes received from the server, formats it so that the user can intuitively check the contents, and displays it on the screen.

[0827] Input: JSON format meeting minutes data

[0828] Output: On-screen transcript

[0829] Specific operation: The device parses the JSON data and logs "Minutes received, display started." The minutes are then formatted in an easy-to-understand format and displayed on the screen.

[0830] Step 10: Review and edit the minutes

[0831] The user can check the minutes displayed on the device screen, edit them in the text area, or add comments as needed. Once editing is complete, the user presses the save button to confirm the changes.

[0832] Input: User edits to the minutes

[0833] Output: Final edited transcript

[0834] Specific operation: The user clicks on a specific comment to edit it and presses the "Save" button. A message saying "Editing completed, minutes saved" is displayed on the terminal.

[0835] (Application example 2)

[0836] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0837] Conventional meeting minutes creation systems require manual input and editing, resulting in inefficient work. They also lack the ability to create minutes that reflect the user's emotions, and are unable to properly reflect the nuances and importance of conversations. Furthermore, they lack advanced functionality, such as operating infotainment systems based on conversation content and emotions in autonomous vehicles, making it difficult to improve the comfort of drivers and passengers.

[0838] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for receiving conversation voice data; speech recognition means for converting the received voice data into text data; natural language processing means for creating minutes based on the converted text data; database management means for saving, retrieving, and referencing the created minutes; means for comparing the saved minutes and materials of related conversations with the new text data and examining the content; means for generating final minutes and sending them to the user; emotion recognition means for analyzing the user's emotion information; and means for operating the vehicle's infotainment system based on the emotion information. This improves the efficiency of automatic minutes creation of conversations in an autonomous vehicle, generates minutes that reflect the user's emotion, and enables optimal operation of the infotainment system based on emotion.

[0839] "Conversational voice data" refers to data that includes the conversational voices of the driver and passengers.

[0840] "Means for receiving" refers to a device or function that receives audio data using a microphone or digital communication means.

[0841] "Speech recognition means" refers to a technology or system for converting voice data into text data.

[0842] "Natural language processing means" refers to technology or systems that analyze text data, understand the context and meaning, and create minutes.

[0843] "Database management means" refers to the systems and technologies used to store, retrieve, and query minutes and related materials.

[0844] A "comparison means" is a technology or system that compares new text data with stored minutes or materials of related conversations and examines their content.

[0845] The "means for generating final minutes and sending them to the user" refers to a technology or system that generates final minutes based on text data and analysis information and sends them to the user's terminal.

[0846] "Emotion recognition means" refers to technology or a system that analyzes a user's emotions based on text data.

[0847] The "means for operating a vehicle infotainment system" refers to a technology or system that appropriately operates the infotainment system based on the user's emotional information.

[0848] This invention describes a system that automatically converts conversational voice data collected in an autonomous vehicle into minutes, and further adjusts the contents of the minutes by recognizing the user's emotions. This system is operated by a server, an on-board computer, and a user terminal in cooperation with each other.

[0849] Server

[0850] 1. Receiving audio data:

[0851] The server receives the conversational audio data sent from the vehicle's onboard computer via an HTTP POST request, and the server receives the request and saves the audio file in a temporary directory.

[0852] 2. Call the speech recognition engine:

[0853] The server passes the saved audio file to a speech recognition engine (e.g., Google Speech-to-Text API) and converts the audio data into text data. It calls the speech recognition API and passes the file path as an argument.

[0854] 3. Receiving converted data and emotion recognition:

[0855] The server receives the text data returned by the speech recognition engine and stores it in memory. Based on this text data, it uses an emotion recognition engine (for example, IBM Watson Tone Analyzer) to analyze the user's emotions. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words.

[0856] 4. Obtaining relevant data:

[0857] The server uses a database management tool (e.g., MySQL) to retrieve minutes and documents of past relevant conversations using SQL queries, and stores the retrieved data in memory.

[0858] 5. Use of natural language processing engines:

[0859] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to a natural language processing engine (e.g., OpenAI GPT-4) for detailed analysis. The natural language processing engine analyzes the meaning of the data and extracts the necessary information.

[0860] 6. Adjusting meeting minutes based on emotional information:

[0861] The server adjusts the content of the minutes it generates based on the user's emotional information obtained by the emotion recognition engine, for example by emphasizing parts of the minutes in which the user expressed strong emotions.

[0862] 7. Generate and send the final minutes:

[0863] Based on the scrutinized data and adjusted emotional information, the server generates the final transcript, which is generated in text format, re-encoded in JSON format, and sent to the vehicle's computer as an HTTP response.

[0864] Vehicle Computer

[0865] 1. Collection and transmission of conversational audio data:

[0866] The on-board computer collects conversational voice data in real time using the vehicle's microphone and transmits the collected voice data to a server.

[0867] 2. Receiving and viewing minutes:

[0868] Once the final minutes data is returned from the server, the on-board computer receives and analyzes it, and displays the minutes on the vehicle's display in a format that allows the user to intuitively review the contents.

[0869] 3. Infotainment system operation:

[0870] Based on the emotional information provided by the server, the in-vehicle computer operates the infotainment system, for example, playing relaxing music if the user is feeling tense.

[0871] User

[0872] 1. Contribution to audio data collection:

[0873] The user simply has to naturally converse in the car, and the on-board computer automatically collects the voice data.

[0874] 2. Review and edit the minutes:

[0875] The final minutes are then displayed on the vehicle's display, where the user can review them, make edits or add additional comments if necessary, and confirm the changes when they are complete.

[0876] Specific examples

[0877] For example, if a driver and a passenger have a conversation like, "We talked about a new project at yesterday's meeting," the voice data is sent to the server via the in-vehicle computer. The server then converts the voice into text using a speech recognition engine and retrieves minutes of past related conversations from a database. It then scrutinizes the old and new data through natural language processing, and recognizes the user's emotions during the conversation using an emotion recognition engine. For example, if the passenger is nervous, the system will adjust the music played to be more relaxing based on the emotional information. The final minutes are then sent back to the user via the in-vehicle display, where they can review them and make any necessary corrections.

[0878] Example prompts for generative AI models

[0879] "Convert the conversation between the driver and passenger into text and analyze the sentiment."

[0880] "Suggest optimal music choices and rerouting based on conversational content and emotional information."

[0881] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0882] Step 1:

[0883] Audio data collection:

[0884] The terminal (on-board computer) uses the in-car microphone to collect real-time conversational voice data of the driver and passengers. The input is analog voice signals from the in-car microphone, which are converted into digital data and temporarily stored in internal memory. The collected voice data is formatted in batch format for transmission to the server.

[0885] Step 2:

[0886] Sending audio data:

[0887] The device sends the collected and formatted audio data to the server as an HTTP POST request. The input is the formatted audio data, and the output is the data transfer to the server. The audio data sent from the device is saved in a temporary directory on the server.

[0888] Step 3:

[0889] Saving audio data:

[0890] The server saves the audio data sent from the device to a temporary directory. The input is the audio data sent from the device, and the output is the saved audio file. The specified directory path is used to save the file.

[0891] Step 4:

[0892] Calling the speech recognition engine:

[0893] The server passes the saved audio file to a speech recognition engine (e.g., Google Speech-to-Text API) and converts the audio data into text data. The input is the path to the saved audio file, and the output is the text data returned by the speech recognition engine. This text data is stored in memory.

[0894] Step 5:

[0895] Calling the emotion recognition engine:

[0896] The server passes the text data obtained from the speech recognition engine to an emotion recognition engine (for example, IBM Watson Tone Analyzer) for analysis. The input is text data, and the output is analyzed emotional information. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words.

[0897] Step 6:

[0898] Retrieving related data:

[0899] The server uses a database management tool to retrieve minutes and related documents from past conversations. The input is an SQL query, and the output is the retrieved minutes and related documents. The retrieved data is stored in memory.

[0900] Step 7:

[0901] Using natural language processing engines:

[0902] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to a natural language processing engine (e.g., OpenAI GPT-4) for detailed analysis. The input is the text data and related data, and the output is the analyzed content information. The natural language processing engine is used to analyze the meaning of the data and extract the necessary information.

[0903] Step 8:

[0904] Adjusting meeting minutes based on emotional information:

[0905] The server adjusts the content of the meeting minutes it generates based on the emotional information obtained by the emotion recognition engine. The input is text data and emotional information, and the output is the adjusted minutes. For example, it may perform processing such as emphasizing parts of statements that express strong emotions.

[0906] Step 9:

[0907] Generate and send final transcripts:

[0908] The server finally generates the adjusted minutes and encodes them in JSON format again. The input is the adjusted minutes data, and the output is the encoded minutes data. This minutes data is sent to the terminal as an HTTP response.

[0909] Step 10:

[0910] Receive and view minutes:

[0911] The terminal receives the final minutes data from the server and analyzes it. The input is the minutes data from the server, and the output is the analyzed minutes content. The minutes data is displayed on the terminal display and formatted so that the user can intuitively check the content.

[0912] Step 11:

[0913] To operate the infotainment system:

[0914] The device operates the infotainment system based on the emotional information provided by the server. The input is the emotional information from the server, and the output is the system's operation command. For example, if the user is nervous, the device will automatically play relaxing music based on that emotional information.

[0915] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0916] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0917] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0918] [Third embodiment]

[0919] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0920] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0921] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0922] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0923] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0924] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0925] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0926] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0927] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0928] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0929] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0930] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0931] The present invention is a system that automatically converts conference audio data into minutes and quickly provides them. This system is operated in cooperation between a server, terminals, and users.

[0932] Server

[0933] 1. Receiving audio data

[0934] The server receives the conference audio data sent from the terminals. The audio data is usually sent via an HTTP request, and the server receives the request and stores the audio file in a temporary storage location.

[0935] 2. Calling the voice recognition engine

[0936] The received voice data is passed to the speech recognition engine, which converts the data into text data and returns the converted text to the server, which stores the text data in its memory.

[0937] 3. Obtaining relevant data

[0938] The server uses database management means to retrieve minutes and materials from past relevant meetings using SQL queries, and stores the retrieved relevant data in memory.

[0939] 4. Use of natural language processing engines

[0940] The server passes the text data received from the speech recognition engine and related data retrieved from the database to the natural language processing engine, which then analyzes the meaning of the data and extracts the necessary information.

[0941] 5. Generate final transcripts

[0942] Based on the scrutinized data, the server generates the final transcript, which is again encoded in JSON format and sent back to the device.

[0943] Terminal (Client)

[0944] 1. Providing an upload interface

[0945] The terminal provides a user with an interface for uploading conference audio data. This interface is implemented as a GUI including a file selection and upload button.

[0946] 2. Sending audio data

[0947] When a user selects an audio file and presses the upload button, the device sends this audio data to the server as an HTTP POST request.

[0948] 3. Receiving and displaying minutes

[0949] Once the final minutes are returned from the server, the device receives this data and displays it to the user. The minutes data is encoded in JSON format, so the device parses it and displays it on the screen.

[0950] User

[0951] 1. Uploading audio data

[0952] Users upload conference audio data through the interface provided by their devices. They select the appropriate audio file in the file selection dialog and click the upload button.

[0953] 2. Review and edit the minutes

[0954] Once the final minutes are displayed on the device, the user can review them and make edits as needed. Editing is done directly in the text area, and once edits are complete, the changes are saved by pressing the save button.

[0955] Specific examples

[0956] For example, if a user uploads an audio file for a "Project Status Meeting on October 5, 2023," the audio data is sent to the server. The server then converts the audio into text using a speech recognition engine and retrieves minutes of related past meetings from the database. It then scrutinizes the old and new data using natural language processing, generates the final minutes, and sends them back to the device. The user can then review the received minutes and make any necessary corrections, completing the minutes quickly and accurately.

[0957] This system will greatly improve the efficiency of creating meeting minutes, increasing accuracy and convenience.

[0958] The processing flow will be explained below.

[0959] Server

[0960] Step 1:

[0961] The server receives the conference audio data sent from the terminal via an HTTP POST request, and the received audio file is saved in a temporary directory.

[0962] Step 2:

[0963] The server passes the saved audio file to a speech recognition engine and converts the audio data into text data. At this time, it calls the speech recognition API and passes the file path as an argument.

[0964] Step 3:

[0965] The server receives the text data returned by the speech recognition engine and stores it in memory. The returned data is usually provided in JSON format.

[0966] Step 4:

[0967] The server accesses the database to retrieve relevant meeting minutes and materials using SQL queries, and stores the retrieved data in memory.

[0968] Step 5:

[0969] The server calls a natural language processing engine to compare and examine the text data obtained through speech recognition with related data retrieved from the database, thereby extracting important information.

[0970] Step 6:

[0971] The server generates the final minutes based on the scrutinized data, which are generated in text format and then encoded again in JSON format.

[0972] Step 7:

[0973] The server sends the generated minutes to the terminal as an HTTP response.

[0974] Terminal (Client)

[0975] Step 1:

[0976] The terminal provides the user with an interface for uploading audio data, which is designed as a GUI containing a file selection dialog and an upload button.

[0977] Step 2:

[0978] When the user selects an audio file and presses the upload button, the device sends the audio data to the server via an HTTP POST request.

[0979] Step 3:

[0980] When the final minutes data is returned from the server, the terminal receives and parses this JSON format data.

[0981] Step 4:

[0982] The parsed minutes are displayed on the device screen as nicely formatted text so that users can intuitively check the contents.

[0983] User

[0984] Step 1:

[0985] After the conference, the user uploads the audio data through the interface provided by the device by selecting the correct audio file in the file selection dialog and clicking the upload button.

[0986] Step 2:

[0987] Once the upload is complete, the user can check the minutes sent back from the server on their device. The displayed minutes reflect the contents of past meetings, making it easy to confirm important points.

[0988] Step 3:

[0989] The user can check the displayed minutes, make corrections or add comments as necessary, and when the corrections are complete, press the save button to confirm the changes.

[0990] The above processing flow makes it possible to efficiently create meeting minutes and provide accurate and prompt minutes.

[0991] Example 1

[0992] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0993] The process of converting meeting audio data into minutes quickly and accurately is time-consuming and labor-intensive when done manually, making it inefficient. Furthermore, manual minutes are prone to errors and often lack sufficient linkage to past meeting information. This creates a need for improved accuracy in minutes and a more efficient creation process. It is also necessary to provide an interface that makes it easy to upload audio data and edit and review minutes.

[0994] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0995] In this invention, the server includes: means for receiving conference audio data; speech recognition means for converting the received audio data into text data; natural language processing means for creating minutes based on the converted text data; database management means for saving, retrieving, and referencing the created minutes; means for comparing the new text data with stored minutes and materials from related meetings and examining the content; means for generating and transmitting final minutes to users; and means for using a client-server architecture that smoothly transmits and receives audio data, generates, and displays minutes. This improves the efficiency of creating meeting minutes and enables the creation of highly accurate minutes. Users can also easily upload audio files and review and edit the minutes.

[0996] "Conference audio data" refers to digital audio files that record what is said during a conference or meeting.

[0997] "Speech recognition means" refers to technology or equipment for converting voice data into text data. Specifically, it uses a speech recognition engine or API.

[0998] "Natural language processing means" refers to technology or equipment that analyzes text data, understands its meaning and content, and extracts appropriate information. Specifically, it uses natural language processing engines and APIs.

[0999] "Database management means" refers to the technology or device used to store, retrieve, and reference data. Specifically, a database management system (DBMS) is used.

[1000] "Client-server architecture" is a system design model that allows communication between a server that provides data and resources and a client that uses those data and resources.

[1001] A "user" is a person who operates the system and uploads audio data and checks and edits minutes.

[1002] A "terminal" is a device that allows a user to access and operate the system. Specifically, it includes computers, smartphones, etc.

[1003] A "minutes" is a document that records what was said and what was decided at a conference or meeting.

[1004] An "audio file upload interface" is a screen or function that a user uses to submit audio data to the system.

[1005] "Text data" is character information converted by a voice recognition means.

[1006] The "JSON format" is a lightweight data interchange format for storing and exchanging data.

[1007] MODE FOR CARRYING OUT THE INVENTION

[1008] The present invention is a system for automatically converting conference audio data into minutes and providing them promptly. A specific embodiment of this system will now be described in detail.

[1009] server

[1010] The server uses the following main hardware and software to perform the following series of data processing and calculations:

[1011] 1. Receiving audio data

[1012] The server receives the conference audio data sent from the device via an HTTP request, processes the request using a Python framework such as Flask, and temporarily stores the audio file on the server's local disk.

[1013] 2. Calling the voice recognition engine

[1014] The server passes the saved voice data to a speech recognition engine, which converts the voice data to text using the Google Cloud Speech-to-Text API, and stores the converted text in the server's memory.

[1015] 3. Obtaining relevant data

[1016] The server retrieves minutes and materials from past relevant meetings from the MySQL database using SQL queries, including filter conditions such as meeting dates and project names, and stores the retrieved data in memory.

[1017] 4. Use of natural language processing engines

[1018] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to the IBM Watson Natural Language Understanding API, which analyzes the meaning of the data, executes a process to extract the necessary information, and stores the extracted results in memory.

[1019] 5. Generate final transcripts

[1020] The server generates the final minutes based on the scrutinized data. It uses the Python Jinja2 template engine to generate the minutes in JSON format and sends it back to the terminal as an HTTP response.

[1021] Terminal

[1022] The terminal provides an interface for users to upload audio data and review / edit the generated minutes. The following hardware and software are used to perform the operation:

[1023] 1. Providing an upload interface

[1024] The terminal provides a GUI for users to upload conference audio data. An HTML form is used to implement a file selection dialog and an upload button, and JavaScript is used to automatically start uploading after file selection.

[1025] 2. Sending audio data

[1026] When a user selects an audio file and presses the upload button, the device uses JavaScript's Fetch API to send this audio data to the server as an HTTP POST request, which includes the binary data of the selected audio file.

[1027] 3. Receiving and displaying minutes

[1028] Once the final minutes are returned from the server, the device receives this data, retrieves the Fetch API response, parses the response data as JSON, and displays it in the user interface using HTML and JavaScript.

[1029] User

[1030] Users use their devices to upload the meeting audio data and check and edit the generated minutes.

[1031] 1. Uploading audio data

[1032] The user uploads the conference audio data using the interface provided by the terminal: open the file selection dialog, select the audio file to upload, and click the upload button.

[1033] 2. Review and edit the minutes

[1034] The user checks the final minutes displayed on the device and edits them as necessary. They edit the content directly in the displayed text area and press the save button when they are done. The edits are captured by a JavaScript event listener and saved in local storage.

[1035] Specific examples

[1036] For example, if a user uploads an audio file for the "Project Status Meeting on October 5, 2023," the following process takes place: The user selects the audio file from the device's GUI and presses the upload button. The server receives an HTTP POST request using the Flask framework and saves the audio file to its local disk. It then sends a request to a speech recognition engine (Google Cloud Speech-to-Text API) and receives the converted text. Next, it retrieves past meeting minutes data from a MySQL database using filter criteria such as "project name" and "meeting date." This data is passed to the IBM Watson Natural Language Understanding API, where the content is examined and analyzed. Finally, the Jinja2 template engine is used to generate the final meeting minutes in JSON format and return them as an HTTP response. The device receives this response data and displays it in the user interface using HTML and JavaScript. The user reviews the displayed minutes and makes any necessary edits in the text area.

[1037] Prompt Sentence Examples

[1038] An example of a prompt is, "Convert the audio file of the project status meeting on October 5, 2023 into text, and display the minutes that have been reviewed and generated in conjunction with past related meeting minutes. The APIs used are speech recognition API and natural language processing API."

[1039] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1040] Step 1:

[1041] The server receives the conference audio data sent from the terminal. Specifically, the terminal creates an HTTP POST request and sends the binary data of the audio file to the server. The server processes this request using a Python framework such as Flask and temporarily saves the received audio file on the local disk. The input is the audio file sent from the terminal, and the output is the audio file saved on the server.

[1042] Step 2:

[1043] The server passes the saved audio file to the speech recognition engine. Specifically, it creates an API request to use the Google Cloud Speech-to-Text API to convert the audio data to text. The server uses the path to the audio file and the API key as input and receives the converted text data as a response. The output is the text data converted from the audio file.

[1044] Step 3:

[1045] The server retrieves minutes and materials from past relevant meetings from a MySQL database. Specifically, it executes an SQL query that includes filter conditions such as the meeting date and project name. The input is the filter conditions for searching the database, and the output is the retrieved meeting minutes and materials data. This data is stored in memory.

[1046] Step 4:

[1047] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to the natural language processing engine. Specifically, it uses the IBM Watson Natural Language Understanding API to analyze the meaning of the data and extract the necessary information. The input is the speech-recognized text data and related data, and the output is the scrutinized text data.

[1048] Step 5:

[1049] The server generates the final minutes based on the parsed data. Specifically, it uses the Python Jinja2 template engine to generate minutes in JSON format. The input is the parsed text data and template information, and the output is the generated minutes data. This minutes data is returned to the terminal as an HTTP response.

[1050] Step 6:

[1051] The terminal receives the final minutes data returned from the server. Specifically, it receives the HTTP response using the Fetch API and parses the response data as JSON. The input is the JSON minutes data returned from the server, and the output is the parsed minutes data.

[1052] Step 7:

[1053] The terminal displays the parsed minutes data on the user interface. Specifically, it uses HTML and JavaScript to insert the parsed data into a text display area. The input is the parsed minutes data, and the output is the minutes displayed on the screen.

[1054] Step 8:

[1055] The user checks the minutes displayed on the terminal and makes corrections as necessary. Specifically, the user edits the content directly in the displayed text area and clicks the save button when finished. The input is the user's edited content, and the output is the corrected minutes.

[1056] Step 9:

[1057] The terminal sends the edited or revised content by the user to the server and updates the final minutes. Specifically, JavaScript is used to collect the revisions and send them to the server as an HTTP POST request. The input is the minutes data edited by the user, and the output is the minutes data updated on the server. This series of processes allows meeting minutes to be created quickly and accurately.

[1058] (Application example 1)

[1059] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1060] Traditionally, work instructions and reports within factories have often been written by hand or orally, which has led to the problem of time-consuming communication and sharing of information. There is also a high risk of missed records and inaccurate reports. This can lead to reduced work efficiency within the factory and affect productivity. Furthermore, in order to provide quick and accurate work reports, workers and engineers have to spend time and effort to keep records, which increases the workload.

[1061] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1062] In this invention, the server includes means for receiving conference voice data, speech recognition means for converting the received voice data into text data, natural language processing means for creating minutes based on the converted text data, database management means for saving, retrieving, and referencing the created minutes, means for comparing the saved minutes and materials of related meetings with the new text data and examining the content, means for generating final minutes and sending them to the user, means for converting voice recordings of work instructions and reports within the factory into text, and means for uploading the generated minutes to a central system within the factory, thereby enabling quick and accurate recording of work instructions and reports within the factory.

[1063] "Conference voice data" refers to digital data of voice generated during a conference or meeting.

[1064] "Means for receiving audio data" refers to an interface or module for receiving and processing audio data from another device or system.

[1065] "Speech recognition means" refers to technology or equipment that analyzes voice data and converts it into text data.

[1066] "Natural language processing means" is a technology that analyzes the meaning of text data and organizes and extracts data.

[1067] A "database management tool" is a system or software used to efficiently store, retrieve, and access data.

[1068] "Means for examining the content" refers to techniques and methods for analyzing the acquired data in detail and extracting the necessary information.

[1069] "Means for transmitting to users" refers to methods or systems for delivering the final generated data or information to users.

[1070] "Means for converting audio recordings of work instructions and reports within a factory into text" refers to technology that records the audio instructions and reports of factory workers and engineers and converts them into text data.

[1071] "Means for uploading the generated minutes to a central system within the factory" refers to an interface or system for transferring the generated minutes to a central system installed within the factory for storage and management.

[1072] The present invention aims to improve the efficiency of labor in factories and the accuracy of reports. This system is operated through collaboration between a server, a terminal, and a user.

[1073] Server

[1074] The server plays a central role in receiving voice data of work instructions and reports within the factory, and processing and storing that data. Specifically, it performs the following processes:

[1075] Receiving audio data

[1076] The server receives the audio data sent from the device via an HTTP request and temporarily stores the received audio file.

[1077] Calling the speech recognition engine

[1078] The voice data is passed to a speech recognition engine and converted into text data. The speech recognition engine used is something like the Google Cloud Speech-to-Text API.

[1079] Retrieving related data

[1080] The server retrieves reports and documents of past related work from the database, searches using SQL queries, and stores the required data in memory.

[1081] Use of natural language processing engines

[1082] The text data received from the speech recognition engine and related data retrieved from the database are passed to a natural language processing engine for detailed analysis, using engines such as TensorFlow and spaCy.

[1083] Generate and send meeting minutes

[1084] Finally, based on the scrutinized data, the server generates a transcript, which is encoded in JSON format and sent back to the device.

[1085] Terminal (Client)

[1086] The terminal is the interface for the user and is responsible for uploading audio data and displaying received minutes.

[1087] Providing an upload interface

[1088] The terminal provides an audio file upload interface, which is implemented as a GUI (Graphical User Interface) with a file selection and upload button.

[1089] Sending audio data

[1090] After selecting an audio file and clicking the upload button, the device will send the audio file to the server as an HTTP POST request.

[1091] Receiving and viewing minutes

[1092] The minutes returned from the server are received, the JSON format data is parsed, and the data is displayed on the screen, allowing the user to check the minutes and edit them as necessary.

[1093] User

[1094] The user is the end of the system, providing the audio data and checking and editing the final minutes.

[1095] Uploading audio data

[1096] Users use the terminal interface to upload audio data of factory operations and reports.

[1097] Review and edit minutes

[1098] The generated minutes can be checked and edited as necessary. For example, by uploading audio data such as "There is a problem with the right-hand conveyor belt. The motor is overheating and needs to be replaced," minutes can be created quickly and accurately.

[1099] Specific prompt examples

[1100] Generate a transcript from the following audio: "There is a problem with the right-hand conveyor belt. The motor is overheating and needs to be replaced."

[1101] In this way, this system enables efficient and accurate work instructions and reports within the factory. The server is a high-performance computer, the terminals are mobile devices with internet access, and the users are assumed to be factory workers and engineers.

[1102] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1103] Step 1: Recording audio data

[1104] Users record audio data at the factory work site. Specifically, engineers and workers use microphones to record work instructions and reports as audio. This audio data is saved on the device.

[1105] Input: User speech

[1106] Output: Audio file saved on device

[1107] Step 2: Upload your audio data

[1108] The user sends the recorded audio data to the server using the upload interface on the device, by selecting the audio file through the GUI and clicking the upload button.

[1109] Input: Audio file saved on the device

[1110] Output: Audio file sent to the server

[1111] Step 3: Receiving audio data

[1112] The server receives the audio data sent from the device via an HTTP request and temporarily stores the received audio data.

[1113] Input: Audio data sent from the device

[1114] Output: Temporarily saved audio file

[1115] Step 4: Calling the speech recognition engine

[1116] The server passes the temporarily saved audio file to a speech recognition engine and converts the audio data into text data, using a speech recognition service such as the Google Cloud Speech-to-Text API.

[1117] Input: Temporarily saved audio file

[1118] Output: Audio information converted to text data

[1119] Step 5: Retrieve related data

[1120] The server retrieves the minutes and documents of past related work from the database, specifically by using SQL queries to search and retrieve the necessary past data.

[1121] Input: SQL query

[1122] Output: Obtained past minutes and document data

[1123] Step 6: Use a natural language processing engine

[1124] The server then passes the text data obtained from the speech recognition engine and related data retrieved from the database to a natural language processing engine for further analysis, such as using TensorFlow or spaCy to analyze the text data and parse its meaning.

[1125] Input: Text data and historical data

[1126] Output: Scanned text data

[1127] Step 7: Generate and send the minutes

[1128] The server generates the final transcript based on the scrutinized data, which is then encoded in JSON format and sent to the device.

[1129] Input: vetted text data

[1130] Output: JSON formatted minutes

[1131] Step 8: Receive and view the minutes

[1132] The terminal receives the minutes in JSON format sent from the server, parses them, and displays them to the user. Specifically, it displays the contents of the minutes on the screen using a GUI.

[1133] Input: JSON formatted minutes

[1134] Output: Transcript displayed on screen

[1135] Step 9: Review and edit the minutes

[1136] The user checks the minutes displayed on the terminal and edits them as necessary. For example, if there are any clerical errors, the user corrects them and saves them.

[1137] Input: Minutes displayed on screen

[1138] Output: Edited transcript

[1139] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1140] The present invention provides a system that automatically converts conference audio data into minutes and further adjusts the content of the minutes by recognizing the emotions of users. This system is operated in cooperation between a server, terminals, and users.

[1141] Server

[1142] 1. Receiving audio data

[1143] The server receives the conference audio data sent from the terminal. The audio data is sent via an HTTP POST request, and the server receives this request and saves the audio file in a temporary directory.

[1144] 2. Calling the voice recognition engine

[1145] The server passes the saved audio file to the speech recognition engine, converts the audio data into text data, and calls the speech recognition API, passing the file path as an argument.

[1146] 3. Receiving converted data and emotion recognition

[1147] The server receives the text data returned by the speech recognition engine and stores it in memory. Based on this text data, it uses an emotion recognition engine to analyze the user's emotions. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words.

[1148] 4. Acquiring relevant data

[1149] The server uses database management means to retrieve minutes and materials from past relevant meetings using SQL queries, and stores the retrieved data in memory.

[1150] 5. Use of natural language processing engines

[1151] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to the natural language processing engine, which then analyzes the meaning of the data and extracts the necessary information.

[1152] 6. Adjusting meeting minutes based on emotional information

[1153] The server adjusts the content of the minutes it generates based on the user's emotional information obtained by the emotion recognition engine, for example by emphasizing parts of the minutes in which the user expressed strong emotions.

[1154] 7. Generate and send final minutes

[1155] Based on the scrutinized data and adjusted sentiment information, the server generates the final transcript, which is generated in text format, encoded again in JSON format, and sent to the device as an HTTP response.

[1156] Terminal (Client)

[1157] 1. Providing an upload interface

[1158] The terminal provides the user with an interface for uploading audio data, which is implemented as a GUI containing a file selection dialog and an upload button.

[1159] 2. Sending audio data

[1160] When the user selects an audio file and presses the upload button, the device sends the audio data to the server via an HTTP POST request.

[1161] 3. Receiving and displaying minutes

[1162] When the final minutes data is returned from the server, the device receives and parses the JSON format data. The parsed minutes are displayed on the device screen and formatted so that the user can intuitively check the contents.

[1163] User

[1164] 1. Uploading audio data

[1165] After the conference, the user uploads the audio data through the interface provided by the device by selecting the correct audio file in the file selection dialog and clicking the upload button.

[1166] 2. Review and edit the minutes

[1167] Once the final minutes are displayed on the device, the user can review them and make edits or add comments as needed. Edits are made directly in the text area, and once corrections are complete, the changes can be confirmed by pressing the save button.

[1168] Specific examples

[1169] For example, if a user uploads an audio file for a "Project Progress Meeting on October 5, 2023," the audio data is sent to the server. The server then converts the audio into text using a speech recognition engine and retrieves minutes of related past meetings from a database. It then scrutinizes old and new data through natural language processing, and recognizes the user's emotions during the meeting using an emotion recognition engine. For example, if there is an emphasis on a particular topic, the system adjusts the display to highlight that part based on the emotional information. The final minutes are then sent back to the user via their device, who can review them and make any necessary corrections.

[1170] This system makes the creation of meeting minutes significantly more efficient than conventional manual operations, and also makes it possible to generate more detailed minutes that reflect the user's emotions during the meeting.

[1171] The processing flow will be explained below.

[1172] Server

[1173] Step 1:

[1174] The server receives the conference audio data sent from the terminal via an HTTP POST request, and the received audio file is saved in a temporary directory.

[1175] Step 2:

[1176] The server passes the saved audio file to a speech recognition engine and converts the audio data into text data. At this time, it calls the speech recognition API and passes the file path as an argument.

[1177] Step 3:

[1178] The server receives the text data returned by the speech recognition engine and stores it in memory. The returned data is usually provided in JSON format.

[1179] Step 4:

[1180] The server passes this text data to an emotion recognition engine, which analyzes the user's emotions. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words, and stores the results in memory.

[1181] Step 5:

[1182] The server accesses the database to retrieve relevant meeting minutes and materials using SQL queries, and stores the retrieved data in memory.

[1183] Step 6:

[1184] The server invokes a natural language processing engine to compare and examine the text data obtained through speech recognition with related data retrieved from the database, thereby extracting important information.

[1185] Step 7:

[1186] The server adjusts the content of the minutes it generates based on the user's emotional information obtained by the emotion recognition engine, for example by emphasizing parts of the minutes in which the user expressed strong emotions.

[1187] Step 8:

[1188] The server generates the final transcript based on the scrutinized data and adjusted sentiment information. The transcript is generated in text format, then encoded again in JSON format and sent to the device as an HTTP response.

[1189] Terminal (Client)

[1190] Step 1:

[1191] The terminal provides the user with an interface for uploading audio data, which is designed as a GUI containing a file selection dialog and an upload button.

[1192] Step 2:

[1193] When the user selects an audio file and presses the upload button, the device sends the audio data to the server via an HTTP POST request.

[1194] Step 3:

[1195] When the final minutes data is returned from the server, the terminal receives and parses this JSON format data.

[1196] Step 4:

[1197] The parsed minutes are displayed on the device screen as nicely formatted text so that users can intuitively check the contents.

[1198] User

[1199] Step 1:

[1200] After the conference, the user uploads the audio data through the interface provided by the device by selecting the correct audio file in the file selection dialog and clicking the upload button.

[1201] Step 2:

[1202] Once the upload is complete, the user can check the minutes sent back from the server on their device. The displayed minutes reflect the contents of past meetings, making it easy to confirm important points.

[1203] Step 3:

[1204] The user can check the displayed minutes, make corrections or add comments as necessary, and when the corrections are complete, press the save button to confirm the changes.

[1205] Examples:

[1206] When a user uploads an audio file for the "Project Progress Meeting on October 5, 2023," the audio data is sent to the server. The server then converts the audio into text using a speech recognition engine and retrieves minutes of past related meetings from a database. It then scrutinizes old and new data through natural language processing, and recognizes the user's emotions during the meeting using an emotion recognition engine. For example, if there is an emphasis on a particular topic, the system adjusts the display to highlight that part based on the emotional information. The final minutes are then sent back to the user via their device, who can review them and make any necessary corrections.

[1207] Example 2

[1208] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1209] Conventional systems that simply transcribe meeting audio data have the problem of not being able to thoroughly examine the content or reflect emotions, resulting in a decline in the quality of the minutes. Creating minutes that reflect the user's emotions is also a time-consuming and difficult process to do efficiently.

[1210] The specification process by the specification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for receiving voice data, voice recognition means for converting the received voice data into text data, emotion recognition means for analyzing the user's emotion based on the converted text data, database management means for acquiring past related data, natural language processing means for analyzing the acquired related data and the text data obtained from the voice recognition means and examining the content, means for adjusting the content of the minutes based on the emotion information obtained by the emotion recognition means, and means for generating the final minutes and sending them to the user. This makes it possible to automatically and efficiently generate highly accurate minutes that reflect the user's emotions.

[1211] "Audio data" refers to a recording file of a conference that is uploaded by a user, and is data that is converted into text data by a voice recognition means.

[1212] "Speech recognition means" refers to software or an engine for analyzing received voice data and converting it into text data.

[1213] The "emotion recognition means" is software or an engine for analyzing the user's emotions based on the converted text data.

[1214] "Database management means" refers to the means for storing, retrieving, and referencing minutes and materials from past related meetings.

[1215] "Natural language processing means" refers to software or an engine for analyzing and examining the acquired association data and the text data obtained from the speech recognition means.

[1216] The "minutes adjustment means" is a means for adjusting the contents of the minutes to be generated based on the emotion information obtained by the emotion recognition means.

[1217] The "minutes generation means" is a means for generating the final minutes and transmitting them to the user.

[1218] A "terminal" is a device or software that allows a user to upload audio files and view and edit the generated minutes.

[1219] "User" refers to a person who uses this system to upload conference audio data and to review and edit the generated minutes.

[1220] The present invention provides a system that automatically converts conference audio data into minutes and adjusts the content of the minutes by recognizing the emotions of users. This system is operated in cooperation between a server, terminals, and users.

[1221] First, let's explain about terminals. A terminal is a device or software that provides a user with an interface for uploading audio data. Specifically, it is implemented as a GUI (Graphical User Interface) that includes a file selection dialog and an upload button. When a user selects an audio file and presses the upload button, the terminal sends the audio data to the server via an HTTP POST request.

[1222] Next, we will explain the server processing. First, the server receives the audio data sent from the terminal. The audio data is sent via an HTTP POST request, and the server receives this request and saves the audio file in a temporary directory. For example, it may be saved in a path such as " / tmp / meeting_audio.wav".

[1223] The server passes the saved audio file to a speech recognition engine (generally a speech recognition API is used) and converts the audio data into text data. Specifically, it calls the Google Cloud Speech-to-Text API and passes the file path as an argument. The returned text data is stored in memory.

[1224] The server then analyzes the user's emotions using an emotion recognition engine, such as IBM Watson Natural Language Understanding, to determine the user's emotions from the content and word usage of the text data. The analysis results are stored in memory.

[1225] The server then issues SQL queries to a database management system (e.g., MySQL) to retrieve minutes and documents from past related meetings. The retrieved data is also stored in memory. This provides information for comparing past minutes with new text data and examining their content.

[1226] The server then uses a natural language processing engine (such as OpenAI GPT-3) to analyze the text data from the speech recognition engine and related data from the database to extract meaning from the data, a process that makes the content of the meeting clearer.

[1227] The server adjusts the content of the meeting minutes it generates based on the user's emotional information obtained from the emotion recognition engine. For example, it highlights parts of the meeting minutes in which the user expressed strong emotions about a particular topic. Once the adjustments are complete, the server generates the final meeting minutes in text format, encodes them again in JSON format, and sends them to the terminal as an HTTP response.

[1228] Finally, the terminal parses the JSON data of the final minutes received from the server, formats it so that the user can intuitively check the contents, and displays it on the screen. The user can review it and make edits or add comments as needed. This series of operations allows the user to obtain efficient and detailed minutes.

[1229] As a specific example, if a user uploads an audio file for the "Project Status Meeting on October 5, 2023," the audio data is sent to the server. The server converts the audio into text using the Google Cloud Speech-to-Text API and retrieves minutes of past related meetings from a database (MySQL). It then uses OpenAI GPT-3 to scrutinize old and new data, and IBM Watson Natural Language Understanding to recognize the user's emotions during the meeting. For example, if there are emphasized opinions on a particular topic, the system adjusts the display to highlight those parts based on the emotional information. The final minutes are then sent back to the user via their device, who can review them and make any necessary corrections.

[1230] An example of a prompt sentence could be "Please upload the audio data of next week's important meeting and automatically generate the minutes."

[1231] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1232] Step 1: Upload your audio data

[1233] The user accesses the interface provided by the device and selects an audio file. When the user clicks the "Upload" button, the audio file is sent from the device to the server. The audio data is sent to the server using the HTTP POST method, and the audio file is included as part of the request.

[1234] Input: An audio file selected by the user

[1235] Output: Audio data sent from the device to the server

[1236] Specific behavior: The user selects "Project Status Meeting on October 5, 2023.wav" and presses the upload button. This file is sent to the server as an HTTP POST request.

[1237] Step 2: Receiving and storing audio data

[1238] The server receives the HTTP POST request sent from the device and temporarily stores the audio data. When temporarily storing the data, an appropriate path (e.g., " / tmp / 20231005_meeting.wav") is specified.

[1239] Input: Audio data sent from the device

[1240] Output: Audio file saved on the server

[1241] Specific operation: The server saves the received audio file to " / tmp / 20231005_meeting.wav" and records a log that the upload is complete.

[1242] Step 3: Performing speech recognition

[1243] The server passes the path of the saved audio file to a speech recognition engine (for example, Google Cloud Speech-to-Text API) to convert the audio data into text data. A request is sent to the API, and when the request is completed, the converted text data is returned as a response.

[1244] Input: Path to the audio file

[1245] Output: Text data

[1246] Specific operation: The server passes " / tmp / 20231005_meeting.wav" to the Google Cloud Speech-to-Text API and records the log "Speech recognition completed, text data received."

[1247] Step 4: Performing Emotion Recognition

[1248] The server passes the acquired text data to an emotion recognition engine (e.g., IBM Watson Natural Language Understanding) to analyze the user's emotions. Emotional information is returned as the analysis result, and the server stores this in its memory.

[1249] Input: Text data

[1250] Output: Emotional information

[1251] Specific operation: The server passes the text data to IBM Watson and records a log indicating "emotion recognition completed, analysis results received."

[1252] Step 5: Retrieve related data

[1253] The server issues SQL queries to a database management system (e.g., MySQL) to retrieve minutes and materials from past related meetings, and the retrieved data is stored in memory.

[1254] Input: SQL query

[1255] Output: Relevant data retrieved from the database

[1256] What happens: The server executes the SQL query "SELECT FROM meeting_minutes WHERE date < '2023-10-05'" and stores the relevant meeting minutes data in memory.

[1257] Step 6: Performing Natural Language Processing

[1258] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to a natural language processing engine (e.g., OpenAI GPT-3) for further analysis. The natural language processing engine then analyzes the data and extracts meaning.

[1259] Input: Text data, related data

[1260] Output: Refined data

[1261] Specific operation: The server passes the text data and related data to GPT-3, records a log of "natural language analysis started", and stores the refined data in memory.

[1262] Step 7: Adjust the minutes

[1263] The server uses the results of a natural language processing engine to adjust the content of the minutes based on the emotional information obtained from the emotion recognition engine, for example by highlighting parts of the minutes where the user expressed strong emotions about a particular topic.

[1264] Input: Emotional information, refined data

[1265] Output: Adjusted minutes data

[1266] Specific operation: Based on the emotion information, the server highlights specific parts of the minutes and records a log indicating that the minutes have been adjusted.

[1267] Step 8: Generate and send the final transcript

[1268] The server generates the adjusted minutes data in text format, encodes it in JSON format, and then sends it to the terminal via an HTTP response.

[1269] Input: Adjusted minutes data

[1270] Output: Final minutes data in JSON format

[1271] Specific operation: The server saves the final minutes to " / tmp / 20231005_minutes.json" and logs "Minutes generated, ready to send" to the terminal. It then sends it to the terminal as an HTTP response.

[1272] Step 9: Receive and view the minutes

[1273] The terminal analyzes the JSON data of the final minutes received from the server, formats it so that the user can intuitively check the contents, and displays it on the screen.

[1274] Input: JSON format meeting minutes data

[1275] Output: On-screen transcript

[1276] Specific operation: The device parses the JSON data and logs "Minutes received, display started." The minutes are then formatted in an easy-to-understand format and displayed on the screen.

[1277] Step 10: Review and edit the minutes

[1278] The user can check the minutes displayed on the device screen, edit them in the text area, or add comments as needed. Once editing is complete, the user presses the save button to confirm the changes.

[1279] Input: User edits to the minutes

[1280] Output: Final edited transcript

[1281] Specific operation: The user clicks on a specific comment to edit it and presses the "Save" button. A message saying "Editing completed, minutes saved" is displayed on the terminal.

[1282] (Application example 2)

[1283] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1284] Conventional meeting minutes creation systems require manual input and editing, resulting in inefficient work. They also lack the ability to create minutes that reflect the user's emotions, and are unable to properly reflect the nuances and importance of conversations. Furthermore, they lack advanced functionality, such as operating infotainment systems based on conversation content and emotions in autonomous vehicles, making it difficult to improve the comfort of drivers and passengers.

[1285] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for receiving conversation voice data; speech recognition means for converting the received voice data into text data; natural language processing means for creating minutes based on the converted text data; database management means for saving, retrieving, and referencing the created minutes; means for comparing the saved minutes and materials of related conversations with the new text data and examining the content; means for generating final minutes and sending them to the user; emotion recognition means for analyzing the user's emotion information; and means for operating the vehicle's infotainment system based on the emotion information. This improves the efficiency of automatic minutes creation of conversations in an autonomous vehicle, generates minutes that reflect the user's emotion, and enables optimal operation of the infotainment system based on emotion.

[1286] "Conversational voice data" refers to data that includes the conversational voices of the driver and passengers.

[1287] "Means for receiving" refers to a device or function that receives audio data using a microphone or digital communication means.

[1288] "Speech recognition means" refers to a technology or system for converting voice data into text data.

[1289] "Natural language processing means" refers to technology or systems that analyze text data, understand the context and meaning, and create minutes.

[1290] "Database management means" refers to the systems and technologies used to store, retrieve, and query minutes and related materials.

[1291] A "comparison means" is a technology or system that compares new text data with stored minutes or materials of related conversations and examines their content.

[1292] The "means for generating final minutes and sending them to the user" refers to a technology or system that generates final minutes based on text data and analysis information and sends them to the user's terminal.

[1293] "Emotion recognition means" refers to technology or a system that analyzes a user's emotions based on text data.

[1294] The "means for operating a vehicle infotainment system" refers to a technology or system that appropriately operates the infotainment system based on the user's emotional information.

[1295] This invention describes a system that automatically converts conversational voice data collected in an autonomous vehicle into minutes, and further adjusts the contents of the minutes by recognizing the user's emotions. This system is operated by a server, an on-board computer, and a user terminal in cooperation with each other.

[1296] Server

[1297] 1. Receiving audio data:

[1298] The server receives the conversational audio data sent from the vehicle's onboard computer via an HTTP POST request, and the server receives the request and saves the audio file in a temporary directory.

[1299] 2. Call the speech recognition engine:

[1300] The server passes the saved audio file to a speech recognition engine (e.g., Google Speech-to-Text API) and converts the audio data into text data. It calls the speech recognition API and passes the file path as an argument.

[1301] 3. Receiving converted data and emotion recognition:

[1302] The server receives the text data returned by the speech recognition engine and stores it in memory. Based on this text data, it uses an emotion recognition engine (for example, IBM Watson Tone Analyzer) to analyze the user's emotions. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words.

[1303] 4. Obtaining relevant data:

[1304] The server uses a database management tool (e.g., MySQL) to retrieve minutes and documents of past relevant conversations using SQL queries, and stores the retrieved data in memory.

[1305] 5. Use of natural language processing engines:

[1306] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to a natural language processing engine (e.g., OpenAI GPT-4) for detailed analysis. The natural language processing engine analyzes the meaning of the data and extracts the necessary information.

[1307] 6. Adjusting meeting minutes based on emotional information:

[1308] The server adjusts the content of the minutes it generates based on the user's emotional information obtained by the emotion recognition engine, for example by emphasizing parts of the minutes in which the user expressed strong emotions.

[1309] 7. Generate and send the final minutes:

[1310] Based on the scrutinized data and adjusted emotional information, the server generates the final transcript, which is generated in text format, re-encoded in JSON format, and sent to the vehicle's computer as an HTTP response.

[1311] Vehicle Computer

[1312] 1. Collection and transmission of conversational audio data:

[1313] The on-board computer collects conversational voice data in real time using the vehicle's microphone and transmits the collected voice data to a server.

[1314] 2. Receiving and viewing minutes:

[1315] Once the final minutes data is returned from the server, the on-board computer receives and analyzes it, and displays the minutes on the vehicle's display in a format that allows the user to intuitively review the contents.

[1316] 3. Infotainment system operation:

[1317] Based on the emotional information provided by the server, the in-vehicle computer operates the infotainment system, for example, playing relaxing music if the user is feeling tense.

[1318] User

[1319] 1. Contribution to audio data collection:

[1320] The user simply has to naturally converse in the car, and the on-board computer automatically collects the voice data.

[1321] 2. Review and edit the minutes:

[1322] The final minutes are then displayed on the vehicle's display, where the user can review them, make edits or add additional comments if necessary, and confirm the changes when they are complete.

[1323] Specific examples

[1324] For example, if a driver and a passenger have a conversation like, "We talked about a new project at yesterday's meeting," the voice data is sent to the server via the in-vehicle computer. The server then converts the voice into text using a speech recognition engine and retrieves minutes of past related conversations from a database. It then scrutinizes the old and new data through natural language processing, and recognizes the user's emotions during the conversation using an emotion recognition engine. For example, if the passenger is nervous, the system will adjust the music played to be more relaxing based on the emotional information. The final minutes are then sent back to the user via the in-vehicle display, where they can review them and make any necessary corrections.

[1325] Example prompts for generative AI models

[1326] "Convert the conversation between the driver and passenger into text and analyze the sentiment."

[1327] "Suggest optimal music choices and rerouting based on conversational content and emotional information."

[1328] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1329] Step 1:

[1330] Audio data collection:

[1331] The terminal (on-board computer) uses the in-car microphone to collect real-time conversational voice data of the driver and passengers. The input is analog voice signals from the in-car microphone, which are converted into digital data and temporarily stored in internal memory. The collected voice data is formatted in batch format for transmission to the server.

[1332] Step 2:

[1333] Sending audio data:

[1334] The device sends the collected and formatted audio data to the server as an HTTP POST request. The input is the formatted audio data, and the output is the data transfer to the server. The audio data sent from the device is saved in a temporary directory on the server.

[1335] Step 3:

[1336] Saving audio data:

[1337] The server saves the audio data sent from the device to a temporary directory. The input is the audio data sent from the device, and the output is the saved audio file. The specified directory path is used to save the file.

[1338] Step 4:

[1339] Calling the speech recognition engine:

[1340] The server passes the saved audio file to a speech recognition engine (e.g., Google Speech-to-Text API) and converts the audio data into text data. The input is the path to the saved audio file, and the output is the text data returned by the speech recognition engine. This text data is stored in memory.

[1341] Step 5:

[1342] Calling the emotion recognition engine:

[1343] The server passes the text data obtained from the speech recognition engine to an emotion recognition engine (for example, IBM Watson Tone Analyzer) for analysis. The input is text data, and the output is analyzed emotional information. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words.

[1344] Step 6:

[1345] Retrieving related data:

[1346] The server uses a database management tool to retrieve minutes and related documents from past conversations. The input is an SQL query, and the output is the retrieved minutes and related documents. The retrieved data is stored in memory.

[1347] Step 7:

[1348] Using natural language processing engines:

[1349] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to a natural language processing engine (e.g., OpenAI GPT-4) for detailed analysis. The input is the text data and related data, and the output is the analyzed content information. The natural language processing engine is used to analyze the meaning of the data and extract the necessary information.

[1350] Step 8:

[1351] Adjusting meeting minutes based on emotional information:

[1352] The server adjusts the content of the meeting minutes it generates based on the emotional information obtained by the emotion recognition engine. The input is text data and emotional information, and the output is the adjusted minutes. For example, it may perform processing such as emphasizing parts of statements that express strong emotions.

[1353] Step 9:

[1354] Generate and send final transcripts:

[1355] The server finally generates the adjusted minutes and encodes them in JSON format again. The input is the adjusted minutes data, and the output is the encoded minutes data. This minutes data is sent to the terminal as an HTTP response.

[1356] Step 10:

[1357] Receive and view minutes:

[1358] The terminal receives the final minutes data from the server and analyzes it. The input is the minutes data from the server, and the output is the analyzed minutes content. The minutes data is displayed on the terminal display and formatted so that the user can intuitively check the content.

[1359] Step 11:

[1360] To operate the infotainment system:

[1361] The device operates the infotainment system based on the emotional information provided by the server. The input is the emotional information from the server, and the output is the system's operation command. For example, if the user is nervous, the device will automatically play relaxing music based on that emotional information.

[1362] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1363] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1364] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1365] [Fourth embodiment]

[1366] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1367] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1368] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1369] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1370] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1371] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1372] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1373] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1374] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1375] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1376] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1377] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1378] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1379] The present invention is a system that automatically converts conference audio data into minutes and quickly provides them. This system is operated in cooperation between a server, terminals, and users.

[1380] Server

[1381] 1. Receiving audio data

[1382] The server receives the conference audio data sent from the terminals. The audio data is usually sent via an HTTP request, and the server receives the request and stores the audio file in a temporary storage location.

[1383] 2. Calling the voice recognition engine

[1384] The received voice data is passed to the speech recognition engine, which converts the data into text data and returns the converted text to the server, which stores the text data in its memory.

[1385] 3. Obtaining relevant data

[1386] The server uses database management means to retrieve minutes and materials from past relevant meetings using SQL queries, and stores the retrieved relevant data in memory.

[1387] 4. Use of natural language processing engines

[1388] The server passes the text data received from the speech recognition engine and related data retrieved from the database to the natural language processing engine, which then analyzes the meaning of the data and extracts the necessary information.

[1389] 5. Generate final transcripts

[1390] Based on the scrutinized data, the server generates the final transcript, which is again encoded in JSON format and sent back to the device.

[1391] Terminal (Client)

[1392] 1. Providing an upload interface

[1393] The terminal provides a user with an interface for uploading conference audio data. This interface is implemented as a GUI including a file selection and upload button.

[1394] 2. Sending audio data

[1395] When a user selects an audio file and presses the upload button, the device sends this audio data to the server as an HTTP POST request.

[1396] 3. Receiving and displaying minutes

[1397] Once the final minutes are returned from the server, the device receives this data and displays it to the user. The minutes data is encoded in JSON format, so the device parses it and displays it on the screen.

[1398] User

[1399] 1. Uploading audio data

[1400] Users upload conference audio data through the interface provided by their devices. They select the appropriate audio file in the file selection dialog and click the upload button.

[1401] 2. Review and edit the minutes

[1402] Once the final minutes are displayed on the device, the user can review them and make edits as needed. Editing is done directly in the text area, and once edits are complete, the changes are saved by pressing the save button.

[1403] Specific examples

[1404] For example, if a user uploads an audio file for a "Project Status Meeting on October 5, 2023," the audio data is sent to the server. The server then converts the audio into text using a speech recognition engine and retrieves minutes of related past meetings from the database. It then scrutinizes the old and new data using natural language processing, generates the final minutes, and sends them back to the device. The user can then review the received minutes and make any necessary corrections, completing the minutes quickly and accurately.

[1405] This system will greatly improve the efficiency of creating meeting minutes, increasing accuracy and convenience.

[1406] The processing flow will be explained below.

[1407] Server

[1408] Step 1:

[1409] The server receives the conference audio data sent from the terminal via an HTTP POST request, and the received audio file is saved in a temporary directory.

[1410] Step 2:

[1411] The server passes the saved audio file to a speech recognition engine and converts the audio data into text data. At this time, it calls the speech recognition API and passes the file path as an argument.

[1412] Step 3:

[1413] The server receives the text data returned by the speech recognition engine and stores it in memory. The returned data is usually provided in JSON format.

[1414] Step 4:

[1415] The server accesses the database to retrieve relevant meeting minutes and materials using SQL queries, and stores the retrieved data in memory.

[1416] Step 5:

[1417] The server calls a natural language processing engine to compare and examine the text data obtained through speech recognition with related data retrieved from the database, thereby extracting important information.

[1418] Step 6:

[1419] The server generates the final minutes based on the scrutinized data, which are generated in text format and then encoded again in JSON format.

[1420] Step 7:

[1421] The server sends the generated minutes to the terminal as an HTTP response.

[1422] Terminal (Client)

[1423] Step 1:

[1424] The terminal provides the user with an interface for uploading audio data, which is designed as a GUI containing a file selection dialog and an upload button.

[1425] Step 2:

[1426] When the user selects an audio file and presses the upload button, the device sends the audio data to the server via an HTTP POST request.

[1427] Step 3:

[1428] When the final minutes data is returned from the server, the terminal receives and parses this JSON format data.

[1429] Step 4:

[1430] The parsed minutes are displayed on the device screen as nicely formatted text so that users can intuitively check the contents.

[1431] User

[1432] Step 1:

[1433] After the conference, the user uploads the audio data through the interface provided by the device by selecting the correct audio file in the file selection dialog and clicking the upload button.

[1434] Step 2:

[1435] Once the upload is complete, the user can check the minutes sent back from the server on their device. The displayed minutes reflect the contents of past meetings, making it easy to confirm important points.

[1436] Step 3:

[1437] The user can check the displayed minutes, make corrections or add comments as necessary, and when the corrections are complete, press the save button to confirm the changes.

[1438] The above processing flow makes it possible to efficiently create meeting minutes and provide accurate and prompt minutes.

[1439] Example 1

[1440] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1441] The process of converting meeting audio data into minutes quickly and accurately is time-consuming and labor-intensive when done manually, making it inefficient. Furthermore, manual minutes are prone to errors and often lack sufficient linkage to past meeting information. This creates a need for improved accuracy in minutes and a more efficient creation process. It is also necessary to provide an interface that makes it easy to upload audio data and edit and review minutes.

[1442] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1443] In this invention, the server includes: means for receiving conference audio data; speech recognition means for converting the received audio data into text data; natural language processing means for creating minutes based on the converted text data; database management means for saving, retrieving, and referencing the created minutes; means for comparing the new text data with stored minutes and materials from related meetings and examining the content; means for generating and transmitting final minutes to users; and means for using a client-server architecture that smoothly transmits and receives audio data, generates, and displays minutes. This improves the efficiency of creating meeting minutes and enables the creation of highly accurate minutes. Users can also easily upload audio files and review and edit the minutes.

[1444] "Conference audio data" refers to digital audio files that record what is said during a conference or meeting.

[1445] "Speech recognition means" refers to technology or equipment for converting voice data into text data. Specifically, it uses a speech recognition engine or API.

[1446] "Natural language processing means" refers to technology or equipment that analyzes text data, understands its meaning and content, and extracts appropriate information. Specifically, it uses natural language processing engines and APIs.

[1447] "Database management means" refers to the technology or device used to store, retrieve, and reference data. Specifically, a database management system (DBMS) is used.

[1448] "Client-server architecture" is a system design model that allows communication between a server that provides data and resources and a client that uses those data and resources.

[1449] A "user" is a person who operates the system and uploads audio data and checks and edits minutes.

[1450] A "terminal" is a device that allows a user to access and operate the system. Specifically, it includes computers, smartphones, etc.

[1451] A "minutes" is a document that records what was said and what was decided at a conference or meeting.

[1452] An "audio file upload interface" is a screen or function that a user uses to submit audio data to the system.

[1453] "Text data" is character information converted by a voice recognition means.

[1454] The "JSON format" is a lightweight data interchange format for storing and exchanging data.

[1455] MODE FOR CARRYING OUT THE INVENTION

[1456] The present invention is a system for automatically converting conference audio data into minutes and providing them promptly. A specific embodiment of this system will now be described in detail.

[1457] server

[1458] The server uses the following main hardware and software to perform the following series of data processing and calculations:

[1459] 1. Receiving audio data

[1460] The server receives the conference audio data sent from the device via an HTTP request, processes the request using a Python framework such as Flask, and temporarily stores the audio file on the server's local disk.

[1461] 2. Calling the voice recognition engine

[1462] The server passes the saved voice data to a speech recognition engine, which converts the voice data to text using the Google Cloud Speech-to-Text API, and stores the converted text in the server's memory.

[1463] 3. Obtaining relevant data

[1464] The server retrieves minutes and materials from past relevant meetings from the MySQL database using SQL queries, including filter conditions such as meeting dates and project names, and stores the retrieved data in memory.

[1465] 4. Use of natural language processing engines

[1466] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to the IBM Watson Natural Language Understanding API, which analyzes the meaning of the data, executes a process to extract the necessary information, and stores the extracted results in memory.

[1467] 5. Generate final transcripts

[1468] The server generates the final minutes based on the scrutinized data. It uses the Python Jinja2 template engine to generate the minutes in JSON format and sends it back to the terminal as an HTTP response.

[1469] Terminal

[1470] The terminal provides an interface for users to upload audio data and review / edit the generated minutes. The following hardware and software are used to perform the operation:

[1471] 1. Providing an upload interface

[1472] The terminal provides a GUI for users to upload conference audio data. An HTML form is used to implement a file selection dialog and an upload button, and JavaScript is used to automatically start uploading after file selection.

[1473] 2. Sending audio data

[1474] When a user selects an audio file and presses the upload button, the device uses JavaScript's Fetch API to send this audio data to the server as an HTTP POST request, which includes the binary data of the selected audio file.

[1475] 3. Receiving and displaying minutes

[1476] Once the final minutes are returned from the server, the device receives this data, retrieves the Fetch API response, parses the response data as JSON, and displays it in the user interface using HTML and JavaScript.

[1477] User

[1478] Users use their devices to upload the meeting audio data and check and edit the generated minutes.

[1479] 1. Uploading audio data

[1480] The user uploads the conference audio data using the interface provided by the terminal: open the file selection dialog, select the audio file to upload, and click the upload button.

[1481] 2. Review and edit the minutes

[1482] The user checks the final minutes displayed on the device and edits them as necessary. They edit the content directly in the displayed text area and press the save button when they are done. The edits are captured by a JavaScript event listener and saved in local storage.

[1483] Specific examples

[1484] For example, if a user uploads an audio file for the "Project Status Meeting on October 5, 2023," the following process takes place: The user selects the audio file from the device's GUI and presses the upload button. The server receives an HTTP POST request using the Flask framework and saves the audio file to its local disk. It then sends a request to a speech recognition engine (Google Cloud Speech-to-Text API) and receives the converted text. Next, it retrieves past meeting minutes data from a MySQL database using filter criteria such as "project name" and "meeting date." This data is passed to the IBM Watson Natural Language Understanding API, where the content is examined and analyzed. Finally, the Jinja2 template engine is used to generate the final meeting minutes in JSON format and return them as an HTTP response. The device receives this response data and displays it in the user interface using HTML and JavaScript. The user reviews the displayed minutes and makes any necessary edits in the text area.

[1485] Prompt Sentence Examples

[1486] An example of a prompt is, "Convert the audio file of the project status meeting on October 5, 2023 into text, and display the minutes that have been reviewed and generated in conjunction with past related meeting minutes. The APIs used are speech recognition API and natural language processing API."

[1487] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1488] Step 1:

[1489] The server receives the conference audio data sent from the terminal. Specifically, the terminal creates an HTTP POST request and sends the binary data of the audio file to the server. The server processes this request using a Python framework such as Flask and temporarily saves the received audio file on the local disk. The input is the audio file sent from the terminal, and the output is the audio file saved on the server.

[1490] Step 2:

[1491] The server passes the saved audio file to the speech recognition engine. Specifically, it creates an API request to use the Google Cloud Speech-to-Text API to convert the audio data to text. The server uses the path to the audio file and the API key as input and receives the converted text data as a response. The output is the text data converted from the audio file.

[1492] Step 3:

[1493] The server retrieves minutes and materials from past relevant meetings from a MySQL database. Specifically, it executes an SQL query that includes filter conditions such as the meeting date and project name. The input is the filter conditions for searching the database, and the output is the retrieved meeting minutes and materials data. This data is stored in memory.

[1494] Step 4:

[1495] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to the natural language processing engine. Specifically, it uses the IBM Watson Natural Language Understanding API to analyze the meaning of the data and extract the necessary information. The input is the speech-recognized text data and related data, and the output is the scrutinized text data.

[1496] Step 5:

[1497] The server generates the final minutes based on the parsed data. Specifically, it uses the Python Jinja2 template engine to generate minutes in JSON format. The input is the parsed text data and template information, and the output is the generated minutes data. This minutes data is returned to the terminal as an HTTP response.

[1498] Step 6:

[1499] The terminal receives the final minutes data returned from the server. Specifically, it receives the HTTP response using the Fetch API and parses the response data as JSON. The input is the JSON minutes data returned from the server, and the output is the parsed minutes data.

[1500] Step 7:

[1501] The terminal displays the parsed minutes data on the user interface. Specifically, it uses HTML and JavaScript to insert the parsed data into a text display area. The input is the parsed minutes data, and the output is the minutes displayed on the screen.

[1502] Step 8:

[1503] The user checks the minutes displayed on the terminal and makes corrections as necessary. Specifically, the user edits the content directly in the displayed text area and clicks the save button when finished. The input is the user's edited content, and the output is the corrected minutes.

[1504] Step 9:

[1505] The terminal sends the edited or revised content by the user to the server and updates the final minutes. Specifically, JavaScript is used to collect the revisions and send them to the server as an HTTP POST request. The input is the minutes data edited by the user, and the output is the minutes data updated on the server. This series of processes allows meeting minutes to be created quickly and accurately.

[1506] (Application example 1)

[1507] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1508] Traditionally, work instructions and reports within factories have often been written by hand or orally, which has led to the problem of time-consuming communication and sharing of information. There is also a high risk of missed records and inaccurate reports. This can lead to reduced work efficiency within the factory and affect productivity. Furthermore, in order to provide quick and accurate work reports, workers and engineers have to spend time and effort to keep records, which increases the workload.

[1509] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1510] In this invention, the server includes means for receiving conference voice data, speech recognition means for converting the received voice data into text data, natural language processing means for creating minutes based on the converted text data, database management means for saving, retrieving, and referencing the created minutes, means for comparing the saved minutes and materials of related meetings with the new text data and examining the content, means for generating final minutes and sending them to the user, means for converting voice recordings of work instructions and reports within the factory into text, and means for uploading the generated minutes to a central system within the factory, thereby enabling quick and accurate recording of work instructions and reports within the factory.

[1511] "Conference voice data" refers to digital data of voice generated during a conference or meeting.

[1512] "Means for receiving audio data" refers to an interface or module for receiving and processing audio data from another device or system.

[1513] "Speech recognition means" refers to technology or equipment that analyzes voice data and converts it into text data.

[1514] "Natural language processing means" is a technology that analyzes the meaning of text data and organizes and extracts data.

[1515] A "database management tool" is a system or software used to efficiently store, retrieve, and access data.

[1516] "Means for examining the content" refers to techniques and methods for analyzing the acquired data in detail and extracting the necessary information.

[1517] "Means for transmitting to users" refers to methods or systems for delivering the final generated data or information to users.

[1518] "Means for converting audio recordings of work instructions and reports within a factory into text" refers to technology that records the audio instructions and reports of factory workers and engineers and converts them into text data.

[1519] "Means for uploading the generated minutes to a central system within the factory" refers to an interface or system for transferring the generated minutes to a central system installed within the factory for storage and management.

[1520] The present invention aims to improve the efficiency of labor in factories and the accuracy of reports. This system is operated through collaboration between a server, a terminal, and a user.

[1521] Server

[1522] The server plays a central role in receiving voice data of work instructions and reports within the factory, and processing and storing that data. Specifically, it performs the following processes:

[1523] Receiving audio data

[1524] The server receives the audio data sent from the device via an HTTP request and temporarily stores the received audio file.

[1525] Calling the speech recognition engine

[1526] The voice data is passed to a speech recognition engine and converted into text data. The speech recognition engine used is something like the Google Cloud Speech-to-Text API.

[1527] Retrieving related data

[1528] The server retrieves reports and documents of past related work from the database, searches using SQL queries, and stores the required data in memory.

[1529] Use of natural language processing engines

[1530] The text data received from the speech recognition engine and related data retrieved from the database are passed to a natural language processing engine for detailed analysis, using engines such as TensorFlow and spaCy.

[1531] Generate and send meeting minutes

[1532] Finally, based on the scrutinized data, the server generates a transcript, which is encoded in JSON format and sent back to the device.

[1533] Terminal (Client)

[1534] The terminal is the interface for the user and is responsible for uploading audio data and displaying received minutes.

[1535] Providing an upload interface

[1536] The terminal provides an audio file upload interface, which is implemented as a GUI (Graphical User Interface) with a file selection and upload button.

[1537] Sending audio data

[1538] After selecting an audio file and clicking the upload button, the device will send the audio file to the server as an HTTP POST request.

[1539] Receiving and viewing minutes

[1540] The minutes returned from the server are received, the JSON format data is parsed, and the data is displayed on the screen, allowing the user to check the minutes and edit them as necessary.

[1541] User

[1542] The user is the end of the system, providing the audio data and checking and editing the final minutes.

[1543] Uploading audio data

[1544] Users use the terminal interface to upload audio data of factory operations and reports.

[1545] Review and edit minutes

[1546] The generated minutes can be checked and edited as necessary. For example, by uploading audio data such as "There is a problem with the right-hand conveyor belt. The motor is overheating and needs to be replaced," minutes can be created quickly and accurately.

[1547] Specific prompt examples

[1548] Generate a transcript from the following audio: "There is a problem with the right-hand conveyor belt. The motor is overheating and needs to be replaced."

[1549] In this way, this system enables efficient and accurate work instructions and reports within the factory. The server is a high-performance computer, the terminals are mobile devices with internet access, and the users are assumed to be factory workers and engineers.

[1550] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1551] Step 1: Recording audio data

[1552] Users record audio data at the factory work site. Specifically, engineers and workers use microphones to record work instructions and reports as audio. This audio data is saved on the device.

[1553] Input: User speech

[1554] Output: Audio file saved on device

[1555] Step 2: Upload your audio data

[1556] The user sends the recorded audio data to the server using the upload interface on the device, by selecting the audio file through the GUI and clicking the upload button.

[1557] Input: Audio file saved on the device

[1558] Output: Audio file sent to the server

[1559] Step 3: Receiving audio data

[1560] The server receives the audio data sent from the device via an HTTP request and temporarily stores the received audio data.

[1561] Input: Audio data sent from the device

[1562] Output: Temporarily saved audio file

[1563] Step 4: Calling the speech recognition engine

[1564] The server passes the temporarily saved audio file to a speech recognition engine and converts the audio data into text data, using a speech recognition service such as the Google Cloud Speech-to-Text API.

[1565] Input: Temporarily saved audio file

[1566] Output: Audio information converted to text data

[1567] Step 5: Retrieve related data

[1568] The server retrieves the minutes and documents of past related work from the database, specifically by using SQL queries to search and retrieve the necessary past data.

[1569] Input: SQL query

[1570] Output: Obtained past minutes and document data

[1571] Step 6: Use a natural language processing engine

[1572] The server then passes the text data obtained from the speech recognition engine and related data retrieved from the database to a natural language processing engine for further analysis, such as using TensorFlow or spaCy to analyze the text data and parse its meaning.

[1573] Input: Text data and historical data

[1574] Output: Scanned text data

[1575] Step 7: Generate and send the minutes

[1576] The server generates the final transcript based on the scrutinized data, which is then encoded in JSON format and sent to the device.

[1577] Input: vetted text data

[1578] Output: JSON formatted minutes

[1579] Step 8: Receive and view the minutes

[1580] The terminal receives the minutes in JSON format sent from the server, parses them, and displays them to the user. Specifically, it displays the contents of the minutes on the screen using a GUI.

[1581] Input: JSON formatted minutes

[1582] Output: Transcript displayed on screen

[1583] Step 9: Review and edit the minutes

[1584] The user checks the minutes displayed on the terminal and edits them as necessary. For example, if there are any clerical errors, the user corrects them and saves them.

[1585] Input: Minutes displayed on screen

[1586] Output: Edited transcript

[1587] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1588] The present invention provides a system that automatically converts conference audio data into minutes and further adjusts the content of the minutes by recognizing the emotions of users. This system is operated in cooperation between a server, terminals, and users.

[1589] Server

[1590] 1. Receiving audio data

[1591] The server receives the conference audio data sent from the terminal. The audio data is sent via an HTTP POST request, and the server receives this request and saves the audio file in a temporary directory.

[1592] 2. Calling the voice recognition engine

[1593] The server passes the saved audio file to the speech recognition engine, converts the audio data into text data, and calls the speech recognition API, passing the file path as an argument.

[1594] 3. Receiving converted data and emotion recognition

[1595] The server receives the text data returned by the speech recognition engine and stores it in memory. Based on this text data, it uses an emotion recognition engine to analyze the user's emotions. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words.

[1596] 4. Acquiring relevant data

[1597] The server uses database management means to retrieve minutes and materials from past relevant meetings using SQL queries, and stores the retrieved data in memory.

[1598] 5. Use of natural language processing engines

[1599] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to the natural language processing engine, which then analyzes the meaning of the data and extracts the necessary information.

[1600] 6. Adjusting meeting minutes based on emotional information

[1601] The server adjusts the content of the minutes it generates based on the user's emotional information obtained by the emotion recognition engine, for example by emphasizing parts of the minutes in which the user expressed strong emotions.

[1602] 7. Generate and send final minutes

[1603] Based on the scrutinized data and adjusted sentiment information, the server generates the final transcript, which is generated in text format, encoded again in JSON format, and sent to the device as an HTTP response.

[1604] Terminal (Client)

[1605] 1. Providing an upload interface

[1606] The terminal provides the user with an interface for uploading audio data, which is implemented as a GUI containing a file selection dialog and an upload button.

[1607] 2. Sending audio data

[1608] When the user selects an audio file and presses the upload button, the device sends the audio data to the server via an HTTP POST request.

[1609] 3. Receiving and displaying minutes

[1610] When the final minutes data is returned from the server, the device receives and parses the JSON format data. The parsed minutes are displayed on the device screen and formatted so that the user can intuitively check the contents.

[1611] User

[1612] 1. Uploading audio data

[1613] After the conference, the user uploads the audio data through the interface provided by the device by selecting the correct audio file in the file selection dialog and clicking the upload button.

[1614] 2. Review and edit the minutes

[1615] Once the final minutes are displayed on the device, the user can review them and make edits or add comments as needed. Edits are made directly in the text area, and once corrections are complete, the changes can be confirmed by pressing the save button.

[1616] Specific examples

[1617] For example, if a user uploads an audio file for a "Project Progress Meeting on October 5, 2023," the audio data is sent to the server. The server then converts the audio into text using a speech recognition engine and retrieves minutes of related past meetings from a database. It then scrutinizes old and new data through natural language processing, and recognizes the user's emotions during the meeting using an emotion recognition engine. For example, if there is an emphasis on a particular topic, the system adjusts the display to highlight that part based on the emotional information. The final minutes are then sent back to the user via their device, who can review them and make any necessary corrections.

[1618] This system makes the creation of meeting minutes significantly more efficient than conventional manual operations, and also makes it possible to generate more detailed minutes that reflect the user's emotions during the meeting.

[1619] The processing flow will be explained below.

[1620] Server

[1621] Step 1:

[1622] The server receives the conference audio data sent from the terminal via an HTTP POST request, and the received audio file is saved in a temporary directory.

[1623] Step 2:

[1624] The server passes the saved audio file to a speech recognition engine and converts the audio data into text data. At this time, it calls the speech recognition API and passes the file path as an argument.

[1625] Step 3:

[1626] The server receives the text data returned by the speech recognition engine and stores it in memory. The returned data is usually provided in JSON format.

[1627] Step 4:

[1628] The server passes this text data to an emotion recognition engine, which analyzes the user's emotions. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words, and stores the results in memory.

[1629] Step 5:

[1630] The server accesses the database to retrieve relevant meeting minutes and materials using SQL queries, and stores the retrieved data in memory.

[1631] Step 6:

[1632] The server invokes a natural language processing engine to compare and examine the text data obtained through speech recognition with related data retrieved from the database, thereby extracting important information.

[1633] Step 7:

[1634] The server adjusts the content of the minutes it generates based on the user's emotional information obtained by the emotion recognition engine, for example by emphasizing parts of the minutes in which the user expressed strong emotions.

[1635] Step 8:

[1636] The server generates the final transcript based on the scrutinized data and adjusted sentiment information. The transcript is generated in text format, then encoded again in JSON format and sent to the device as an HTTP response.

[1637] Terminal (Client)

[1638] Step 1:

[1639] The terminal provides the user with an interface for uploading audio data, which is designed as a GUI containing a file selection dialog and an upload button.

[1640] Step 2:

[1641] When the user selects an audio file and presses the upload button, the device sends the audio data to the server via an HTTP POST request.

[1642] Step 3:

[1643] When the final minutes data is returned from the server, the terminal receives and parses this JSON format data.

[1644] Step 4:

[1645] The parsed minutes are displayed on the device screen as nicely formatted text so that users can intuitively check the contents.

[1646] User

[1647] Step 1:

[1648] After the conference, the user uploads the audio data through the interface provided by the device by selecting the correct audio file in the file selection dialog and clicking the upload button.

[1649] Step 2:

[1650] Once the upload is complete, the user can check the minutes sent back from the server on their device. The displayed minutes reflect the contents of past meetings, making it easy to confirm important points.

[1651] Step 3:

[1652] The user can check the displayed minutes, make corrections or add comments as necessary, and when the corrections are complete, press the save button to confirm the changes.

[1653] Examples:

[1654] When a user uploads an audio file for the "Project Progress Meeting on October 5, 2023," the audio data is sent to the server. The server then converts the audio into text using a speech recognition engine and retrieves minutes of past related meetings from a database. It then scrutinizes old and new data through natural language processing, and recognizes the user's emotions during the meeting using an emotion recognition engine. For example, if there is an emphasis on a particular topic, the system adjusts the display to highlight that part based on the emotional information. The final minutes are then sent back to the user via their device, who can review them and make any necessary corrections.

[1655] Example 2

[1656] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1657] Conventional systems that simply transcribe meeting audio data have the problem of not being able to thoroughly examine the content or reflect emotions, resulting in a decline in the quality of the minutes. Creating minutes that reflect the user's emotions is also a time-consuming and difficult process to do efficiently.

[1658] The specification process by the specification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for receiving voice data, voice recognition means for converting the received voice data into text data, emotion recognition means for analyzing the user's emotion based on the converted text data, database management means for acquiring past related data, natural language processing means for analyzing the acquired related data and the text data obtained from the voice recognition means and examining the content, means for adjusting the content of the minutes based on the emotion information obtained by the emotion recognition means, and means for generating the final minutes and sending them to the user. This makes it possible to automatically and efficiently generate highly accurate minutes that reflect the user's emotions.

[1659] "Audio data" refers to a recording file of a conference that is uploaded by a user, and is data that is converted into text data by a voice recognition means.

[1660] "Speech recognition means" refers to software or an engine for analyzing received voice data and converting it into text data.

[1661] The "emotion recognition means" is software or an engine for analyzing the user's emotions based on the converted text data.

[1662] "Database management means" refers to the means for storing, retrieving, and referencing minutes and materials from past related meetings.

[1663] "Natural language processing means" refers to software or an engine for analyzing and examining the acquired association data and the text data obtained from the speech recognition means.

[1664] The "minutes adjustment means" is a means for adjusting the contents of the minutes to be generated based on the emotion information obtained by the emotion recognition means.

[1665] The "minutes generation means" is a means for generating the final minutes and transmitting them to the user.

[1666] A "terminal" is a device or software that allows a user to upload audio files and view and edit the generated minutes.

[1667] "User" refers to a person who uses this system to upload conference audio data and to review and edit the generated minutes.

[1668] The present invention provides a system that automatically converts conference audio data into minutes and adjusts the content of the minutes by recognizing the emotions of users. This system is operated in cooperation between a server, terminals, and users.

[1669] First, let's explain about terminals. A terminal is a device or software that provides a user with an interface for uploading audio data. Specifically, it is implemented as a GUI (Graphical User Interface) that includes a file selection dialog and an upload button. When a user selects an audio file and presses the upload button, the terminal sends the audio data to the server via an HTTP POST request.

[1670] Next, we will explain the server processing. First, the server receives the audio data sent from the terminal. The audio data is sent via an HTTP POST request, and the server receives this request and saves the audio file in a temporary directory. For example, it may be saved in a path such as " / tmp / meeting_audio.wav".

[1671] The server passes the saved audio file to a speech recognition engine (generally a speech recognition API is used) and converts the audio data into text data. Specifically, it calls the Google Cloud Speech-to-Text API and passes the file path as an argument. The returned text data is stored in memory.

[1672] The server then analyzes the user's emotions using an emotion recognition engine, such as IBM Watson Natural Language Understanding, to determine the user's emotions from the content and word usage of the text data. The analysis results are stored in memory.

[1673] The server then issues SQL queries to a database management system (e.g., MySQL) to retrieve minutes and documents from past related meetings. The retrieved data is also stored in memory. This provides information for comparing past minutes with new text data and examining their content.

[1674] The server then uses a natural language processing engine (such as OpenAI GPT-3) to analyze the text data from the speech recognition engine and related data from the database to extract meaning from the data, a process that makes the content of the meeting clearer.

[1675] The server adjusts the content of the meeting minutes it generates based on the user's emotional information obtained from the emotion recognition engine. For example, it highlights parts of the meeting minutes in which the user expressed strong emotions about a particular topic. Once the adjustments are complete, the server generates the final meeting minutes in text format, encodes them again in JSON format, and sends them to the terminal as an HTTP response.

[1676] Finally, the terminal parses the JSON data of the final minutes received from the server, formats it so that the user can intuitively check the contents, and displays it on the screen. The user can review it and make edits or add comments as needed. This series of operations allows the user to obtain efficient and detailed minutes.

[1677] As a specific example, if a user uploads an audio file for the "Project Status Meeting on October 5, 2023," the audio data is sent to the server. The server converts the audio into text using the Google Cloud Speech-to-Text API and retrieves minutes of past related meetings from a database (MySQL). It then uses OpenAI GPT-3 to scrutinize old and new data, and IBM Watson Natural Language Understanding to recognize the user's emotions during the meeting. For example, if there are emphasized opinions on a particular topic, the system adjusts the display to highlight those parts based on the emotional information. The final minutes are then sent back to the user via their device, who can review them and make any necessary corrections.

[1678] An example of a prompt sentence could be "Please upload the audio data of next week's important meeting and automatically generate the minutes."

[1679] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1680] Step 1: Upload your audio data

[1681] The user accesses the interface provided by the device and selects an audio file. When the user clicks the "Upload" button, the audio file is sent from the device to the server. The audio data is sent to the server using the HTTP POST method, and the audio file is included as part of the request.

[1682] Input: An audio file selected by the user

[1683] Output: Audio data sent from the device to the server

[1684] Specific behavior: The user selects "Project Status Meeting on October 5, 2023.wav" and presses the upload button. This file is sent to the server as an HTTP POST request.

[1685] Step 2: Receiving and storing audio data

[1686] The server receives the HTTP POST request sent from the device and temporarily stores the audio data. When temporarily storing the data, an appropriate path (e.g., " / tmp / 20231005_meeting.wav") is specified.

[1687] Input: Audio data sent from the device

[1688] Output: Audio file saved on the server

[1689] Specific operation: The server saves the received audio file to " / tmp / 20231005_meeting.wav" and records a log that the upload is complete.

[1690] Step 3: Performing speech recognition

[1691] The server passes the path of the saved audio file to a speech recognition engine (for example, Google Cloud Speech-to-Text API) to convert the audio data into text data. A request is sent to the API, and when the request is completed, the converted text data is returned as a response.

[1692] Input: Path to the audio file

[1693] Output: Text data

[1694] Specific operation: The server passes " / tmp / 20231005_meeting.wav" to the Google Cloud Speech-to-Text API and records the log "Speech recognition completed, text data received."

[1695] Step 4: Performing Emotion Recognition

[1696] The server passes the acquired text data to an emotion recognition engine (e.g., IBM Watson Natural Language Understanding) to analyze the user's emotions. Emotional information is returned as the analysis result, and the server stores this in its memory.

[1697] Input: Text data

[1698] Output: Emotional information

[1699] Specific operation: The server passes the text data to IBM Watson and records a log indicating "emotion recognition completed, analysis results received."

[1700] Step 5: Retrieve related data

[1701] The server issues SQL queries to a database management system (e.g., MySQL) to retrieve minutes and materials from past related meetings, and the retrieved data is stored in memory.

[1702] Input: SQL query

[1703] Output: Relevant data retrieved from the database

[1704] What happens: The server executes the SQL query "SELECT FROM meeting_minutes WHERE date < '2023-10-05'" and stores the relevant meeting minutes data in memory.

[1705] Step 6: Performing Natural Language Processing

[1706] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to a natural language processing engine (e.g., OpenAI GPT-3) for further analysis. The natural language processing engine then analyzes the data and extracts meaning.

[1707] Input: Text data, related data

[1708] Output: Refined data

[1709] Specific operation: The server passes the text data and related data to GPT-3, records a log of "natural language analysis started", and stores the refined data in memory.

[1710] Step 7: Adjust the minutes

[1711] The server uses the results of a natural language processing engine to adjust the content of the minutes based on the emotional information obtained from the emotion recognition engine, for example by highlighting parts of the minutes where the user expressed strong emotions about a particular topic.

[1712] Input: Emotional information, refined data

[1713] Output: Adjusted minutes data

[1714] Specific operation: Based on the emotion information, the server highlights specific parts of the minutes and records a log indicating that the minutes have been adjusted.

[1715] Step 8: Generate and send the final transcript

[1716] The server generates the adjusted minutes data in text format, encodes it in JSON format, and then sends it to the terminal via an HTTP response.

[1717] Input: Adjusted minutes data

[1718] Output: Final minutes data in JSON format

[1719] Specific operation: The server saves the final minutes to " / tmp / 20231005_minutes.json" and logs "Minutes generated, ready to send" to the terminal. It then sends it to the terminal as an HTTP response.

[1720] Step 9: Receive and view the minutes

[1721] The terminal analyzes the JSON data of the final minutes received from the server, formats it so that the user can intuitively check the contents, and displays it on the screen.

[1722] Input: JSON format meeting minutes data

[1723] Output: On-screen transcript

[1724] Specific operation: The device parses the JSON data and logs "Minutes received, display started." The minutes are then formatted in an easy-to-understand format and displayed on the screen.

[1725] Step 10: Review and edit the minutes

[1726] The user can check the minutes displayed on the device screen, edit them in the text area, or add comments as needed. Once editing is complete, the user presses the save button to confirm the changes.

[1727] Input: User edits to the minutes

[1728] Output: Final edited transcript

[1729] Specific operation: The user clicks on a specific comment to edit it and presses the "Save" button. A message saying "Editing completed, minutes saved" is displayed on the terminal.

[1730] (Application example 2)

[1731] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1732] Conventional meeting minutes creation systems require manual input and editing, resulting in inefficient work. They also lack the ability to create minutes that reflect the user's emotions, and are unable to properly reflect the nuances and importance of conversations. Furthermore, they lack advanced functionality, such as operating infotainment systems based on conversation content and emotions in autonomous vehicles, making it difficult to improve the comfort of drivers and passengers.

[1733] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for receiving conversation voice data; speech recognition means for converting the received voice data into text data; natural language processing means for creating minutes based on the converted text data; database management means for saving, retrieving, and referencing the created minutes; means for comparing the saved minutes and materials of related conversations with the new text data and examining the content; means for generating final minutes and sending them to the user; emotion recognition means for analyzing the user's emotion information; and means for operating the vehicle's infotainment system based on the emotion information. This improves the efficiency of automatic minutes creation of conversations in an autonomous vehicle, generates minutes that reflect the user's emotion, and enables optimal operation of the infotainment system based on emotion.

[1734] "Conversational voice data" refers to data that includes the conversational voices of the driver and passengers.

[1735] "Means for receiving" refers to a device or function that receives audio data using a microphone or digital communication means.

[1736] "Speech recognition means" refers to a technology or system for converting voice data into text data.

[1737] "Natural language processing means" refers to technology or systems that analyze text data, understand the context and meaning, and create minutes.

[1738] "Database management means" refers to the systems and technologies used to store, retrieve, and query minutes and related materials.

[1739] A "comparison means" is a technology or system that compares new text data with stored minutes or materials of related conversations and examines their content.

[1740] The "means for generating final minutes and sending them to the user" refers to a technology or system that generates final minutes based on text data and analysis information and sends them to the user's terminal.

[1741] "Emotion recognition means" refers to technology or a system that analyzes a user's emotions based on text data.

[1742] The "means for operating a vehicle infotainment system" refers to a technology or system that appropriately operates the infotainment system based on the user's emotional information.

[1743] This invention describes a system that automatically converts conversational voice data collected in an autonomous vehicle into minutes, and further adjusts the contents of the minutes by recognizing the user's emotions. This system is operated by a server, an on-board computer, and a user terminal in cooperation with each other.

[1744] Server

[1745] 1. Receiving audio data:

[1746] The server receives the conversational audio data sent from the vehicle's onboard computer via an HTTP POST request, and the server receives the request and saves the audio file in a temporary directory.

[1747] 2. Call the speech recognition engine:

[1748] The server passes the saved audio file to a speech recognition engine (e.g., Google Speech-to-Text API) and converts the audio data into text data. It calls the speech recognition API and passes the file path as an argument.

[1749] 3. Receiving converted data and emotion recognition:

[1750] The server receives the text data returned by the speech recognition engine and stores it in memory. Based on this text data, it uses an emotion recognition engine (for example, IBM Watson Tone Analyzer) to analyze the user's emotions. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words.

[1751] 4. Obtaining relevant data:

[1752] The server uses a database management tool (e.g., MySQL) to retrieve minutes and documents of past relevant conversations using SQL queries, and stores the retrieved data in memory.

[1753] 5. Use of natural language processing engines:

[1754] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to a natural language processing engine (e.g., OpenAI GPT-4) for detailed analysis. The natural language processing engine analyzes the meaning of the data and extracts the necessary information.

[1755] 6. Adjusting meeting minutes based on emotional information:

[1756] The server adjusts the content of the minutes it generates based on the user's emotional information obtained by the emotion recognition engine, for example by emphasizing parts of the minutes in which the user expressed strong emotions.

[1757] 7. Generate and send the final minutes:

[1758] Based on the scrutinized data and adjusted emotional information, the server generates the final transcript, which is generated in text format, re-encoded in JSON format, and sent to the vehicle's computer as an HTTP response.

[1759] Vehicle Computer

[1760] 1. Collection and transmission of conversational audio data:

[1761] The on-board computer collects conversational voice data in real time using the vehicle's microphone and transmits the collected voice data to a server.

[1762] 2. Receiving and viewing minutes:

[1763] Once the final minutes data is returned from the server, the on-board computer receives and analyzes it, and displays the minutes on the vehicle's display in a format that allows the user to intuitively review the contents.

[1764] 3. Infotainment system operation:

[1765] Based on the emotional information provided by the server, the in-vehicle computer operates the infotainment system, for example, playing relaxing music if the user is feeling tense.

[1766] User

[1767] 1. Contribution to audio data collection:

[1768] The user simply has to naturally converse in the car, and the on-board computer automatically collects the voice data.

[1769] 2. Review and edit the minutes:

[1770] The final minutes are then displayed on the vehicle's display, where the user can review them, make edits or add additional comments if necessary, and confirm the changes when they are complete.

[1771] Specific examples

[1772] For example, if a driver and a passenger have a conversation like, "We talked about a new project at yesterday's meeting," the voice data is sent to the server via the in-vehicle computer. The server then converts the voice into text using a speech recognition engine and retrieves minutes of past related conversations from a database. It then scrutinizes the old and new data through natural language processing, and recognizes the user's emotions during the conversation using an emotion recognition engine. For example, if the passenger is nervous, the system will adjust the music played to be more relaxing based on the emotional information. The final minutes are then sent back to the user via the in-vehicle display, where they can review them and make any necessary corrections.

[1773] Example prompts for generative AI models

[1774] "Convert the conversation between the driver and passenger into text and analyze the sentiment."

[1775] "Suggest optimal music choices and rerouting based on conversational content and emotional information."

[1776] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1777] Step 1:

[1778] Audio data collection:

[1779] The terminal (on-board computer) uses the in-car microphone to collect real-time conversational voice data of the driver and passengers. The input is analog voice signals from the in-car microphone, which are converted into digital data and temporarily stored in internal memory. The collected voice data is formatted in batch format for transmission to the server.

[1780] Step 2:

[1781] Sending audio data:

[1782] The device sends the collected and formatted audio data to the server as an HTTP POST request. The input is the formatted audio data, and the output is the data transfer to the server. The audio data sent from the device is saved in a temporary directory on the server.

[1783] Step 3:

[1784] Saving audio data:

[1785] The server saves the audio data sent from the device to a temporary directory. The input is the audio data sent from the device, and the output is the saved audio file. The specified directory path is used to save the file.

[1786] Step 4:

[1787] Calling the speech recognition engine:

[1788] The server passes the saved audio file to a speech recognition engine (e.g., Google Speech-to-Text API) and converts the audio data into text data. The input is the path to the saved audio file, and the output is the text data returned by the speech recognition engine. This text data is stored in memory.

[1789] Step 5:

[1790] Calling the emotion recognition engine:

[1791] The server passes the text data obtained from the speech recognition engine to an emotion recognition engine (for example, IBM Watson Tone Analyzer) for analysis. The input is text data, and the output is analyzed emotional information. The emotion recognition engine determines the user's emotions from the content of the text data and the use of words.

[1792] Step 6:

[1793] Retrieving related data:

[1794] The server uses a database management tool to retrieve minutes and related documents from past conversations. The input is an SQL query, and the output is the retrieved minutes and related documents. The retrieved data is stored in memory.

[1795] Step 7:

[1796] Using natural language processing engines:

[1797] The server passes the text data obtained from the speech recognition engine and related data obtained from the database to a natural language processing engine (e.g., OpenAI GPT-4) for detailed analysis. The input is the text data and related data, and the output is the analyzed content information. The natural language processing engine is used to analyze the meaning of the data and extract the necessary information.

[1798] Step 8:

[1799] Adjusting meeting minutes based on emotional information:

[1800] The server adjusts the content of the meeting minutes it generates based on the emotional information obtained by the emotion recognition engine. The input is text data and emotional information, and the output is the adjusted minutes. For example, it may perform processing such as emphasizing parts of statements that express strong emotions.

[1801] Step 9:

[1802] Generate and send final transcripts:

[1803] The server finally generates the adjusted minutes and encodes them in JSON format again. The input is the adjusted minutes data, and the output is the encoded minutes data. This minutes data is sent to the terminal as an HTTP response.

[1804] Step 10:

[1805] Receive and view minutes:

[1806] The terminal receives the final minutes data from the server and analyzes it. The input is the minutes data from the server, and the output is the analyzed minutes content. The minutes data is displayed on the terminal display and formatted so that the user can intuitively check the content.

[1807] Step 11:

[1808] To operate the infotainment system:

[1809] The device operates the infotainment system based on the emotional information provided by the server. The input is the emotional information from the server, and the output is the system's operation command. For example, if the user is nervous, the device will automatically play relaxing music based on that emotional information.

[1810] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1811] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1812] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1813] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1814] FIG. 9 illustrates an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and behaviors arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1815] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1816] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1817] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1818] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1819] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1820] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1821] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1822] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1823] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[1824] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1825] The hardware resource for executing a specific process can be any of the following types of processors: A processor, for example, is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1826] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1827] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1828] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1829] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1830] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1831] The following is further disclosed regarding the above embodiment.

[1832] (Claim 1)

[1833] means for receiving conference audio data;

[1834] a speech recognition means for converting received speech data into text data;

[1835] a natural language processing means for creating minutes based on the converted text data;

[1836] A database management means for storing, retrieving and referencing the minutes prepared;

[1837] A means to compare the new text data with the minutes and materials of the archived related meetings and to examine the content.

[1838] means for generating and transmitting the final minutes to the user;

[1839] A system including:

[1840] (Claim 2)

[1841] 10. The system of claim 1, further comprising a terminal that provides an audio file upload interface for converting the audio data into a final transcript.

[1842] (Claim 3)

[1843] 2. The system according to claim 1, further comprising means for displaying the created minutes on a screen so that the user can check and edit them.

[1844] "Example 1"

[1845] (Claim 1)

[1846] means for receiving conference audio data;

[1847] a speech recognition means for converting received speech data into text data;

[1848] a natural language processing means for creating minutes based on the converted text data;

[1849] A database management means for storing, retrieving and referencing the minutes prepared;

[1850] A means to compare the new text data with the minutes and materials of the archived related meetings and to examine the content.

[1851] means for generating and transmitting the final minutes to the user;

[1852] A means for using a client-server architecture to smoothly transmit audio data, receive audio data, generate transcripts, and display them;

[1853] A system including:

[1854] (Claim 2)

[1855] 10. The system of claim 1, further comprising a terminal that provides an audio file upload interface for converting the audio data into a final transcript.

[1856] (Claim 3)

[1857] 2. The system according to claim 1, further comprising means for displaying the created minutes on a screen so that the user can check and edit them.

[1858] "Application Example 1"

[1859] (Claim 1)

[1860] means for receiving conference audio data;

[1861] a speech recognition means for converting received speech data into text data;

[1862] a natural language processing means for creating minutes based on the converted text data;

[1863] A database management means for storing, retrieving and referencing the minutes prepared;

[1864] A means to compare the new text data with the minutes and materials of the archived related meetings and to examine the content.

[1865] means for generating and transmitting the final minutes to the user;

[1866] A means of converting voice recordings of work instructions and reports within the factory into text,

[1867] a means of uploading the generated minutes to a central system within the factory;

[1868] A system including:

[1869] (Claim 2)

[1870] 10. The system of claim 1, further comprising a terminal that provides an audio file upload interface for converting the audio data into a final transcript.

[1871] (Claim 3)

[1872] 2. The system according to claim 1, further comprising means for displaying the created minutes on a screen so that the user can check and edit them.

[1873] "Example 2: Combining Emotion Engines"

[1874] (Claim 1)

[1875] means for receiving audio data;

[1876] a speech recognition means for converting received speech data into text data;

[1877] emotion recognition means for analyzing the emotion of the user based on the converted text data;

[1878] a database management means for retrieving historical relevant data;

[1879] natural language processing means for analyzing the acquired related data and the ...

Claims

1. means for receiving conference audio data; a speech recognition means for converting received speech data into text data; a natural language processing means for creating minutes based on the converted text data; A database management means for storing, retrieving and referencing the minutes prepared; A means to compare the new text data with the minutes and materials of the archived related meetings and to examine the content. means for generating and transmitting the final minutes to the user; A system including:

2. 10. The system of claim 1, further comprising a terminal that provides an interface for uploading audio files to convert the audio data into final minutes.

3. 2. The system according to claim 1, further comprising means for displaying the created minutes on a screen so that the user can check and edit the minutes.

Images