system

Abstract

Provide a system. [Solution] means for receiving audio data of the conference; means for converting voice data into text data; a means for summarizing text data; a means for formatting the summarized text into minutes format; The system includes a means for transmitting formatted minutes.

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] In conventional meetings, creating minutes is a time-consuming and laborious task. Furthermore, manual transcription and summarization are likely to result in errors and missing information. Furthermore, while there is a need to share meeting content accurately and quickly as minutes, conventional methods have not been able to fully achieve this. The objective of the present invention is to provide a system that solves these problems and enables efficient and accurate creation of minutes. [Means for solving the problem]

[0005] The present invention provides a system that includes a means for receiving audio data of a meeting and converting it into text data, a means for summarizing the text data, a means for formatting the summarized text into a minutes format, and a means for transmitting the formatted minutes. This system uploads the audio data to a server, transcribes the audio using an AI speech recognition engine, and summarizes the text data using a natural language processing engine. Furthermore, the system solves the above-mentioned problems by formatting the summarized text into minutes in a standard format and quickly transmitting the generated minutes to a user.

[0006] "Conference audio data" refers to audio files recorded during a conference that contain participants' comments and discussions.

[0007] The term "receiving means" refers to a combination of hardware and software for receiving audio data from an external device.

[0008] "Text data" refers to data in which voice data is expressed as text information.

[0009] "Means for converting" refers to speech recognition technology for converting voice data into text data.

[0010] "Summarization methods" refer to natural language processing algorithms that shorten text data and extract only the important information.

[0011] A "minutes format" is a standard format for organizing the contents of a meeting and recording them in a standard format.

[0012] "Formatting means" refers to a template engine or format conversion technology that adapts the summarized text to the minutes format.

[0013] "Means of transmission" refers to the communication means or distribution platform used to deliver the formatted minutes to the user's device. [Brief explanation of the drawings]

[0014] [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

[0015] 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.

[0016] First, the terms used in the following description will be explained.

[0017] 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).

[0018] 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.

[0019] 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.

[0020] 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.

[0021] 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."

[0022] [First embodiment]

[0023] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0024] 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.

[0025] 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).

[0026] 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.

[0027] 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.

[0028] 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.

[0029] 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.

[0030] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0031] 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.

[0032] 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.

[0033] 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.

[0034] 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."

[0035] In accordance with an embodiment of the present invention, there is provided a system for efficiently transcribing and summarizing audio data from a conference to create and distribute minutes, which uses a server, user terminals, and a data communication network.

[0036] First, after the conference ends, the user uploads the conference audio data to the server using a dedicated application. The dedicated application is installed on the user's device (PC, smartphone, tablet, etc.) and has the function to specify the audio file and send it to the server.

[0037] Server Operation

[0038] Server: When voice data is received, it is first temporarily stored. Next, the voice data is passed to an AI voice recognition engine to convert it into text data. The AI ​​voice recognition engine used here uses a commercial or open source API (e.g., a general-purpose voice recognition API) that provides highly accurate voice recognition.

[0039] The resulting text data is passed to a natural language processing engine, which then generates a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. The technologies used in this process apply natural language processing algorithms such as topic modeling and key phrase extraction.

[0040] The server then formats the summarized text into minutes, using a template engine to fit the summary text into a predefined minutes format, generating minutes in a standard format that users can easily review and share.

[0041] Finally, the server sends the generated minutes to the user. This process includes sending an email and sending a notification via a dedicated application. A notification is displayed on the user's device, and the generated minutes file (e.g., PDF or Word format) can be downloaded.

[0042] Specific examples

[0043] For example, in the case of a project meeting in which audio data "regarding the progress of development of the next product" is recorded, the following operations are performed.

[0044] 1. User: Uses a dedicated application to upload the meeting recording data to the server.

[0045] 2. Server: Receives the voice data and transcribes it using a commercial speech recognition API. For example, the server can obtain text data such as, "Development of our next product is progressing smoothly. We are currently in the prototype stage, and testing is scheduled to begin next month."

[0046] 3. Server: The obtained text data is passed to a natural language processing engine to generate a summary. The summary result is something like, "Development of the next product is progressing smoothly, and testing is scheduled to begin next month."

[0047] 4. Server: Format the summarized text into a minutes format. For example, it can be formatted into the following format using a template engine.

[0048] ★Agenda: Progress of next product development

[0049] Development is progressing smoothly

[0050] Testing is scheduled to begin next month

[0051] 5. Server: Converts the completed minutes into PDF format and sends them to the user's device. The user receives a notification and can download and view the minutes.

[0052] In this way, the present invention improves the efficiency and accuracy of creating meeting minutes.

[0053] The processing flow will be explained below.

[0054] Step 1:

[0055] User: After the conference ends, the user launches the dedicated application, selects the conference audio data file, and uploads it to the server. When the user clicks the "Upload" button, the audio data is sent to the server as an HTTP request.

[0056] Step 2:

[0057] Server: The server receives the HTTP request and temporarily stores the audio data in storage. Meta information such as the storage location and file name is also recorded in a database.

[0058] Step 3:

[0059] Server: The server passes the saved voice data to the AI ​​voice recognition engine. At this time, it generates an API request and calls the voice recognition API by specifying the file path of the voice data.

[0060] Step 4:

[0061] Server: Receives the transcription results returned from the speech recognition API and saves them as text data. For example, the resulting text data might be something like, "Let's start the meeting. We'll discuss the next project."

[0062] Step 5:

[0063] Server: Passes the transcribed text data to the natural language processing engine and makes an API request to generate a summary, again passing the text data as input.

[0064] Step 6:

[0065] Server: Receives the summary results returned by the natural language processing engine and saves them as text. For example, a summary such as "Meeting begins. Project discussion."

[0066] Step 7:

[0067] Server: The summary text is formatted into a minutes format using a template engine. The summary text is embedded in a pre-prepared template to generate a highly readable format.

[0068] Step 8:

[0069] Server: Converts the formatted minutes into PDF or Word format, creating a file format that users can easily view and share.

[0070] Step 9:

[0071] Server: Sends the generated minutes file to the user's device. A download link is provided to the user via email or a notification function in a dedicated application.

[0072] Step 10:

[0073] User: The user receives a notification on their device and can download and review the generated minutes via the app or email. If necessary, they can make corrections or add additional comments.

[0074] These are the specific steps for transcribing, summarizing, creating, and distributing meeting minutes based on audio data from a meeting. This process allows for efficient and accurate minutes to be created.

[0075] Example 1

[0076] 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."

[0077] Efficiently transcribing and summarizing meeting audio data to create and distribute meeting minutes requires time and effort, and there are issues with the accuracy of the results. It is also often difficult to standardize the format of the generated minutes.

[0078] 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.

[0079] In this invention, the server includes means for receiving conference audio data, means for converting the audio data into text data using highly accurate speech recognition technology, means for summarizing the text data using natural language processing technology, means for formatting the summarized text into a minutes format using a template engine, and means for delivering the formatted minutes to user terminals. This makes it possible to automatically generate minutes from conference audio data efficiently and with high accuracy and deliver them in a unified format.

[0080] "Conference audio data" refers to data in which audio information generated during a conference is recorded in digital format.

[0081] "Speech recognition technology" is a technology for converting speech information into text data, and is a technology that extracts text information from speech using highly accurate algorithms.

[0082] "Natural language processing technology" is a technology for analyzing text data to extract summaries and important information, and uses generative AI models and other algorithms.

[0083] A "template engine" is a software technology for fitting data into a standardized format, and is a technology for formatting any generated text into a pre-specified format.

[0084] A "user terminal" is an electronic device used by a user, such as a computer or smartphone, that can communicate with a server via a network.

[0085] The present invention relates to a system for efficiently transcribing and summarizing audio data from a meeting to create and distribute minutes, which is implemented using a server, user terminals, and a data communication network.

[0086] Hardware and software used

[0087] This system uses the following hardware and software:

[0088] 1. User terminal: Refers to an electronic device that can communicate with the server, such as a computer, smartphone, or tablet.

[0089] 2. Server: A server that contains the hardware required to run speech recognition and natural language processing technologies.

[0090] 3. Data communication network: Internet communication environment for uploading audio data and distributing minutes.

[0091] Specific software used includes:

[0092] 1. Speech Recognition API: Use Google® Speech-to-Text API or other speech recognition APIs.

[0093] 2. Natural language processing engine: Uses OpenAI (registered trademark) GPT-3 (registered trademark) and other natural language processing technologies.

[0094] 3. Template engine: Use a template engine such as Jinja2.

[0095] System action

[0096] The processing of this system is carried out as follows.

[0097] 1. After the conference ends, the user uploads the conference audio data to the server using a dedicated application.

[0098] Using the dedicated software installed on the user's terminal, the user specifies the audio file and clicks the "Upload" button.

[0099] 2. The server receives the audio data and temporarily stores it.

[0100] The server receives the audio files over the network and stores them securely in storage.

[0101] 3. The server uses a speech recognition API to convert the voice data into text data.

[0102] For example, pass an audio file to the Google Speech-to-Text API to generate text data.

[0103] 4. The server passes the text data to a natural language processing engine to generate a summary.

[0104] A generative AI model such as OpenAI GPT-3 is given the prompt "Summarize the key points of this text" to generate a summary.

[0105] 5. The server formats the summarized text into minutes format using a template engine.

[0106] The summary is formatted into a standard format for minutes using the Jinja2 template engine.

[0107] 6. The server distributes the generated minutes to the user terminal.

[0108] The generated minutes (e.g., PDF format) are sent to the user using the email sending function or notification function via a dedicated application.

[0109] Specific examples

[0110] For example, if audio data "regarding the development progress of the next product" is recorded at a project meeting, the process proceeds as follows.

[0111] 1. The user uploads an audio data file (e.g., progress_meeting.wav) using a dedicated application.

[0112] 2. The server receives this audio data and performs character recognition using the Google Speech-to-Text API.

[0113] Example of text data obtained: "Development of our next product is progressing smoothly. It is currently in the prototype stage, and testing is scheduled to begin next month."

[0114] 3. The server passes this text data to OpenAI GPT-3 to generate a summary.

[0115] Example prompt: "Summarize the key points of this passage."

[0116] Generated summary: "Development of next product is progressing well, testing will begin next month."

[0117] 4. The server formats the summary into minutes format using the Jinja2 template engine.

[0118] Example format:

[0119] ★Agenda: Progress of next product development

[0120] Development is progressing smoothly

[0121] Testing is scheduled to begin next month

[0122] 5. The server converts the formatted minutes into PDF format and sends them to the user.

[0123] A notification will be sent to the user's device, and the user can download the PDF minutes and check the contents.

[0124] This makes it possible to automatically generate minutes from meeting audio data efficiently and with high accuracy, and distribute them in a unified format.

[0125] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0126] Step 1:

[0127] After the conference ends, the user uploads the conference audio data to the server using a dedicated application. The user selects the audio data file (e.g., "meeting_record.wav") on the application screen and clicks the "Upload" button.

[0128] Input: Specify the audio data and give instructions to upload it on the user's device.

[0129] Output: The specified audio data is sent to the server via a data communication network.

[0130] Step 2:

[0131] The server temporarily stores the received audio data. The server saves the received file in a specific directory and records the saved path and file name in the database.

[0132] Input: Audio data sent from the user device.

[0133] Output: The audio data is saved to file storage and the information is recorded in a database.

[0134] Step 3:

[0135] The server passes the saved voice data to a voice recognition API and converts it into text data. The server then sends a request to the voice recognition API (e.g., Google Speech-to-Text API) with the path to the audio file as an argument. The API analyzes the voice data and returns the corresponding text data.

[0136] Input: The file path of the saved audio data and the request to the speech recognition API.

[0137] Output: Text data generated from the audio data.

[0138] Step 4:

[0139] The server receives the text data returned by the speech recognition API and stores it in a database. The server inserts this text data into the appropriate tables and records it in a log file.

[0140] Input: Text data returned from the speech recognition API.

[0141] Output: Text data is saved to a database and logged.

[0142] Step 5:

[0143] The server passes the text data stored in the database to a natural language processing engine to generate a summary. The server extracts the text data and passes it to a generative AI model (e.g., OpenAI GPT-3) along with the prompt, "Please summarize the key points of this passage."

[0144] Input: Text data stored in the database and prompt statements.

[0145] Output: A summary returned by the generative AI model.

[0146] Step 6:

[0147] The server receives the generated summary and stores it in a database. The server inserts the summary into the database and records it in a log file.

[0148] Input: A summary returned by the generative AI model.

[0149] Output: The summary sentences are stored in the database and logged.

[0150] Step 7:

[0151] The server passes the abstract to a template engine, which formats it into minutes format. The server passes the abstract to a template engine (e.g., Jinja2) as an argument, which formats it into minutes format.

[0152] Input: Generated abstract and minutes format template.

[0153] Output: Formatted meeting minutes text.

[0154] Step 8:

[0155] The server converts the formatted minutes into PDF format and saves it in file storage. The server passes the text obtained from the template engine to the PDF generation library to generate a PDF file.

[0156] Input: Formatted meeting minutes text.

[0157] Output: Meeting minutes file in PDF format.

[0158] Step 9:

[0159] The server delivers the PDF minutes to the user's device. The server then sends an email with the PDF file attached, and notifies the user through a dedicated application. The user's device receives the notification and displays a download link to the user.

[0160] Input: PDF file of meeting minutes and user contact information.

[0161] Output: A notification will be sent to the user's device, and the minutes will be available for download.

[0162] (Application example 1)

[0163] 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."

[0164] There is a need for a method to efficiently collect and summarize conversations and evaluations during operation of autonomous vehicles, and quickly generate and distribute them as operational reports. Currently, manual recording of conversations during operation and subsequent manual summarization are required, which is extremely time-consuming and labor-intensive. In addition, it is difficult for operation managers to grasp the operational status of multiple vehicles in a timely manner. Therefore, a highly efficient and automated system for recording and summarizing operational conversations is needed.

[0165] 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.

[0166] In this invention, the server includes means for receiving voice data, means for converting the voice data into text data, means for summarizing the text data using a natural language processing engine and formatting it into a standard operation report format, and means for converting the formatted report into PDF format and sending it by email. This makes it possible to automatically collect voice data from inside an autonomous vehicle and quickly generate and distribute summarized operation reports.

[0167] "Audio data" refers to data that records conversations, sounds, etc. in digital format.

[0168] "Text data" is digital data that is obtained by converting voice data into character information.

[0169] A "summary" refers to the extraction of important information from text data and reconstructing it in a short form.

[0170] "Format" refers to the form and layout of a document or data, which is formatted according to specific standards or rules.

[0171] A "report" is a document that summarizes the results and progress of a specific event or activity.

[0172] "Mounted vehicle device" refers to a device installed in an autonomous vehicle that records and uploads voice data.

[0173] A "cloud server" is a remote server provided over the Internet for storing and processing data.

[0174] A "natural language processing engine" is a software engine for analyzing text data and performing high-level processing such as summarization and classification.

[0175] "PDF format" is an abbreviation for Portable Document Format, a document format that can be displayed and printed across platforms.

[0176] "Email" refers to a means of exchanging messages electronically over the Internet.

[0177] In one embodiment of the present invention, a system is provided for automatically recording and summarizing conversations and evaluations during the operation of an autonomous vehicle to generate and distribute operational reports, which uses a cloud server, an in-vehicle device, a user terminal, and a data communication network.

[0178] System Configuration

[0179] 1. Vehicle-mounted equipment:

[0180] The vehicle will be equipped with a device that records voice data inside the vehicle and automatically uploads the audio files to a cloud server, collecting conversations and evaluations that occur during operation in real time.

[0181] 2. Cloud Server:

[0182] When voice data is received, it is first temporarily stored, then passed to an AI voice recognition engine (such as Google Cloud Speech-to-Text) to convert it into text data.

[0183] The resulting text data is passed to a natural language processing engine (such as OpenAI GPT or Google Cloud Natural Language API) to generate a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. The technologies used in this process apply natural language processing algorithms such as topic modeling and key phrase extraction.

[0184] The server then formats the summarized text into an operational report format, using a template engine to fit the summary text into a predefined operational report format, generating a standardized report that can be easily viewed and shared by users.

[0185] Finally, the server converts the generated operation report into PDF format and sends it to the user via email. A notification is sent to the user's terminal, and the generated report file becomes available for download.

[0186] Specific examples

[0187] For example, if audio data is recorded while an autonomous vehicle is operating that says, "Today's operation went very smoothly. Thanks to the new sensor system, obstacle avoidance is faster and more accurate. Fuel efficiency has also improved, resulting in lower fuel costs than expected," the following actions will be taken.

[0188] 1. Vehicle-mounted device: Audio data recorded inside the vehicle is automatically uploaded to a cloud server.

[0189] 2. Cloud server: Receives the voice data and transcribes it using Google Cloud Speech-to-Text. For example, the resulting text might be something like, "Today's operation went very smoothly. Thanks to the new sensor system, obstacle avoidance is now faster and more accurate. Fuel efficiency has also improved, resulting in lower fuel costs than expected."

[0190] 3. Cloud server: The obtained text data is passed to a natural language processing engine such as OpenAI GPT, which generates a summary. The summary result will be something like, "The vehicle travels smoothly, and the sensor system allows for fast and accurate obstacle avoidance. Fuel efficiency has improved, reducing fuel costs."

[0191] 4. Cloud server: The summarized text is formatted into an operational report format. For example, it is formatted into the following format using a template engine.

[0192] ★Operation report:

[0193] The operation is smooth

[0194] Sensor system enables fast and accurate obstacle avoidance

[0195] Improved fuel efficiency and reduced fuel costs

[0196] 5. Cloud server: The completed operation report is converted to PDF format and sent to the user via email. A notification is sent to the user's device, and the user can download and view the report.

[0197] This system will enable automated collection and summarization of conversations and evaluations that occur in real time while autonomous vehicles are operating, and will enable efficient creation and distribution of operational reports.

[0198] Example prompts to input to a generative AI model:

[0199] "Today's operation went very smoothly. Thanks to the new sensor system, obstacle avoidance is faster and more accurate. Fuel efficiency has also improved, resulting in lower fuel costs than expected."

[0200] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0201] Step 1:

[0202] An in-vehicle device collects audio data recorded inside the vehicle, which is then automatically uploaded to a cloud server. The input is the audio inside the vehicle, and the output is the upload of the audio file to the cloud server.

[0203] Step 2:

[0204] The server receives voice data from the vehicle's onboard device. The received voice data is temporarily stored and passed to an AI voice recognition engine for conversion into text data. The input is an audio file, and the output is text data. Specifically, the voice data is converted into text format using a service such as Google Cloud Speech-to-Text.

[0205] Step 3:

[0206] The server passes the converted text data to a natural language processing engine to generate a summary. The input is text data, and the output is summarized text. Specifically, OpenAI GPT and Google Cloud Natural Language API are used to analyze the text data, extract important information, and summarize it.

[0207] Step 4:

[0208] The server formats the summarized text into an operational report format. A template engine is used to fit the summary text into a predefined operational report format. The input is the summarized text, and the output is a formatted operational report. Specifically, the server generates an operational report by inserting the text into a format defined by a tool such as Tex or ReportLab.

[0209] Step 5:

[0210] The server converts the formatted operation report into PDF format and sends it to the user via email. The input is the formatted operation report, and the output is the sent PDF file. Specifically, the PDF file is generated using ReportLab and sent via email using SendGrid or Amazon SES. The user receives a notification and can download and view the generated report file.

[0211] 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.

[0212] As an embodiment of the present invention, we provide a system for efficiently transcribing and summarizing audio data from a meeting, and creating and distributing minutes that incorporate sentiment analysis. This system uses a server, user terminals, and a data communication network. Furthermore, by incorporating an emotion recognition engine, the emotions of the meeting participants can also be reflected in the minutes.

[0213] First, after the conference ends, the user uploads the conference audio data to the server using a dedicated application. The dedicated application is installed on the user's device (PC, smartphone, tablet, etc.) and has the function to specify the audio file and send it to the server.

[0214] Server Operation

[0215] Server: When voice data is received, it is first temporarily stored. Next, the voice data is passed to an AI voice recognition engine to convert it into text data. The AI ​​voice recognition engine used here uses a commercial or open source API (e.g., a general-purpose voice recognition API) that provides highly accurate voice recognition.

[0216] The obtained text data is also passed to the emotion engine, which analyzes the emotions of the meeting participants. The emotion engine recognizes the emotion corresponding to each part of the text data and adds an emotion tag. The technology used in this process applies natural language processing algorithms such as voice emotion recognition and text emotion analysis.

[0217] The server then passes the emotion-tagged text data to a natural language processing engine to generate a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. During this process, the importance and summary content are adjusted based on the emotion data.

[0218] The server formats the summarized text into a minutes format. Using a template engine, the summary text is formatted into a predefined minutes format. Furthermore, sentiment tags and contextual information are added to the text, allowing for emotional changes to be detected.

[0219] Finally, the server sends the generated minutes to the user. This process includes sending an email and sending a notification via a dedicated application. A notification is displayed on the user's device, and the generated minutes file (e.g., PDF or Word format) can be downloaded.

[0220] Specific examples

[0221] For example, in the case of a project meeting in which audio data "regarding the progress of development of the next product" is recorded, the following operations are performed.

[0222] 1. User: Uses a dedicated application to upload the meeting recording data to the server.

[0223] 2. Server: Receives the voice data and transcribes it using a commercial speech recognition API. For example, the server can obtain text data such as, "Development of our next product is progressing smoothly. We are currently in the prototype stage, and testing is scheduled to begin next month."

[0224] 3. Server: The emotion engine is used to analyze emotions from the text data and add tags. For example, the server can obtain emotion-tagged text such as, "Development of the next product is progressing smoothly [Enjoyed]. It is currently in the prototype stage [Neutral], and testing is scheduled to begin next month [Expected]."

[0225] 4. Server: The emotion-tagged text data is passed to a natural language processing engine, which generates a summary. The summary result is something like, "Development of the next product is progressing smoothly, and testing is scheduled to begin next month."

[0226] 5. Server: Format the summarized text into a minutes format. For example, it can be formatted into the following format using a template engine.

[0227] ★Agenda: Progress of next product development

[0228] Development is going smoothly [Joy]

[0229] Testing is expected to begin next month [expected]

[0230] 6. Server: Converts the completed minutes into PDF format and sends them to the user's device. The user receives a notification and can download and view the minutes.

[0231] In this way, the present invention not only improves the efficiency and accuracy of creating meeting minutes, but also provides richer information by reflecting the emotions of the participants.

[0232] The processing flow will be explained below.

[0233] Step 1:

[0234] User: After the conference ends, the user launches the dedicated application, selects the conference audio data file, and uploads it to the server. When the user clicks the "Upload" button, the audio data is sent to the server as an HTTP request.

[0235] Step 2:

[0236] Server: The server receives the HTTP request and temporarily stores the audio data in storage. Meta information such as the storage location and file name is also recorded in a database.

[0237] Step 3:

[0238] Server: The server passes the saved voice data to the AI ​​voice recognition engine. At this time, it generates an API request and calls the voice recognition API by specifying the file path of the voice data.

[0239] Step 4:

[0240] Server: Receives the transcription results returned from the speech recognition API and saves them as text data. For example, the resulting text data might be something like, "Let's start the meeting. We'll discuss the next project."

[0241] Step 5:

[0242] Server: The transcribed text data is passed to the emotion engine for sentiment analysis. The emotion engine recognizes the sentiment associated with each part of the text data and adds emotion tags. For example, emotion-tagged text data such as "Let's start the meeting [Neutral]. We will discuss the next project [Expected]" is obtained.

[0243] Step 6:

[0244] Server: The emotion-tagged text data is passed to a natural language processing engine to generate a summary. The importance of the summary is adjusted based on the emotion data. For example, a summary such as "Project discussion [expectations]" is generated.

[0245] Step 7:

[0246] Server: The summarized text is formatted into minutes format using a template engine. Sentiment tags are also embedded appropriately.

[0247] Step 8:

[0248] Server: Convert the formatted minutes into PDF or Word format. During the conversion, the emotional information is displayed visually in an easy-to-understand manner.

[0249] Step 9:

[0250] Server: Sends the generated minutes file to the user's device. A download link is provided to the user via email or a notification function in a dedicated application.

[0251] Step 10:

[0252] User: The user receives a notification on their device and can download and review the generated minutes via the app or email. If necessary, they can make corrections or add additional comments.

[0253] These are the specific processing steps of the system that combines the emotion engine. This not only improves the efficiency of creating meeting minutes, but also provides rich information that reflects the emotions of the participants.

[0254] Example 2

[0255] 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."

[0256] There is a need to efficiently transcribe and summarize meeting audio data and generate minutes that reflect the emotions of meeting participants. However, there are currently no meeting minutes creation systems that include a means to add emotion tags. As a result, it is difficult to provide rich information that incorporates the emotions of participants rather than simply transcribing the content of meetings.

[0257] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0258] In this invention, the server includes means for receiving conference audio data, means for converting the audio data into text data, means for attaching emotion tags to the text data, means for summarizing the emotion-tagged text data, means for formatting the summarized text into minutes format, and means for transmitting the formatted minutes. This makes it possible to accurately and efficiently record the contents of the conference as minutes and provide rich information that also reflects the emotions of the participants.

[0259] "Conference audio data" means electronic data containing speech and other utterances recorded during a conference.

[0260] "Means for receiving" refers to the function or method for transferring voice data from the user terminal to the server.

[0261] The "means for converting voice data into text data" refers to a method or device that includes voice recognition technology for analyzing voice and converting it into text information.

[0262] "Means for assigning emotion tags to text data" refers to a technology or method that analyzes the emotion of speech in recognized text information and adds appropriate emotion labels.

[0263] A "means for summarizing emotion-tagged text data" is a technology or algorithm that analyzes emotion-tagged text information, extracts important content, and summarizes it concisely.

[0264] "Means for formatting the summarized text into minutes format" refers to a template engine or other formatting method for organizing the summarized content into the prescribed minutes format.

[0265] "Means for sending formatted minutes" includes communication means and processes for delivering the generated minutes to the user via email or a dedicated application.

[0266] A "server" is a computer system that receives, processes, stores, analyzes, and distributes audio data.

[0267] A "user terminal" is a device such as a personal computer, smartphone, or tablet that is used to record and upload audio data to a server.

[0268] A "natural language processing engine" is software that includes artificial intelligence technology that analyzes text data and performs advanced processing such as generating summaries and analyzing sentiment.

[0269] This invention provides a system that efficiently transcribes and summarizes audio data from a conference, and generates and distributes minutes that reflect the emotions of the conference participants. This system uses a server, user terminals, and a data communication network.

[0270] The system operates using the following hardware and software: User terminals include devices such as personal computers, smartphones, and tablets. The server uses commercial or open-source AI speech recognition engines (e.g., general-purpose speech recognition APIs), emotion recognition engines (e.g., emotion analysis APIs using natural language processing technology), natural language processing engines (e.g., generative AI models), and template engines (e.g., Jinja2).

[0271] After the conference ends, users use a dedicated application to upload the conference audio data to the server. The dedicated application is installed on the user's device (PC, smartphone, tablet, etc.) and has the function of selecting an audio file and sending it to a specified server.

[0272] When the server receives the voice data, it first temporarily stores it. Next, it passes the voice data to an AI voice recognition engine to convert it into text data. The resulting text data is also passed to an emotion engine, which analyzes the emotions of the meeting participants. The emotion engine recognizes the emotion corresponding to each part of the text data and adds an emotion tag. The technology used in this process applies natural language processing algorithms such as voice emotion recognition and text emotion analysis.

[0273] The server then inputs the emotion-tagged text data into a natural language processing engine to generate a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. During this process, the importance and summary content are adjusted based on the emotion data. The server then formats the summarized text into a minutes format. Using a template engine, the summary text is fitted into a predefined minutes format. Emotion tags and contextual information are also added, and the text is formatted to show changes in emotion.

[0274] Finally, the server sends the generated minutes to the user. This process includes sending an email and sending a notification via a dedicated application. A notification is displayed on the user's device, and the generated minutes file (e.g., PDF or Word format) can be downloaded.

[0275] Specific examples

[0276] For example, in a project meeting in which audio data on the "progress of development of the next product" is recorded, the following process takes place: The user uses a dedicated application to upload the recording of the meeting (file name: meeting_audio.wav) to the server. The server receives the audio data and transcribes it using a commercial speech recognition API. For example, text data such as "Development of the next product is progressing smoothly. It is currently in the prototype stage, and testing is scheduled to begin next month" is obtained. The server then uses an emotion engine to analyze the emotions from the text data and add tags. For example, emotion-tagged text such as "Development of the next product is progressing smoothly [Joy]. It is currently in the prototype stage [Neutral], and testing is scheduled to begin next month [Expectation]" is obtained. The server then passes the emotion-tagged text data to a natural language processing engine to generate a summary. The summary result is "Development of the next product is progressing smoothly, and testing is scheduled to begin next month." The server then formats the summarized text into meeting minutes format. For example, a template engine can be used to format it into the following format:

[0277] ★Agenda: Progress of next product development

[0278] Development is going smoothly [Joy]

[0279] Testing is expected to begin next month [expected]

[0280] Finally, the server converts the completed minutes into PDF format and sends them to the user's device. The user receives a notification and can download and view the minutes.

[0281] Prompt Sentence Examples

[0282] "Please generate minutes that include the emotions of the participants from audio data of a meeting regarding the development progress of the next product."

[0283] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0284] Step 1:

[0285] The user opens the dedicated application. The user selects the conference audio data file (e.g., meeting_audio.wav) and clicks the upload button. The device sends the selected audio data to the server. The input is the conference audio data, and the output is the audio data sent to the server.

[0286] Step 2:

[0287] The server receives the voice data sent from the device and temporarily stores it. The storage destination is data storage (e.g., storage service), the input is the sent voice data, and the output is the voice data stored in the data storage. Specifically, the server receives an HTTP request and writes binary data to storage.

[0288] Step 3:

[0289] The server passes the saved voice data to the AI ​​voice recognition engine. Specifically, this is the process of sending the voice data to a voice recognition API and converting it into text data. The input is the saved voice data, and the output is text data (e.g., "Development of the next product is going well..."). The server makes an API request and receives text data as a response.

[0290] Step 4:

[0291] The server inputs the acquired text data into the emotion engine for sentiment analysis. The emotion engine recognizes the emotion corresponding to each part of the text and assigns an emotion tag. The input is text data, and the output is emotion-tagged text data (e.g., "Development of the next product is going smoothly [joy]..."). Specifically, it calls the sentiment analysis API and reflects the results in the text.

[0292] Step 5:

[0293] The server passes the emotion-tagged text data to a natural language processing engine, which generates a summary of the text. The natural language processing engine extracts important information based on the emotion tags and creates a summary text. The input is emotion-tagged text data, and the output is summary text (e.g., "Development of the next product is progressing smoothly, and testing is scheduled to begin next month"). Specifically, the server generates a summary using a natural language processing model.

[0294] Step 6:

[0295] The server passes the summarized text to a template engine to format it into minutes of meeting format. A template engine (e.g., Jinja2) is used to fit the summary text into a predefined format. The input is the summary text, and the output is text formatted in minutes of meeting format (e.g., "Development progress of the next product; development is going well [joy]..."). Specifically, the text is input into the template engine, and formatted text is generated.

[0296] Step 7:

[0297] The server converts the formatted minutes into PDF format. A PDF generation program (e.g., ReportLab) is used to create a PDF from text data. The input is text formatted in minutes format, and the output is a PDF minutes file. Specifically, the server converts the text into PDF using a PDF generation library.

[0298] Step 8:

[0299] The server sends the generated PDF to the user. The server uses the email sending function or the notification function of a dedicated application. The input is a PDF format minutes file, and the output is a notification or email to the user's device. Specifically, the file is sent using the email sending API or push notification function.

[0300] Step 9:

[0301] The terminal receives the notification from the server and displays it to the user. The user checks the notification, clicks the provided link to download the minutes, and checks the contents. The input is the notification sent from the server, and the output is the minutes file downloaded to the user. Specifically, the terminal displays the notification message and allows the user to operate the link.

[0302] (Application example 2)

[0303] 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."

[0304] Previously, creating meeting minutes was often done manually, which was inefficient and time-consuming. It was also difficult to create minutes that reflected the emotions of the meeting participants, making it difficult to accurately capture the atmosphere and key points of the meeting. This made it difficult to share information in real time during operational meetings at factories and other locations, which sometimes hindered efficient decision-making.

[0305] 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 conference audio data, means for converting the audio data into text data, means for summarizing the text data, means for formatting the summarized text into a minutes format, means for performing sentiment analysis, means for adding sentiment tags, and means for transmitting the formatted minutes. This enables the automated and efficient creation of conference minutes, and further enables the creation of minutes that reflect the emotions of the conference participants. This enables rapid and accurate information sharing and promotes efficient decision-making.

[0306] "Meeting audio data" refers to audio files that record statements and discussions made during a meeting.

[0307] A "means for receiving" refers to a system or method for obtaining data from another device or network.

[0308] "Text data" refers to character string data converted from voice data by a voice recognition system.

[0309] A "summarization means" is a method or system for extracting important information from given text data and summarizing it concisely.

[0310] A "minutes format" is a predefined format for organizing and recording the contents of a meeting.

[0311] A "formatting means" is a method or system for converting and preparing data or information into a particular form or format.

[0312] A "transmitting means" is a system or method for sending data to another device or network.

[0313] "Sentiment analysis" is a technology that detects emotions from text data or audio data and evaluates the type and intensity of those emotions.

[0314] An "emotion tag" is a label or mark that indicates the emotion contained within a piece of text.

[0315] As an embodiment of this invention, we will explain an example of an "operational meeting support app for factory robots" installed on factory robots. The following explanation is based on the process of efficiently collecting meeting audio data, transcribing it, summarizing it, and analyzing its sentiment, and finally generating and distributing meeting minutes with sentiment tags.

[0316] System Configuration

[0317] The system uses factory robots, servers, and a data communication network. The factory robots are equipped with built-in microphones that collect audio data from meetings. The server processes the audio data and uses a highly accurate speech recognition API, emotion recognition engine, and natural language processing engine. The main software and hardware components are described below.

[0318] Hardware

[0319] Factory robots: Collect audio data using built-in microphones.

[0320] Server: A high-performance server used for data processing and storage.

[0321] software

[0322] Speech Recognition API: A commercial or open-source API for high-accuracy speech recognition (e.g., Google Speech-to-Text API).

[0323] Emotion recognition engine: Technology for analyzing emotions in text data and adding tags (e.g., Microsoft® Azure® Cognitive Services' Text Analytics).

[0324] Natural language processing engine: A technology for summarizing text data (e.g., OpenAI GPT-3).

[0325] Template engine: Software for formatting summary text into minutes format.

[0326] PDF generation software: Software to convert minutes into PDF format (e.g. pdfkit).

[0327] Processing flow

[0328] 1. Collection of audio data

[0329] The factory robots are equipped with built-in microphones to collect audio data from meetings in real time, and the audio files are temporarily stored inside the robots and then sent to a server via a network.

[0330] 2. Transcription

[0331] When the server receives the audio data, it converts it into text using the Google Cloud Speech-to-Text API, allowing the contents of the meeting to be recorded as text.

[0332] 3. Sentiment analysis

[0333] The resulting text data is passed to the Microsoft Azure Cognitive Services Text Analytics API for emotional tagging, which recognizes the emotions (happiness, sadness, anticipation, etc.) that correspond to each part of the text data and adds emotional tags.

[0334] 4. Summary Generation

[0335] The sentiment-tagged text data is then summarised using OpenAI GPT-3, which performs text analysis to extract and summarise key information concisely.

[0336] 5. Formatting of minutes

[0337] The generated summary and emotion tags are then formatted into a predefined minutes format using a template engine, allowing participants to understand the contents of the meeting, including changes in emotion, at a glance.

[0338] 6. Results Distribution

[0339] The final minutes are converted to PDF format using PDF generation software, then uploaded to email or cloud storage and notified to the relevant parties.

[0340] Specific examples

[0341] For example, if the following audio data is collected:

[0342] "Development of our next product is progressing smoothly and is currently in the prototype stage. Testing is scheduled to begin next month. All team members are highly motivated."

[0343] An example prompt based on this is:

[0344] Example prompt sentence:

[0345] Generate a summary for the following text: "Development of our next product is progressing smoothly and is currently in the prototype stage. Testing is scheduled to begin next month. All team members are highly motivated."

[0346] The final summary results are as follows:

[0347] "Development of the next product is progressing smoothly, and testing is scheduled to begin next month. The entire team is highly motivated."

[0348] As described above, the present invention automates and improves the efficiency of creating meeting minutes, and also provides information that reflects the emotions of the participants, thereby realizing richer information sharing.

[0349] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0350] Step 1:

[0351] The user collects audio data from meetings in real time using the factory robot's built-in microphone. The collected audio data is temporarily stored inside the robot. The audio data is then sent to a server via a network. The input is audio data, and the output is an audio file sent to the server.

[0352] Step 2:

[0353] The server receives the voice data and temporarily stores it. The received voice data is passed to a high-precision voice recognition API (e.g., Google Speech-to-Text API), which converts the voice data into text data. The input is voice data, and the output is transcribed text data. Specifically, the server sends a voice recognition request, and the API analyzes the voice data and converts it into text.

[0354] Step 3:

[0355] The server passes the obtained text data to an emotion recognition engine (for example, the Text Analytics API from Microsoft Azure Cognitive Services) to analyze the emotions in the text data. The emotion recognition engine detects the emotions corresponding to each part of the text and adds emotion tags. The input is text data, and the output is emotion-tagged text data. Specifically, the server sends an emotion analysis request, and the API analyzes the text and adds emotion tags.

[0356] Step 4:

[0357] The server passes emotion-tagged text data to a natural language processing engine (e.g., OpenAI GPT-3) to extract key information and generate a summary. The input is emotion-tagged text data, and the output is the summary text. Specifically, a summary generation request is sent, and the generative AI model analyzes the text and creates a summary.

[0358] Step 5:

[0359] The server passes the summarized text and emotion tags to a template engine to format it into a predefined minutes format. The input is the summary text and emotion-tagged text data, and the output is text formatted in the minutes format. Specifically, the template engine rearranges the text data into a predetermined format and performs additional processing to visually display the emotion tag information.

[0360] Step 6:

[0361] The server converts the formatted minutes into PDF format and sends it to the user's terminal. The input is text formatted in the minutes format, and the output is a PDF minutes file. Specifically, the server converts the text data into a PDF file using PDF generation software (e.g., pdfkit), and then sends the PDF file by email or uploads it to cloud storage.

[0362] These are the processing steps of the factory robot operation meeting support app, which efficiently transcribes the audio data of meetings, automatically generates and distributes meeting minutes that have been summarized and analyzed for sentiment.

[0363] 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.

[0364] 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.

[0365] 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.

[0366] [Second embodiment]

[0367] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0368] 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.

[0369] 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).

[0370] 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.

[0371] 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.

[0372] 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).

[0373] 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.

[0374] 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.

[0375] 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.

[0376] 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.

[0377] 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.

[0378] 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."

[0379] In accordance with an embodiment of the present invention, there is provided a system for efficiently transcribing and summarizing audio data from a conference to create and distribute minutes, which uses a server, user terminals, and a data communication network.

[0380] First, after the conference ends, the user uploads the conference audio data to the server using a dedicated application. The dedicated application is installed on the user's device (PC, smartphone, tablet, etc.) and has the function to specify the audio file and send it to the server.

[0381] Server Operation

[0382] Server: When voice data is received, it is first temporarily stored. Next, the voice data is passed to an AI voice recognition engine to convert it into text data. The AI ​​voice recognition engine used here uses a commercial or open source API (e.g., a general-purpose voice recognition API) that provides highly accurate voice recognition.

[0383] The resulting text data is passed to a natural language processing engine, which then generates a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. The technologies used in this process apply natural language processing algorithms such as topic modeling and key phrase extraction.

[0384] The server then formats the summarized text into minutes, using a template engine to fit the summary text into a predefined minutes format, generating minutes in a standard format that users can easily review and share.

[0385] Finally, the server sends the generated minutes to the user. This process includes sending an email and sending a notification via a dedicated application. A notification is displayed on the user's device, and the generated minutes file (e.g., PDF or Word format) can be downloaded.

[0386] Specific examples

[0387] For example, in the case of a project meeting in which audio data "regarding the progress of development of the next product" is recorded, the following operations are performed.

[0388] 1. User: Uses a dedicated application to upload the meeting recording data to the server.

[0389] 2. Server: Receives the voice data and transcribes it using a commercial speech recognition API. For example, the server can obtain text data such as, "Development of our next product is progressing smoothly. We are currently in the prototype stage, and testing is scheduled to begin next month."

[0390] 3. Server: The obtained text data is passed to a natural language processing engine to generate a summary. The summary result is something like, "Development of the next product is progressing smoothly, and testing is scheduled to begin next month."

[0391] 4. Server: Format the summarized text into a minutes format. For example, it can be formatted into the following format using a template engine.

[0392] ★Agenda: Progress of next product development

[0393] Development is progressing smoothly

[0394] Testing is scheduled to begin next month

[0395] 5. Server: Converts the completed minutes into PDF format and sends them to the user's device. The user receives a notification and can download and view the minutes.

[0396] In this way, the present invention improves the efficiency and accuracy of creating meeting minutes.

[0397] The processing flow will be explained below.

[0398] Step 1:

[0399] User: After the conference ends, the user launches the dedicated application, selects the conference audio data file, and uploads it to the server. When the user clicks the "Upload" button, the audio data is sent to the server as an HTTP request.

[0400] Step 2:

[0401] Server: The server receives the HTTP request and temporarily stores the audio data in storage. Meta information such as the storage location and file name is also recorded in a database.

[0402] Step 3:

[0403] Server: The server passes the saved voice data to the AI ​​voice recognition engine. At this time, it generates an API request and calls the voice recognition API by specifying the file path of the voice data.

[0404] Step 4:

[0405] Server: Receives the transcription results returned from the speech recognition API and saves them as text data. For example, the resulting text data might be something like, "Let's start the meeting. We'll discuss the next project."

[0406] Step 5:

[0407] Server: Passes the transcribed text data to the natural language processing engine and makes an API request to generate a summary, again passing the text data as input.

[0408] Step 6:

[0409] Server: Receives the summary results returned by the natural language processing engine and saves them as text. For example, a summary such as "Meeting begins. Project discussion."

[0410] Step 7:

[0411] Server: The summary text is formatted into a minutes format using a template engine. The summary text is embedded in a pre-prepared template to generate a highly readable format.

[0412] Step 8:

[0413] Server: Converts the formatted minutes into PDF or Word format, creating a file format that users can easily view and share.

[0414] Step 9:

[0415] Server: Sends the generated minutes file to the user's device. A download link is provided to the user via email or a notification function in a dedicated application.

[0416] Step 10:

[0417] User: The user receives a notification on their device and can download and review the generated minutes via the app or email. If necessary, they can make corrections or add additional comments.

[0418] These are the specific steps for transcribing, summarizing, creating, and distributing meeting minutes based on audio data from a meeting. This process allows for efficient and accurate minutes to be created.

[0419] Example 1

[0420] 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."

[0421] Efficiently transcribing and summarizing meeting audio data to create and distribute meeting minutes requires time and effort, and there are issues with the accuracy of the results. It is also often difficult to standardize the format of the generated minutes.

[0422] 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.

[0423] In this invention, the server includes means for receiving conference audio data, means for converting the audio data into text data using highly accurate speech recognition technology, means for summarizing the text data using natural language processing technology, means for formatting the summarized text into a minutes format using a template engine, and means for delivering the formatted minutes to user terminals. This makes it possible to automatically generate minutes from conference audio data efficiently and with high accuracy and deliver them in a unified format.

[0424] "Conference audio data" refers to data in which audio information generated during a conference is recorded in digital format.

[0425] "Speech recognition technology" is a technology for converting speech information into text data, and is a technology that extracts text information from speech using highly accurate algorithms.

[0426] "Natural language processing technology" is a technology for analyzing text data to extract summaries and important information, and uses generative AI models and other algorithms.

[0427] A "template engine" is a software technology for fitting data into a standardized format, and is a technology for formatting any generated text into a pre-specified format.

[0428] A "user terminal" is an electronic device used by a user, such as a computer or smartphone, that can communicate with a server via a network.

[0429] The present invention relates to a system for efficiently transcribing and summarizing audio data from a meeting to create and distribute minutes, which is implemented using a server, user terminals, and a data communication network.

[0430] Hardware and software used

[0431] This system uses the following hardware and software:

[0432] 1. User terminal: Refers to an electronic device that can communicate with the server, such as a computer, smartphone, or tablet.

[0433] 2. Server: A server that contains the hardware required to run speech recognition and natural language processing technologies.

[0434] 3. Data communication network: Internet communication environment for uploading audio data and distributing minutes.

[0435] Specific software used includes:

[0436] 1. Speech Recognition API: Use Google Speech-to-Text API or other speech recognition APIs.

[0437] 2. Natural Language Processing Engine: Uses OpenAI GPT-3 and other natural language processing technologies.

[0438] 3. Template engine: Use a template engine such as Jinja2.

[0439] System action

[0440] The processing of this system is carried out as follows.

[0441] 1. After the conference ends, the user uploads the conference audio data to the server using a dedicated application.

[0442] Using the dedicated software installed on the user's terminal, the user specifies the audio file and clicks the "Upload" button.

[0443] 2. The server receives the audio data and temporarily stores it.

[0444] The server receives the audio files over the network and stores them securely in storage.

[0445] 3. The server uses a speech recognition API to convert the voice data into text data.

[0446] For example, pass an audio file to the Google Speech-to-Text API to generate text data.

[0447] 4. The server passes the text data to a natural language processing engine to generate a summary.

[0448] A generative AI model such as OpenAI GPT-3 is given the prompt "Summarize the key points of this text" to generate a summary.

[0449] 5. The server formats the summarized text into minutes format using a template engine.

[0450] The summary is formatted into a standard format for minutes using the Jinja2 template engine.

[0451] 6. The server distributes the generated minutes to the user terminal.

[0452] The generated minutes (e.g., PDF format) are sent to the user using the email sending function or notification function via a dedicated application.

[0453] Specific examples

[0454] For example, if audio data "regarding the development progress of the next product" is recorded at a project meeting, the process proceeds as follows.

[0455] 1. The user uploads an audio data file (e.g., progress_meeting.wav) using a dedicated application.

[0456] 2. The server receives this audio data and performs character recognition using the Google Speech-to-Text API.

[0457] Example of text data obtained: "Development of our next product is progressing smoothly. It is currently in the prototype stage, and testing is scheduled to begin next month."

[0458] 3. The server passes this text data to OpenAI GPT-3 to generate a summary.

[0459] Example prompt: "Summarize the key points of this passage."

[0460] Generated summary: "Development of next product is progressing well, testing will begin next month."

[0461] 4. The server formats the summary into minutes format using the Jinja2 template engine.

[0462] Example format:

[0463] ★Agenda: Progress of next product development

[0464] Development is progressing smoothly

[0465] Testing is scheduled to begin next month

[0466] 5. The server converts the formatted minutes into PDF format and sends them to the user.

[0467] A notification will be sent to the user's device, and the user can download the PDF minutes and check the contents.

[0468] This makes it possible to automatically generate minutes from meeting audio data efficiently and with high accuracy, and distribute them in a unified format.

[0469] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0470] Step 1:

[0471] After the conference ends, the user uploads the conference audio data to the server using a dedicated application. The user selects the audio data file (e.g., "meeting_record.wav") on the application screen and clicks the "Upload" button.

[0472] Input: Specify the audio data and give instructions to upload it on the user's device.

[0473] Output: The specified audio data is sent to the server via a data communication network.

[0474] Step 2:

[0475] The server temporarily stores the received audio data. The server saves the received file in a specific directory and records the saved path and file name in the database.

[0476] Input: Audio data sent from the user device.

[0477] Output: The audio data is saved to file storage and the information is recorded in a database.

[0478] Step 3:

[0479] The server passes the saved voice data to a voice recognition API and converts it into text data. The server then sends a request to the voice recognition API (e.g., Google Speech-to-Text API) with the path to the audio file as an argument. The API analyzes the voice data and returns the corresponding text data.

[0480] Input: The file path of the saved audio data and the request to the speech recognition API.

[0481] Output: Text data generated from the audio data.

[0482] Step 4:

[0483] The server receives the text data returned by the speech recognition API and stores it in a database. The server inserts this text data into the appropriate tables and records it in a log file.

[0484] Input: Text data returned from the speech recognition API.

[0485] Output: Text data is saved to a database and logged.

[0486] Step 5:

[0487] The server passes the text data stored in the database to a natural language processing engine to generate a summary. The server extracts the text data and passes it to a generative AI model (e.g., OpenAI GPT-3) along with the prompt, "Please summarize the key points of this passage."

[0488] Input: Text data stored in the database and prompt statements.

[0489] Output: A summary returned by the generative AI model.

[0490] Step 6:

[0491] The server receives the generated summary and stores it in a database. The server inserts the summary into the database and records it in a log file.

[0492] Input: A summary returned by the generative AI model.

[0493] Output: The summary sentences are stored in the database and logged.

[0494] Step 7:

[0495] The server passes the abstract to a template engine, which formats it into minutes format. The server passes the abstract to a template engine (e.g., Jinja2) as an argument, which formats it into minutes format.

[0496] Input: Generated abstract and minutes format template.

[0497] Output: Formatted meeting minutes text.

[0498] Step 8:

[0499] The server converts the formatted minutes into PDF format and saves it in file storage. The server passes the text obtained from the template engine to the PDF generation library to generate a PDF file.

[0500] Input: Formatted meeting minutes text.

[0501] Output: Meeting minutes file in PDF format.

[0502] Step 9:

[0503] The server delivers the PDF minutes to the user's device. The server then sends an email with the PDF file attached, and notifies the user through a dedicated application. The user's device receives the notification and displays a download link to the user.

[0504] Input: PDF file of meeting minutes and user contact information.

[0505] Output: A notification will be sent to the user's device, and the minutes will be available for download.

[0506] (Application example 1)

[0507] 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."

[0508] There is a need for a method to efficiently collect and summarize conversations and evaluations during operation of autonomous vehicles, and quickly generate and distribute them as operational reports. Currently, manual recording of conversations during operation and subsequent manual summarization are required, which is extremely time-consuming and labor-intensive. In addition, it is difficult for operation managers to grasp the operational status of multiple vehicles in a timely manner. Therefore, a highly efficient and automated system for recording and summarizing operational conversations is needed.

[0509] 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.

[0510] In this invention, the server includes means for receiving voice data, means for converting the voice data into text data, means for summarizing the text data using a natural language processing engine and formatting it into a standard operation report format, and means for converting the formatted report into PDF format and sending it by email. This makes it possible to automatically collect voice data from inside an autonomous vehicle and quickly generate and distribute summarized operation reports.

[0511] "Audio data" refers to data that records conversations, sounds, etc. in digital format.

[0512] "Text data" is digital data that is obtained by converting voice data into character information.

[0513] A "summary" refers to the extraction of important information from text data and reconstructing it in a short form.

[0514] "Format" refers to the form and layout of a document or data, which is formatted according to specific standards or rules.

[0515] A "report" is a document that summarizes the results and progress of a specific event or activity.

[0516] "Mounted vehicle device" refers to a device installed in an autonomous vehicle that records and uploads voice data.

[0517] A "cloud server" is a remote server provided over the Internet for storing and processing data.

[0518] A "natural language processing engine" is a software engine for analyzing text data and performing high-level processing such as summarization and classification.

[0519] "PDF format" is an abbreviation for Portable Document Format, a document format that can be displayed and printed across platforms.

[0520] "Email" refers to a means of exchanging messages electronically over the Internet.

[0521] In one embodiment of the present invention, a system is provided for automatically recording and summarizing conversations and evaluations during the operation of an autonomous vehicle to generate and distribute operational reports, which uses a cloud server, an in-vehicle device, a user terminal, and a data communication network.

[0522] System Configuration

[0523] 1. Vehicle-mounted equipment:

[0524] The vehicle will be equipped with a device that records voice data inside the vehicle and automatically uploads the audio files to a cloud server, collecting conversations and evaluations that occur during operation in real time.

[0525] 2. Cloud Server:

[0526] When voice data is received, it is first temporarily stored, then passed to an AI voice recognition engine (such as Google Cloud Speech-to-Text) to convert it into text data.

[0527] The resulting text data is passed to a natural language processing engine (such as OpenAI GPT or Google Cloud Natural Language API) to generate a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. The technologies used in this process apply natural language processing algorithms such as topic modeling and key phrase extraction.

[0528] The server then formats the summarized text into an operational report format, using a template engine to fit the summary text into a predefined operational report format, generating a standardized report that can be easily viewed and shared by users.

[0529] Finally, the server converts the generated operation report into PDF format and sends it to the user via email. A notification is sent to the user's terminal, and the generated report file becomes available for download.

[0530] Specific examples

[0531] For example, if audio data is recorded while an autonomous vehicle is operating that says, "Today's operation went very smoothly. Thanks to the new sensor system, obstacle avoidance is faster and more accurate. Fuel efficiency has also improved, resulting in lower fuel costs than expected," the following actions will be taken.

[0532] 1. Vehicle-mounted device: Audio data recorded inside the vehicle is automatically uploaded to a cloud server.

[0533] 2. Cloud server: Receives the voice data and transcribes it using Google Cloud Speech-to-Text. For example, the resulting text might be something like, "Today's operation went very smoothly. Thanks to the new sensor system, obstacle avoidance is now faster and more accurate. Fuel efficiency has also improved, resulting in lower fuel costs than expected."

[0534] 3. Cloud server: The obtained text data is passed to a natural language processing engine such as OpenAI GPT, which generates a summary. The summary result will be something like, "The vehicle travels smoothly, and the sensor system allows for fast and accurate obstacle avoidance. Fuel efficiency has improved, reducing fuel costs."

[0535] 4. Cloud server: The summarized text is formatted into an operational report format. For example, it is formatted into the following format using a template engine.

[0536] ★Operation report:

[0537] The operation is smooth

[0538] Sensor system enables fast and accurate obstacle avoidance

[0539] Improved fuel efficiency and reduced fuel costs

[0540] 5. Cloud server: The completed operation report is converted to PDF format and sent to the user via email. A notification is sent to the user's device, and the user can download and view the report.

[0541] This system will enable automated collection and summarization of conversations and evaluations that occur in real time while autonomous vehicles are operating, and will enable efficient creation and distribution of operational reports.

[0542] Example prompts to input to a generative AI model:

[0543] "Today's operation went very smoothly. Thanks to the new sensor system, obstacle avoidance is faster and more accurate. Fuel efficiency has also improved, resulting in lower fuel costs than expected."

[0544] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0545] Step 1:

[0546] An in-vehicle device collects audio data recorded inside the vehicle, which is then automatically uploaded to a cloud server. The input is the audio inside the vehicle, and the output is the upload of the audio file to the cloud server.

[0547] Step 2:

[0548] The server receives voice data from the vehicle's onboard device. The received voice data is temporarily stored and passed to an AI voice recognition engine for conversion into text data. The input is an audio file, and the output is text data. Specifically, the voice data is converted into text format using a service such as Google Cloud Speech-to-Text.

[0549] Step 3:

[0550] The server passes the converted text data to a natural language processing engine to generate a summary. The input is text data, and the output is summarized text. Specifically, OpenAI GPT and Google Cloud Natural Language API are used to analyze the text data, extract important information, and summarize it.

[0551] Step 4:

[0552] The server formats the summarized text into an operational report format. A template engine is used to fit the summary text into a predefined operational report format. The input is the summarized text, and the output is a formatted operational report. Specifically, the server generates an operational report by inserting the text into a format defined by a tool such as Tex or ReportLab.

[0553] Step 5:

[0554] The server converts the formatted operation report into PDF format and sends it to the user via email. The input is the formatted operation report, and the output is the sent PDF file. Specifically, the PDF file is generated using ReportLab and sent via email using SendGrid or Amazon SES. The user receives a notification and can download and view the generated report file.

[0555] 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.

[0556] As an embodiment of the present invention, we provide a system for efficiently transcribing and summarizing audio data from a meeting, and creating and distributing minutes that incorporate sentiment analysis. This system uses a server, user terminals, and a data communication network. Furthermore, by incorporating an emotion recognition engine, the emotions of the meeting participants can also be reflected in the minutes.

[0557] First, after the conference ends, the user uploads the conference audio data to the server using a dedicated application. The dedicated application is installed on the user's device (PC, smartphone, tablet, etc.) and has the function to specify the audio file and send it to the server.

[0558] Server Operation

[0559] Server: When voice data is received, it is first temporarily stored. Next, the voice data is passed to an AI voice recognition engine to convert it into text data. The AI ​​voice recognition engine used here uses a commercial or open source API (e.g., a general-purpose voice recognition API) that provides highly accurate voice recognition.

[0560] The obtained text data is also passed to the emotion engine, which analyzes the emotions of the meeting participants. The emotion engine recognizes the emotion corresponding to each part of the text data and adds an emotion tag. The technology used in this process applies natural language processing algorithms such as voice emotion recognition and text emotion analysis.

[0561] The server then passes the emotion-tagged text data to a natural language processing engine to generate a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. During this process, the importance and summary content are adjusted based on the emotion data.

[0562] The server formats the summarized text into a minutes format. Using a template engine, the summary text is formatted into a predefined minutes format. Furthermore, sentiment tags and contextual information are added to the text, allowing for emotional changes to be detected.

[0563] Finally, the server sends the generated minutes to the user. This process includes sending an email and sending a notification via a dedicated application. A notification is displayed on the user's device, and the generated minutes file (e.g., PDF or Word format) can be downloaded.

[0564] Specific examples

[0565] For example, in the case of a project meeting in which audio data "regarding the progress of development of the next product" is recorded, the following operations are performed.

[0566] 1. User: Uses a dedicated application to upload the meeting recording data to the server.

[0567] 2. Server: Receives the voice data and transcribes it using a commercial speech recognition API. For example, the server can obtain text data such as, "Development of our next product is progressing smoothly. We are currently in the prototype stage, and testing is scheduled to begin next month."

[0568] 3. Server: The emotion engine is used to analyze emotions from the text data and add tags. For example, the server can obtain emotion-tagged text such as, "Development of the next product is progressing smoothly [Enjoyed]. It is currently in the prototype stage [Neutral], and testing is scheduled to begin next month [Expected]."

[0569] 4. Server: The emotion-tagged text data is passed to a natural language processing engine, which generates a summary. The summary result is something like, "Development of the next product is progressing smoothly, and testing is scheduled to begin next month."

[0570] 5. Server: Format the summarized text into a minutes format. For example, it can be formatted into the following format using a template engine.

[0571] ★Agenda: Progress of next product development

[0572] Development is going smoothly [Joy]

[0573] Testing is expected to begin next month [expected]

[0574] 6. Server: Converts the completed minutes into PDF format and sends them to the user's device. The user receives a notification and can download and view the minutes.

[0575] In this way, the present invention not only improves the efficiency and accuracy of creating meeting minutes, but also provides richer information by reflecting the emotions of the participants.

[0576] The processing flow will be explained below.

[0577] Step 1:

[0578] User: After the conference ends, the user launches the dedicated application, selects the conference audio data file, and uploads it to the server. When the user clicks the "Upload" button, the audio data is sent to the server as an HTTP request.

[0579] Step 2:

[0580] Server: The server receives the HTTP request and temporarily stores the audio data in storage. Meta information such as the storage location and file name is also recorded in a database.

[0581] Step 3:

[0582] Server: The server passes the saved voice data to the AI ​​voice recognition engine. At this time, it generates an API request and calls the voice recognition API by specifying the file path of the voice data.

[0583] Step 4:

[0584] Server: Receives the transcription results returned from the speech recognition API and saves them as text data. For example, the resulting text data might be something like, "Let's start the meeting. We'll discuss the next project."

[0585] Step 5:

[0586] Server: The transcribed text data is passed to the emotion engine for sentiment analysis. The emotion engine recognizes the sentiment associated with each part of the text data and adds emotion tags. For example, emotion-tagged text data such as "Let's start the meeting [Neutral]. We will discuss the next project [Expected]" is obtained.

[0587] Step 6:

[0588] Server: The emotion-tagged text data is passed to a natural language processing engine to generate a summary. The importance of the summary is adjusted based on the emotion data. For example, a summary such as "Project discussion [expectations]" is generated.

[0589] Step 7:

[0590] Server: The summarized text is formatted into minutes format using a template engine. Sentiment tags are also embedded appropriately.

[0591] Step 8:

[0592] Server: Convert the formatted minutes into PDF or Word format. During the conversion, the emotional information is displayed visually in an easy-to-understand manner.

[0593] Step 9:

[0594] Server: Sends the generated minutes file to the user's device. A download link is provided to the user via email or a notification function in a dedicated application.

[0595] Step 10:

[0596] User: The user receives a notification on their device and can download and review the generated minutes via the app or email. If necessary, they can make corrections or add additional comments.

[0597] These are the specific processing steps of the system that combines the emotion engine. This not only improves the efficiency of creating meeting minutes, but also provides rich information that reflects the emotions of the participants.

[0598] Example 2

[0599] 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."

[0600] There is a need to efficiently transcribe and summarize meeting audio data and generate minutes that reflect the emotions of meeting participants. However, there are currently no meeting minutes creation systems that include a means to add emotion tags. As a result, it is difficult to provide rich information that incorporates the emotions of participants rather than simply transcribing the content of meetings.

[0601] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0602] In this invention, the server includes means for receiving conference audio data, means for converting the audio data into text data, means for attaching emotion tags to the text data, means for summarizing the emotion-tagged text data, means for formatting the summarized text into minutes format, and means for transmitting the formatted minutes. This makes it possible to accurately and efficiently record the contents of the conference as minutes and provide rich information that also reflects the emotions of the participants.

[0603] "Conference audio data" means electronic data containing speech and other utterances recorded during a conference.

[0604] "Means for receiving" refers to the function or method for transferring voice data from the user terminal to the server.

[0605] The "means for converting voice data into text data" refers to a method or device that includes voice recognition technology for analyzing voice and converting it into text information.

[0606] "Means for assigning emotion tags to text data" refers to a technology or method that analyzes the emotion of speech in recognized text information and adds appropriate emotion labels.

[0607] A "means for summarizing emotion-tagged text data" is a technology or algorithm that analyzes emotion-tagged text information, extracts important content, and summarizes it concisely.

[0608] "Means for formatting the summarized text into minutes format" refers to a template engine or other formatting method for organizing the summarized content into the prescribed minutes format.

[0609] "Means for sending formatted minutes" includes communication means and processes for delivering the generated minutes to the user via email or a dedicated application.

[0610] A "server" is a computer system that receives, processes, stores, analyzes, and distributes audio data.

[0611] A "user terminal" is a device such as a personal computer, smartphone, or tablet that is used to record and upload audio data to a server.

[0612] A "natural language processing engine" is software that includes artificial intelligence technology that analyzes text data and performs advanced processing such as generating summaries and analyzing sentiment.

[0613] This invention provides a system that efficiently transcribes and summarizes audio data from a conference, and generates and distributes minutes that reflect the emotions of the conference participants. This system uses a server, user terminals, and a data communication network.

[0614] The system operates using the following hardware and software: User terminals include devices such as personal computers, smartphones, and tablets. The server uses commercial or open-source AI speech recognition engines (e.g., general-purpose speech recognition APIs), emotion recognition engines (e.g., emotion analysis APIs using natural language processing technology), natural language processing engines (e.g., generative AI models), and template engines (e.g., Jinja2).

[0615] After the conference ends, users use a dedicated application to upload the conference audio data to the server. The dedicated application is installed on the user's device (PC, smartphone, tablet, etc.) and has the function of selecting an audio file and sending it to a specified server.

[0616] When the server receives the voice data, it first temporarily stores it. Next, it passes the voice data to an AI voice recognition engine to convert it into text data. The resulting text data is also passed to an emotion engine, which analyzes the emotions of the meeting participants. The emotion engine recognizes the emotion corresponding to each part of the text data and adds an emotion tag. The technology used in this process applies natural language processing algorithms such as voice emotion recognition and text emotion analysis.

[0617] The server then inputs the emotion-tagged text data into a natural language processing engine to generate a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. During this process, the importance and summary content are adjusted based on the emotion data. The server then formats the summarized text into a minutes format. Using a template engine, the summary text is fitted into a predefined minutes format. Emotion tags and contextual information are also added, and the text is formatted to show changes in emotion.

[0618] Finally, the server sends the generated minutes to the user. This process includes sending an email and sending a notification via a dedicated application. A notification is displayed on the user's device, and the generated minutes file (e.g., PDF or Word format) can be downloaded.

[0619] Specific examples

[0620] For example, in a project meeting in which audio data on the "progress of development of the next product" is recorded, the following process takes place: The user uses a dedicated application to upload the recording of the meeting (file name: meeting_audio.wav) to the server. The server receives the audio data and transcribes it using a commercial speech recognition API. For example, text data such as "Development of the next product is progressing smoothly. It is currently in the prototype stage, and testing is scheduled to begin next month" is obtained. The server then uses an emotion engine to analyze the emotions from the text data and add tags. For example, emotion-tagged text such as "Development of the next product is progressing smoothly [Joy]. It is currently in the prototype stage [Neutral], and testing is scheduled to begin next month [Expectation]" is obtained. The server then passes the emotion-tagged text data to a natural language processing engine to generate a summary. The summary result is "Development of the next product is progressing smoothly, and testing is scheduled to begin next month." The server then formats the summarized text into meeting minutes format. For example, a template engine can be used to format it into the following format:

[0621] ★Agenda: Progress of next product development

[0622] Development is going smoothly [Joy]

[0623] Testing is expected to begin next month [expected]

[0624] Finally, the server converts the completed minutes into PDF format and sends them to the user's device. The user receives a notification and can download and view the minutes.

[0625] Prompt Sentence Examples

[0626] "Please generate minutes that include the emotions of the participants from audio data of a meeting regarding the development progress of the next product."

[0627] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0628] Step 1:

[0629] The user opens the dedicated application. The user selects the conference audio data file (e.g., meeting_audio.wav) and clicks the upload button. The device sends the selected audio data to the server. The input is the conference audio data, and the output is the audio data sent to the server.

[0630] Step 2:

[0631] The server receives the voice data sent from the device and temporarily stores it. The storage destination is data storage (e.g., storage service), the input is the sent voice data, and the output is the voice data stored in the data storage. Specifically, the server receives an HTTP request and writes binary data to storage.

[0632] Step 3:

[0633] The server passes the saved voice data to the AI ​​voice recognition engine. Specifically, this is the process of sending the voice data to a voice recognition API and converting it into text data. The input is the saved voice data, and the output is text data (e.g., "Development of the next product is going well..."). The server makes an API request and receives text data as a response.

[0634] Step 4:

[0635] The server inputs the acquired text data into the emotion engine for sentiment analysis. The emotion engine recognizes the emotion corresponding to each part of the text and assigns an emotion tag. The input is text data, and the output is emotion-tagged text data (e.g., "Development of the next product is going smoothly [joy]..."). Specifically, it calls the sentiment analysis API and reflects the results in the text.

[0636] Step 5:

[0637] The server passes the emotion-tagged text data to a natural language processing engine, which generates a summary of the text. The natural language processing engine extracts important information based on the emotion tags and creates a summary text. The input is emotion-tagged text data, and the output is summary text (e.g., "Development of the next product is progressing smoothly, and testing is scheduled to begin next month"). Specifically, the server generates a summary using a natural language processing model.

[0638] Step 6:

[0639] The server passes the summarized text to a template engine to format it into minutes of meeting format. A template engine (e.g., Jinja2) is used to fit the summary text into a predefined format. The input is the summary text, and the output is text formatted in minutes of meeting format (e.g., "Development progress of the next product; development is going well [joy]..."). Specifically, the text is input into the template engine, and formatted text is generated.

[0640] Step 7:

[0641] The server converts the formatted minutes into PDF format. A PDF generation program (e.g., ReportLab) is used to create a PDF from text data. The input is text formatted in minutes format, and the output is a PDF minutes file. Specifically, the server converts the text into PDF using a PDF generation library.

[0642] Step 8:

[0643] The server sends the generated PDF to the user. The server uses the email sending function or the notification function of a dedicated application. The input is a PDF format minutes file, and the output is a notification or email to the user's device. Specifically, the file is sent using the email sending API or push notification function.

[0644] Step 9:

[0645] The terminal receives the notification from the server and displays it to the user. The user checks the notification, clicks the provided link to download the minutes, and checks the contents. The input is the notification sent from the server, and the output is the minutes file downloaded to the user. Specifically, the terminal displays the notification message and allows the user to operate the link.

[0646] (Application example 2)

[0647] 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."

[0648] Previously, creating meeting minutes was often done manually, which was inefficient and time-consuming. It was also difficult to create minutes that reflected the emotions of the meeting participants, making it difficult to accurately capture the atmosphere and key points of the meeting. This made it difficult to share information in real time during operational meetings at factories and other locations, which sometimes hindered efficient decision-making.

[0649] 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 conference audio data, means for converting the audio data into text data, means for summarizing the text data, means for formatting the summarized text into a minutes format, means for performing sentiment analysis, means for adding sentiment tags, and means for transmitting the formatted minutes. This enables the automated and efficient creation of conference minutes, and further enables the creation of minutes that reflect the emotions of the conference participants. This enables rapid and accurate information sharing and promotes efficient decision-making.

[0650] "Meeting audio data" refers to audio files that record statements and discussions made during a meeting.

[0651] A "means for receiving" refers to a system or method for obtaining data from another device or network.

[0652] "Text data" refers to character string data converted from voice data by a voice recognition system.

[0653] A "summarization means" is a method or system for extracting important information from given text data and summarizing it concisely.

[0654] A "minutes format" is a predefined format for organizing and recording the contents of a meeting.

[0655] A "formatting means" is a method or system for converting and preparing data or information into a particular form or format.

[0656] A "transmitting means" is a system or method for sending data to another device or network.

[0657] "Sentiment analysis" is a technology that detects emotions from text data or audio data and evaluates the type and intensity of those emotions.

[0658] An "emotion tag" is a label or mark that indicates the emotion contained within a piece of text.

[0659] As an embodiment of this invention, we will explain an example of an "operational meeting support app for factory robots" installed on factory robots. The following explanation is based on the process of efficiently collecting meeting audio data, transcribing it, summarizing it, and analyzing its sentiment, and finally generating and distributing meeting minutes with sentiment tags.

[0660] System Configuration

[0661] The system uses factory robots, servers, and a data communication network. The factory robots are equipped with built-in microphones that collect audio data from meetings. The server processes the audio data and uses a highly accurate speech recognition API, emotion recognition engine, and natural language processing engine. The main software and hardware components are described below.

[0662] Hardware

[0663] Factory robots: Collect audio data using built-in microphones.

[0664] Server: A high-performance server used for data processing and storage.

[0665] software

[0666] Speech Recognition API: A commercial or open-source API for high-accuracy speech recognition (e.g., Google Speech-to-Text API).

[0667] Emotion recognition engine: Technology for analyzing emotions in text data and adding tags (e.g., Microsoft Azure Cognitive Services' Text Analytics).

[0668] Natural language processing engine: A technology for summarizing text data (e.g., OpenAI GPT-3).

[0669] Template engine: Software for formatting summary text into minutes format.

[0670] PDF generation software: Software to convert minutes into PDF format (e.g. pdfkit).

[0671] Processing flow

[0672] 1. Collection of audio data

[0673] The factory robots are equipped with built-in microphones to collect audio data from meetings in real time, and the audio files are temporarily stored inside the robots and then sent to a server via a network.

[0674] 2. Transcription

[0675] When the server receives the audio data, it converts it into text using the Google Cloud Speech-to-Text API, allowing the contents of the meeting to be recorded as text.

[0676] 3. Sentiment analysis

[0677] The resulting text data is passed to the Microsoft Azure Cognitive Services Text Analytics API for emotional tagging, which recognizes the emotions (happiness, sadness, anticipation, etc.) that correspond to each part of the text data and adds emotional tags.

[0678] 4. Summary Generation

[0679] The sentiment-tagged text data is then summarised using OpenAI GPT-3, which performs text analysis to extract and summarise key information concisely.

[0680] 5. Formatting of minutes

[0681] The generated summary and emotion tags are then formatted into a predefined minutes format using a template engine, allowing participants to understand the contents of the meeting, including changes in emotion, at a glance.

[0682] 6. Results Distribution

[0683] The final minutes are converted to PDF format using PDF generation software, then uploaded to email or cloud storage and notified to the relevant parties.

[0684] Specific examples

[0685] For example, if the following audio data is collected:

[0686] "Development of our next product is progressing smoothly and is currently in the prototype stage. Testing is scheduled to begin next month. All team members are highly motivated."

[0687] An example prompt based on this is:

[0688] Example prompt sentence:

[0689] Generate a summary for the following text: "Development of our next product is progressing smoothly and is currently in the prototype stage. Testing is scheduled to begin next month. All team members are highly motivated."

[0690] The final summary results are as follows:

[0691] "Development of the next product is progressing smoothly, and testing is scheduled to begin next month. The entire team is highly motivated."

[0692] As described above, the present invention automates and improves the efficiency of creating meeting minutes, and also provides information that reflects the emotions of the participants, thereby realizing richer information sharing.

[0693] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0694] Step 1:

[0695] The user collects audio data from meetings in real time using the factory robot's built-in microphone. The collected audio data is temporarily stored inside the robot. The audio data is then sent to a server via a network. The input is audio data, and the output is an audio file sent to the server.

[0696] Step 2:

[0697] The server receives the voice data and temporarily stores it. The received voice data is passed to a high-precision voice recognition API (e.g., Google Speech-to-Text API), which converts the voice data into text data. The input is voice data, and the output is transcribed text data. Specifically, the server sends a voice recognition request, and the API analyzes the voice data and converts it into text.

[0698] Step 3:

[0699] The server passes the obtained text data to an emotion recognition engine (for example, the Text Analytics API from Microsoft Azure Cognitive Services) to analyze the emotions in the text data. The emotion recognition engine detects the emotions corresponding to each part of the text and adds emotion tags. The input is text data, and the output is emotion-tagged text data. Specifically, the server sends an emotion analysis request, and the API analyzes the text and adds emotion tags.

[0700] Step 4:

[0701] The server passes emotion-tagged text data to a natural language processing engine (e.g., OpenAI GPT-3) to extract key information and generate a summary. The input is emotion-tagged text data, and the output is the summary text. Specifically, a summary generation request is sent, and the generative AI model analyzes the text and creates a summary.

[0702] Step 5:

[0703] The server passes the summarized text and emotion tags to a template engine to format it into a predefined minutes format. The input is the summary text and emotion-tagged text data, and the output is text formatted in the minutes format. Specifically, the template engine rearranges the text data into a predetermined format and performs additional processing to visually display the emotion tag information.

[0704] Step 6:

[0705] The server converts the formatted minutes into PDF format and sends it to the user's terminal. The input is text formatted in the minutes format, and the output is a PDF minutes file. Specifically, the server converts the text data into a PDF file using PDF generation software (e.g., pdfkit), and then sends the PDF file by email or uploads it to cloud storage.

[0706] These are the processing steps of the factory robot operation meeting support app, which efficiently transcribes the audio data of meetings, automatically generates and distributes meeting minutes that have been summarized and analyzed for sentiment.

[0707] 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.

[0708] 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.

[0709] 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.

[0710] [Third embodiment]

[0711] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0712] 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.

[0713] 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).

[0714] 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.

[0715] 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.

[0716] 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).

[0717] 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.

[0718] 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.

[0719] 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.

[0720] 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.

[0721] 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.

[0722] 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."

[0723] In accordance with an embodiment of the present invention, there is provided a system for efficiently transcribing and summarizing audio data from a conference to create and distribute minutes, which uses a server, user terminals, and a data communication network.

[0724] First, after the conference ends, the user uploads the conference audio data to the server using a dedicated application. The dedicated application is installed on the user's device (PC, smartphone, tablet, etc.) and has the function to specify the audio file and send it to the server.

[0725] Server Operation

[0726] Server: When voice data is received, it is first temporarily stored. Next, the voice data is passed to an AI voice recognition engine to convert it into text data. The AI ​​voice recognition engine used here uses a commercial or open source API (e.g., a general-purpose voice recognition API) that provides highly accurate voice recognition.

[0727] The resulting text data is passed to a natural language processing engine, which then generates a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. The technologies used in this process apply natural language processing algorithms such as topic modeling and key phrase extraction.

[0728] The server then formats the summarized text into minutes, using a template engine to fit the summary text into a predefined minutes format, generating minutes in a standard format that users can easily review and share.

[0729] Finally, the server sends the generated minutes to the user. This process includes sending an email and sending a notification via a dedicated application. A notification is displayed on the user's device, and the generated minutes file (e.g., PDF or Word format) can be downloaded.

[0730] Specific examples

[0731] For example, in the case of a project meeting in which audio data "regarding the progress of development of the next product" is recorded, the following operations are performed.

[0732] 1. User: Uses a dedicated application to upload the meeting recording data to the server.

[0733] 2. Server: Receives the voice data and transcribes it using a commercial speech recognition API. For example, the server can obtain text data such as, "Development of our next product is progressing smoothly. We are currently in the prototype stage, and testing is scheduled to begin next month."

[0734] 3. Server: The obtained text data is passed to a natural language processing engine to generate a summary. The summary result is something like, "Development of the next product is progressing smoothly, and testing is scheduled to begin next month."

[0735] 4. Server: Format the summarized text into a minutes format. For example, it can be formatted into the following format using a template engine.

[0736] ★Agenda: Progress of next product development

[0737] Development is progressing smoothly

[0738] Testing is scheduled to begin next month

[0739] 5. Server: Converts the completed minutes into PDF format and sends them to the user's device. The user receives a notification and can download and view the minutes.

[0740] In this way, the present invention improves the efficiency and accuracy of creating meeting minutes.

[0741] The processing flow will be explained below.

[0742] Step 1:

[0743] User: After the conference ends, the user launches the dedicated application, selects the conference audio data file, and uploads it to the server. When the user clicks the "Upload" button, the audio data is sent to the server as an HTTP request.

[0744] Step 2:

[0745] Server: The server receives the HTTP request and temporarily stores the audio data in storage. Meta information such as the storage location and file name is also recorded in a database.

[0746] Step 3:

[0747] Server: The server passes the saved voice data to the AI ​​voice recognition engine. At this time, it generates an API request and calls the voice recognition API by specifying the file path of the voice data.

[0748] Step 4:

[0749] Server: Receives the transcription results returned from the speech recognition API and saves them as text data. For example, the resulting text data might be something like, "Let's start the meeting. We'll discuss the next project."

[0750] Step 5:

[0751] Server: Passes the transcribed text data to the natural language processing engine and makes an API request to generate a summary, again passing the text data as input.

[0752] Step 6:

[0753] Server: Receives the summary results returned by the natural language processing engine and saves them as text. For example, a summary such as "Meeting begins. Project discussion."

[0754] Step 7:

[0755] Server: The summary text is formatted into a minutes format using a template engine. The summary text is embedded in a pre-prepared template to generate a highly readable format.

[0756] Step 8:

[0757] Server: Converts the formatted minutes into PDF or Word format, creating a file format that users can easily view and share.

[0758] Step 9:

[0759] Server: Sends the generated minutes file to the user's device. A download link is provided to the user via email or a notification function in a dedicated application.

[0760] Step 10:

[0761] User: The user receives a notification on their device and can download and review the generated minutes via the app or email. If necessary, they can make corrections or add additional comments.

[0762] These are the specific steps for transcribing, summarizing, creating, and distributing meeting minutes based on audio data from a meeting. This process allows for efficient and accurate minutes to be created.

[0763] Example 1

[0764] 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."

[0765] Efficiently transcribing and summarizing meeting audio data to create and distribute meeting minutes requires time and effort, and there are issues with the accuracy of the results. It is also often difficult to standardize the format of the generated minutes.

[0766] 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.

[0767] In this invention, the server includes means for receiving conference audio data, means for converting the audio data into text data using highly accurate speech recognition technology, means for summarizing the text data using natural language processing technology, means for formatting the summarized text into a minutes format using a template engine, and means for delivering the formatted minutes to user terminals. This makes it possible to automatically generate minutes from conference audio data efficiently and with high accuracy and deliver them in a unified format.

[0768] "Conference audio data" refers to data in which audio information generated during a conference is recorded in digital format.

[0769] "Speech recognition technology" is a technology for converting speech information into text data, and is a technology that extracts text information from speech using highly accurate algorithms.

[0770] "Natural language processing technology" is a technology for analyzing text data to extract summaries and important information, and uses generative AI models and other algorithms.

[0771] A "template engine" is a software technology for fitting data into a standardized format, and is a technology for formatting any generated text into a pre-specified format.

[0772] A "user terminal" is an electronic device used by a user, such as a computer or smartphone, that can communicate with a server via a network.

[0773] The present invention relates to a system for efficiently transcribing and summarizing audio data from a meeting to create and distribute minutes, which is implemented using a server, user terminals, and a data communication network.

[0774] Hardware and software used

[0775] This system uses the following hardware and software:

[0776] 1. User terminal: Refers to an electronic device that can communicate with the server, such as a computer, smartphone, or tablet.

[0777] 2. Server: A server that contains the hardware required to run speech recognition and natural language processing technologies.

[0778] 3. Data communication network: Internet communication environment for uploading audio data and distributing minutes.

[0779] Specific software used includes:

[0780] 1. Speech Recognition API: Use Google Speech-to-Text API or other speech recognition APIs.

[0781] 2. Natural Language Processing Engine: Uses OpenAI GPT-3 and other natural language processing technologies.

[0782] 3. Template engine: Use a template engine such as Jinja2.

[0783] System action

[0784] The processing of this system is carried out as follows.

[0785] 1. After the conference ends, the user uploads the conference audio data to the server using a dedicated application.

[0786] Using the dedicated software installed on the user's terminal, the user specifies the audio file and clicks the "Upload" button.

[0787] 2. The server receives the audio data and temporarily stores it.

[0788] The server receives the audio files over the network and stores them securely in storage.

[0789] 3. The server uses a speech recognition API to convert the voice data into text data.

[0790] For example, pass an audio file to the Google Speech-to-Text API to generate text data.

[0791] 4. The server passes the text data to a natural language processing engine to generate a summary.

[0792] A generative AI model such as OpenAI GPT-3 is given the prompt "Summarize the key points of this text" to generate a summary.

[0793] 5. The server formats the summarized text into minutes format using a template engine.

[0794] The summary is formatted into a standard format for minutes using the Jinja2 template engine.

[0795] 6. The server distributes the generated minutes to the user terminal.

[0796] The generated minutes (e.g., PDF format) are sent to the user using the email sending function or notification function via a dedicated application.

[0797] Specific examples

[0798] For example, if audio data "regarding the development progress of the next product" is recorded at a project meeting, the process proceeds as follows.

[0799] 1. The user uploads an audio data file (e.g., progress_meeting.wav) using a dedicated application.

[0800] 2. The server receives this audio data and performs character recognition using the Google Speech-to-Text API.

[0801] Example of text data obtained: "Development of our next product is progressing smoothly. It is currently in the prototype stage, and testing is scheduled to begin next month."

[0802] 3. The server passes this text data to OpenAI GPT-3 to generate a summary.

[0803] Example prompt: "Summarize the key points of this passage."

[0804] Generated summary: "Development of next product is progressing well, testing will begin next month."

[0805] 4. The server formats the summary into minutes format using the Jinja2 template engine.

[0806] Example format:

[0807] ★Agenda: Progress of next product development

[0808] Development is progressing smoothly

[0809] Testing is scheduled to begin next month

[0810] 5. The server converts the formatted minutes into PDF format and sends them to the user.

[0811] A notification will be sent to the user's device, and the user can download the PDF minutes and check the contents.

[0812] This makes it possible to automatically generate minutes from meeting audio data efficiently and with high accuracy, and distribute them in a unified format.

[0813] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0814] Step 1:

[0815] After the conference ends, the user uploads the conference audio data to the server using a dedicated application. The user selects the audio data file (e.g., "meeting_record.wav") on the application screen and clicks the "Upload" button.

[0816] Input: Specify the audio data and give instructions to upload it on the user's device.

[0817] Output: The specified audio data is sent to the server via a data communication network.

[0818] Step 2:

[0819] The server temporarily stores the received audio data. The server saves the received file in a specific directory and records the saved path and file name in the database.

[0820] Input: Audio data sent from the user device.

[0821] Output: The audio data is saved to file storage and the information is recorded in a database.

[0822] Step 3:

[0823] The server passes the saved voice data to a voice recognition API and converts it into text data. The server then sends a request to the voice recognition API (e.g., Google Speech-to-Text API) with the path to the audio file as an argument. The API analyzes the voice data and returns the corresponding text data.

[0824] Input: The file path of the saved audio data and the request to the speech recognition API.

[0825] Output: Text data generated from the audio data.

[0826] Step 4:

[0827] The server receives the text data returned by the speech recognition API and stores it in a database. The server inserts this text data into the appropriate tables and records it in a log file.

[0828] Input: Text data returned from the speech recognition API.

[0829] Output: Text data is saved to a database and logged.

[0830] Step 5:

[0831] The server passes the text data stored in the database to a natural language processing engine to generate a summary. The server extracts the text data and passes it to a generative AI model (e.g., OpenAI GPT-3) along with the prompt, "Please summarize the key points of this passage."

[0832] Input: Text data stored in the database and prompt statements.

[0833] Output: A summary returned by the generative AI model.

[0834] Step 6:

[0835] The server receives the generated summary and stores it in a database. The server inserts the summary into the database and records it in a log file.

[0836] Input: A summary returned by the generative AI model.

[0837] Output: The summary sentences are stored in the database and logged.

[0838] Step 7:

[0839] The server passes the abstract to a template engine, which formats it into minutes format. The server passes the abstract to a template engine (e.g., Jinja2) as an argument, which formats it into minutes format.

[0840] Input: Generated abstract and minutes format template.

[0841] Output: Formatted meeting minutes text.

[0842] Step 8:

[0843] The server converts the formatted minutes into PDF format and saves it in file storage. The server passes the text obtained from the template engine to the PDF generation library to generate a PDF file.

[0844] Input: Formatted meeting minutes text.

[0845] Output: Meeting minutes file in PDF format.

[0846] Step 9:

[0847] The server delivers the PDF minutes to the user's device. The server then sends an email with the PDF file attached, and notifies the user through a dedicated application. The user's device receives the notification and displays a download link to the user.

[0848] Input: PDF file of meeting minutes and user contact information.

[0849] Output: A notification will be sent to the user's device, and the minutes will be available for download.

[0850] (Application example 1)

[0851] 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."

[0852] There is a need for a method to efficiently collect and summarize conversations and evaluations during operation of autonomous vehicles, and quickly generate and distribute them as operational reports. Currently, manual recording of conversations during operation and subsequent manual summarization are required, which is extremely time-consuming and labor-intensive. In addition, it is difficult for operation managers to grasp the operational status of multiple vehicles in a timely manner. Therefore, a highly efficient and automated system for recording and summarizing operational conversations is needed.

[0853] 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.

[0854] In this invention, the server includes means for receiving voice data, means for converting the voice data into text data, means for summarizing the text data using a natural language processing engine and formatting it into a standard operation report format, and means for converting the formatted report into PDF format and sending it by email. This makes it possible to automatically collect voice data from inside an autonomous vehicle and quickly generate and distribute summarized operation reports.

[0855] "Audio data" refers to data that records conversations, sounds, etc. in digital format.

[0856] "Text data" is digital data that is obtained by converting voice data into character information.

[0857] A "summary" refers to the extraction of important information from text data and reconstructing it in a short form.

[0858] "Format" refers to the form and layout of a document or data, which is formatted according to specific standards or rules.

[0859] A "report" is a document that summarizes the results and progress of a specific event or activity.

[0860] "Mounted vehicle device" refers to a device installed in an autonomous vehicle that records and uploads voice data.

[0861] A "cloud server" is a remote server provided over the Internet for storing and processing data.

[0862] A "natural language processing engine" is a software engine for analyzing text data and performing high-level processing such as summarization and classification.

[0863] "PDF format" is an abbreviation for Portable Document Format, a document format that can be displayed and printed across platforms.

[0864] "Email" refers to a means of exchanging messages electronically over the Internet.

[0865] In one embodiment of the present invention, a system is provided for automatically recording and summarizing conversations and evaluations during the operation of an autonomous vehicle to generate and distribute operational reports, which uses a cloud server, an in-vehicle device, a user terminal, and a data communication network.

[0866] System Configuration

[0867] 1. Vehicle-mounted equipment:

[0868] The vehicle will be equipped with a device that records voice data inside the vehicle and automatically uploads the audio files to a cloud server, collecting conversations and evaluations that occur during operation in real time.

[0869] 2. Cloud Server:

[0870] When voice data is received, it is first temporarily stored, then passed to an AI voice recognition engine (such as Google Cloud Speech-to-Text) to convert it into text data.

[0871] The resulting text data is passed to a natural language processing engine (such as OpenAI GPT or Google Cloud Natural Language API) to generate a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. The technologies used in this process apply natural language processing algorithms such as topic modeling and key phrase extraction.

[0872] The server then formats the summarized text into an operational report format, using a template engine to fit the summary text into a predefined operational report format, generating a standardized report that can be easily viewed and shared by users.

[0873] Finally, the server converts the generated operation report into PDF format and sends it to the user via email. A notification is sent to the user's terminal, and the generated report file becomes available for download.

[0874] Specific examples

[0875] For example, if audio data is recorded while an autonomous vehicle is operating that says, "Today's operation went very smoothly. Thanks to the new sensor system, obstacle avoidance is faster and more accurate. Fuel efficiency has also improved, resulting in lower fuel costs than expected," the following actions will be taken.

[0876] 1. Vehicle-mounted device: Audio data recorded inside the vehicle is automatically uploaded to a cloud server.

[0877] 2. Cloud server: Receives the voice data and transcribes it using Google Cloud Speech-to-Text. For example, the resulting text might be something like, "Today's operation went very smoothly. Thanks to the new sensor system, obstacle avoidance is now faster and more accurate. Fuel efficiency has also improved, resulting in lower fuel costs than expected."

[0878] 3. Cloud server: The obtained text data is passed to a natural language processing engine such as OpenAI GPT, which generates a summary. The summary result will be something like, "The vehicle travels smoothly, and the sensor system allows for fast and accurate obstacle avoidance. Fuel efficiency has improved, reducing fuel costs."

[0879] 4. Cloud server: The summarized text is formatted into an operational report format. For example, it is formatted into the following format using a template engine.

[0880] ★Operation report:

[0881] The operation is smooth

[0882] Sensor system enables fast and accurate obstacle avoidance

[0883] Improved fuel efficiency and reduced fuel costs

[0884] 5. Cloud server: The completed operation report is converted to PDF format and sent to the user via email. A notification is sent to the user's device, and the user can download and view the report.

[0885] This system will enable automated collection and summarization of conversations and evaluations that occur in real time while autonomous vehicles are operating, and will enable efficient creation and distribution of operational reports.

[0886] Example prompts to input to a generative AI model:

[0887] "Today's operation went very smoothly. Thanks to the new sensor system, obstacle avoidance is faster and more accurate. Fuel efficiency has also improved, resulting in lower fuel costs than expected."

[0888] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0889] Step 1:

[0890] An in-vehicle device collects audio data recorded inside the vehicle, which is then automatically uploaded to a cloud server. The input is the audio inside the vehicle, and the output is the upload of the audio file to the cloud server.

[0891] Step 2:

[0892] The server receives voice data from the vehicle's onboard device. The received voice data is temporarily stored and passed to an AI voice recognition engine for conversion into text data. The input is an audio file, and the output is text data. Specifically, the voice data is converted into text format using a service such as Google Cloud Speech-to-Text.

[0893] Step 3:

[0894] The server passes the converted text data to a natural language processing engine to generate a summary. The input is text data, and the output is summarized text. Specifically, OpenAI GPT and Google Cloud Natural Language API are used to analyze the text data, extract important information, and summarize it.

[0895] Step 4:

[0896] The server formats the summarized text into an operational report format. A template engine is used to fit the summary text into a predefined operational report format. The input is the summarized text, and the output is a formatted operational report. Specifically, the server generates an operational report by inserting the text into a format defined by a tool such as Tex or ReportLab.

[0897] Step 5:

[0898] The server converts the formatted operation report into PDF format and sends it to the user via email. The input is the formatted operation report, and the output is the sent PDF file. Specifically, the PDF file is generated using ReportLab and sent via email using SendGrid or Amazon SES. The user receives a notification and can download and view the generated report file.

[0899] 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.

[0900] As an embodiment of the present invention, we provide a system for efficiently transcribing and summarizing audio data from a meeting, and creating and distributing minutes that incorporate sentiment analysis. This system uses a server, user terminals, and a data communication network. Furthermore, by incorporating an emotion recognition engine, the emotions of the meeting participants can also be reflected in the minutes.

[0901] First, after the conference ends, the user uploads the conference audio data to the server using a dedicated application. The dedicated application is installed on the user's device (PC, smartphone, tablet, etc.) and has the function to specify the audio file and send it to the server.

[0902] Server Operation

[0903] Server: When voice data is received, it is first temporarily stored. Next, the voice data is passed to an AI voice recognition engine to convert it into text data. The AI ​​voice recognition engine used here uses a commercial or open source API (e.g., a general-purpose voice recognition API) that provides highly accurate voice recognition.

[0904] The obtained text data is also passed to the emotion engine, which analyzes the emotions of the meeting participants. The emotion engine recognizes the emotion corresponding to each part of the text data and adds an emotion tag. The technology used in this process applies natural language processing algorithms such as voice emotion recognition and text emotion analysis.

[0905] The server then passes the emotion-tagged text data to a natural language processing engine to generate a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. During this process, the importance and summary content are adjusted based on the emotion data.

[0906] The server formats the summarized text into a minutes format. Using a template engine, the summary text is formatted into a predefined minutes format. Furthermore, sentiment tags and contextual information are added to the text, allowing for emotional changes to be detected.

[0907] Finally, the server sends the generated minutes to the user. This process includes sending an email and sending a notification via a dedicated application. A notification is displayed on the user's device, and the generated minutes file (e.g., PDF or Word format) can be downloaded.

[0908] Specific examples

[0909] For example, in the case of a project meeting in which audio data "regarding the progress of development of the next product" is recorded, the following operations are performed.

[0910] 1. User: Uses a dedicated application to upload the meeting recording data to the server.

[0911] 2. Server: Receives the voice data and transcribes it using a commercial speech recognition API. For example, the server can obtain text data such as, "Development of our next product is progressing smoothly. We are currently in the prototype stage, and testing is scheduled to begin next month."

[0912] 3. Server: The emotion engine is used to analyze emotions from the text data and add tags. For example, the server can obtain emotion-tagged text such as, "Development of the next product is progressing smoothly [Enjoyed]. It is currently in the prototype stage [Neutral], and testing is scheduled to begin next month [Expected]."

[0913] 4. Server: The emotion-tagged text data is passed to a natural language processing engine, which generates a summary. The summary result is something like, "Development of the next product is progressing smoothly, and testing is scheduled to begin next month."

[0914] 5. Server: Format the summarized text into a minutes format. For example, it can be formatted into the following format using a template engine.

[0915] ★Agenda: Progress of next product development

[0916] Development is going smoothly [Joy]

[0917] Testing is expected to begin next month [expected]

[0918] 6. Server: Converts the completed minutes into PDF format and sends them to the user's device. The user receives a notification and can download and view the minutes.

[0919] In this way, the present invention not only improves the efficiency and accuracy of creating meeting minutes, but also provides richer information by reflecting the emotions of the participants.

[0920] The processing flow will be explained below.

[0921] Step 1:

[0922] User: After the conference ends, the user launches the dedicated application, selects the conference audio data file, and uploads it to the server. When the user clicks the "Upload" button, the audio data is sent to the server as an HTTP request.

[0923] Step 2:

[0924] Server: The server receives the HTTP request and temporarily stores the audio data in storage. Meta information such as the storage location and file name is also recorded in a database.

[0925] Step 3:

[0926] Server: The server passes the saved voice data to the AI ​​voice recognition engine. At this time, it generates an API request and calls the voice recognition API by specifying the file path of the voice data.

[0927] Step 4:

[0928] Server: Receives the transcription results returned from the speech recognition API and saves them as text data. For example, the resulting text data might be something like, "Let's start the meeting. We'll discuss the next project."

[0929] Step 5:

[0930] Server: The transcribed text data is passed to the emotion engine for sentiment analysis. The emotion engine recognizes the sentiment associated with each part of the text data and adds emotion tags. For example, emotion-tagged text data such as "Let's start the meeting [Neutral]. We will discuss the next project [Expected]" is obtained.

[0931] Step 6:

[0932] Server: The emotion-tagged text data is passed to a natural language processing engine to generate a summary. The importance of the summary is adjusted based on the emotion data. For example, a summary such as "Project discussion [expectations]" is generated.

[0933] Step 7:

[0934] Server: The summarized text is formatted into minutes format using a template engine. Sentiment tags are also embedded appropriately.

[0935] Step 8:

[0936] Server: Convert the formatted minutes into PDF or Word format. During the conversion, the emotional information is displayed visually in an easy-to-understand manner.

[0937] Step 9:

[0938] Server: Sends the generated minutes file to the user's device. A download link is provided to the user via email or a notification function in a dedicated application.

[0939] Step 10:

[0940] User: The user receives a notification on their device and can download and review the generated minutes via the app or email. If necessary, they can make corrections or add additional comments.

[0941] These are the specific processing steps of the system that combines the emotion engine. This not only improves the efficiency of creating meeting minutes, but also provides rich information that reflects the emotions of the participants.

[0942] Example 2

[0943] 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."

[0944] There is a need to efficiently transcribe and summarize meeting audio data and generate minutes that reflect the emotions of meeting participants. However, there are currently no meeting minutes creation systems that include a means to add emotion tags. As a result, it is difficult to provide rich information that incorporates the emotions of participants rather than simply transcribing the content of meetings.

[0945] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0946] In this invention, the server includes means for receiving conference audio data, means for converting the audio data into text data, means for attaching emotion tags to the text data, means for summarizing the emotion-tagged text data, means for formatting the summarized text into minutes format, and means for transmitting the formatted minutes. This makes it possible to accurately and efficiently record the contents of the conference as minutes and provide rich information that also reflects the emotions of the participants.

[0947] "Conference audio data" means electronic data containing speech and other utterances recorded during a conference.

[0948] "Means for receiving" refers to the function or method for transferring voice data from the user terminal to the server.

[0949] The "means for converting voice data into text data" refers to a method or device that includes voice recognition technology for analyzing voice and converting it into text information.

[0950] "Means for assigning emotion tags to text data" refers to a technology or method that analyzes the emotion of speech in recognized text information and adds appropriate emotion labels.

[0951] A "means for summarizing emotion-tagged text data" is a technology or algorithm that analyzes emotion-tagged text information, extracts important content, and summarizes it concisely.

[0952] "Means for formatting the summarized text into minutes format" refers to a template engine or other formatting method for organizing the summarized content into the prescribed minutes format.

[0953] "Means for sending formatted minutes" includes communication means and processes for delivering the generated minutes to the user via email or a dedicated application.

[0954] A "server" is a computer system that receives, processes, stores, analyzes, and distributes audio data.

[0955] A "user terminal" is a device such as a personal computer, smartphone, or tablet that is used to record and upload audio data to a server.

[0956] A "natural language processing engine" is software that includes artificial intelligence technology that analyzes text data and performs advanced processing such as generating summaries and analyzing sentiment.

[0957] This invention provides a system that efficiently transcribes and summarizes audio data from a conference, and generates and distributes minutes that reflect the emotions of the conference participants. This system uses a server, user terminals, and a data communication network.

[0958] The system operates using the following hardware and software: User terminals include devices such as personal computers, smartphones, and tablets. The server uses commercial or open-source AI speech recognition engines (e.g., general-purpose speech recognition APIs), emotion recognition engines (e.g., emotion analysis APIs using natural language processing technology), natural language processing engines (e.g., generative AI models), and template engines (e.g., Jinja2).

[0959] After the conference ends, users use a dedicated application to upload the conference audio data to the server. The dedicated application is installed on the user's device (PC, smartphone, tablet, etc.) and has the function of selecting an audio file and sending it to a specified server.

[0960] When the server receives the voice data, it first temporarily stores it. Next, it passes the voice data to an AI voice recognition engine to convert it into text data. The resulting text data is also passed to an emotion engine, which analyzes the emotions of the meeting participants. The emotion engine recognizes the emotion corresponding to each part of the text data and adds an emotion tag. The technology used in this process applies natural language processing algorithms such as voice emotion recognition and text emotion analysis.

[0961] The server then inputs the emotion-tagged text data into a natural language processing engine to generate a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. During this process, the importance and summary content are adjusted based on the emotion data. The server then formats the summarized text into a minutes format. Using a template engine, the summary text is fitted into a predefined minutes format. Emotion tags and contextual information are also added, and the text is formatted to show changes in emotion.

[0962] Finally, the server sends the generated minutes to the user. This process includes sending an email and sending a notification via a dedicated application. A notification is displayed on the user's device, and the generated minutes file (e.g., PDF or Word format) can be downloaded.

[0963] Specific examples

[0964] For example, in a project meeting in which audio data on the "progress of development of the next product" is recorded, the following process takes place: The user uses a dedicated application to upload the recording of the meeting (file name: meeting_audio.wav) to the server. The server receives the audio data and transcribes it using a commercial speech recognition API. For example, text data such as "Development of the next product is progressing smoothly. It is currently in the prototype stage, and testing is scheduled to begin next month" is obtained. The server then uses an emotion engine to analyze the emotions from the text data and add tags. For example, emotion-tagged text such as "Development of the next product is progressing smoothly [Joy]. It is currently in the prototype stage [Neutral], and testing is scheduled to begin next month [Expectation]" is obtained. The server then passes the emotion-tagged text data to a natural language processing engine to generate a summary. The summary result is "Development of the next product is progressing smoothly, and testing is scheduled to begin next month." The server then formats the summarized text into meeting minutes format. For example, a template engine can be used to format it into the following format:

[0965] ★Agenda: Progress of next product development

[0966] Development is going smoothly [Joy]

[0967] Testing is expected to begin next month [expected]

[0968] Finally, the server converts the completed minutes into PDF format and sends them to the user's device. The user receives a notification and can download and view the minutes.

[0969] Prompt Sentence Examples

[0970] "Please generate minutes that include the emotions of the participants from audio data of a meeting regarding the development progress of the next product."

[0971] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0972] Step 1:

[0973] The user opens the dedicated application. The user selects the conference audio data file (e.g., meeting_audio.wav) and clicks the upload button. The device sends the selected audio data to the server. The input is the conference audio data, and the output is the audio data sent to the server.

[0974] Step 2:

[0975] The server receives the voice data sent from the device and temporarily stores it. The storage destination is data storage (e.g., storage service), the input is the sent voice data, and the output is the voice data stored in the data storage. Specifically, the server receives an HTTP request and writes binary data to storage.

[0976] Step 3:

[0977] The server passes the saved voice data to the AI ​​voice recognition engine. Specifically, this is the process of sending the voice data to a voice recognition API and converting it into text data. The input is the saved voice data, and the output is text data (e.g., "Development of the next product is going well..."). The server makes an API request and receives text data as a response.

[0978] Step 4:

[0979] The server inputs the acquired text data into the emotion engine for sentiment analysis. The emotion engine recognizes the emotion corresponding to each part of the text and assigns an emotion tag. The input is text data, and the output is emotion-tagged text data (e.g., "Development of the next product is going smoothly [joy]..."). Specifically, it calls the sentiment analysis API and reflects the results in the text.

[0980] Step 5:

[0981] The server passes the emotion-tagged text data to a natural language processing engine, which generates a summary of the text. The natural language processing engine extracts important information based on the emotion tags and creates a summary text. The input is emotion-tagged text data, and the output is summary text (e.g., "Development of the next product is progressing smoothly, and testing is scheduled to begin next month"). Specifically, the server generates a summary using a natural language processing model.

[0982] Step 6:

[0983] The server passes the summarized text to a template engine to format it into minutes of meeting format. A template engine (e.g., Jinja2) is used to fit the summary text into a predefined format. The input is the summary text, and the output is text formatted in minutes of meeting format (e.g., "Development progress of the next product; development is going well [joy]..."). Specifically, the text is input into the template engine, and formatted text is generated.

[0984] Step 7:

[0985] The server converts the formatted minutes into PDF format. A PDF generation program (e.g., ReportLab) is used to create a PDF from text data. The input is text formatted in minutes format, and the output is a PDF minutes file. Specifically, the server converts the text into PDF using a PDF generation library.

[0986] Step 8:

[0987] The server sends the generated PDF to the user. The server uses the email sending function or the notification function of a dedicated application. The input is a PDF format minutes file, and the output is a notification or email to the user's device. Specifically, the file is sent using the email sending API or push notification function.

[0988] Step 9:

[0989] The terminal receives the notification from the server and displays it to the user. The user checks the notification, clicks the provided link to download the minutes, and checks the contents. The input is the notification sent from the server, and the output is the minutes file downloaded to the user. Specifically, the terminal displays the notification message and allows the user to operate the link.

[0990] (Application example 2)

[0991] 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."

[0992] Previously, creating meeting minutes was often done manually, which was inefficient and time-consuming. It was also difficult to create minutes that reflected the emotions of the meeting participants, making it difficult to accurately capture the atmosphere and key points of the meeting. This made it difficult to share information in real time during operational meetings at factories and other locations, which sometimes hindered efficient decision-making.

[0993] 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 conference audio data, means for converting the audio data into text data, means for summarizing the text data, means for formatting the summarized text into a minutes format, means for performing sentiment analysis, means for adding sentiment tags, and means for transmitting the formatted minutes. This enables the automated and efficient creation of conference minutes, and further enables the creation of minutes that reflect the emotions of the conference participants. This enables rapid and accurate information sharing and promotes efficient decision-making.

[0994] "Meeting audio data" refers to audio files that record statements and discussions made during a meeting.

[0995] A "means for receiving" refers to a system or method for obtaining data from another device or network.

[0996] "Text data" refers to character string data converted from voice data by a voice recognition system.

[0997] A "summarization means" is a method or system for extracting important information from given text data and summarizing it concisely.

[0998] A "minutes format" is a predefined format for organizing and recording the contents of a meeting.

[0999] A "formatting means" is a method or system for converting and preparing data or information into a particular form or format.

[1000] A "transmitting means" is a system or method for sending data to another device or network.

[1001] "Sentiment analysis" is a technology that detects emotions from text data or audio data and evaluates the type and intensity of those emotions.

[1002] An "emotion tag" is a label or mark that indicates the emotion contained within a piece of text.

[1003] As an embodiment of this invention, we will explain an example of an "operational meeting support app for factory robots" installed on factory robots. The following explanation is based on the process of efficiently collecting meeting audio data, transcribing it, summarizing it, and analyzing its sentiment, and finally generating and distributing meeting minutes with sentiment tags.

[1004] System Configuration

[1005] The system uses factory robots, servers, and a data communication network. The factory robots are equipped with built-in microphones that collect audio data from meetings. The server processes the audio data and uses a highly accurate speech recognition API, emotion recognition engine, and natural language processing engine. The main software and hardware components are described below.

[1006] Hardware

[1007] Factory robots: Collect audio data using built-in microphones.

[1008] Server: A high-performance server used for data processing and storage.

[1009] software

[1010] Speech Recognition API: A commercial or open-source API for high-accuracy speech recognition (e.g., Google Speech-to-Text API).

[1011] Emotion recognition engine: Technology for analyzing emotions in text data and adding tags (e.g., Microsoft Azure Cognitive Services' Text Analytics).

[1012] Natural language processing engine: A technology for summarizing text data (e.g., OpenAI GPT-3).

[1013] Template engine: Software for formatting summary text into minutes format.

[1014] PDF generation software: Software to convert minutes into PDF format (e.g. pdfkit).

[1015] Processing flow

[1016] 1. Collection of audio data

[1017] The factory robots are equipped with built-in microphones to collect audio data from meetings in real time, and the audio files are temporarily stored inside the robots and then sent to a server via a network.

[1018] 2. Transcription

[1019] When the server receives the audio data, it converts it into text using the Google Cloud Speech-to-Text API, allowing the contents of the meeting to be recorded as text.

[1020] 3. Sentiment analysis

[1021] The resulting text data is passed to the Microsoft Azure Cognitive Services Text Analytics API for emotional tagging, which recognizes the emotions (happiness, sadness, anticipation, etc.) that correspond to each part of the text data and adds emotional tags.

[1022] 4. Summary Generation

[1023] The sentiment-tagged text data is then summarised using OpenAI GPT-3, which performs text analysis to extract and summarise key information concisely.

[1024] 5. Formatting of minutes

[1025] The generated summary and emotion tags are then formatted into a predefined minutes format using a template engine, allowing participants to understand the contents of the meeting, including changes in emotion, at a glance.

[1026] 6. Results Distribution

[1027] The final minutes are converted to PDF format using PDF generation software, then uploaded to email or cloud storage and notified to the relevant parties.

[1028] Specific examples

[1029] For example, if the following audio data is collected:

[1030] "Development of our next product is progressing smoothly and is currently in the prototype stage. Testing is scheduled to begin next month. All team members are highly motivated."

[1031] An example prompt based on this is:

[1032] Example prompt sentence:

[1033] Generate a summary for the following text: "Development of our next product is progressing smoothly and is currently in the prototype stage. Testing is scheduled to begin next month. All team members are highly motivated."

[1034] The final summary results are as follows:

[1035] "Development of the next product is progressing smoothly, and testing is scheduled to begin next month. The entire team is highly motivated."

[1036] As described above, the present invention automates and improves the efficiency of creating meeting minutes, and also provides information that reflects the emotions of the participants, thereby realizing richer information sharing.

[1037] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1038] Step 1:

[1039] The user collects audio data from meetings in real time using the factory robot's built-in microphone. The collected audio data is temporarily stored inside the robot. The audio data is then sent to a server via a network. The input is audio data, and the output is an audio file sent to the server.

[1040] Step 2:

[1041] The server receives the voice data and temporarily stores it. The received voice data is passed to a high-precision voice recognition API (e.g., Google Speech-to-Text API), which converts the voice data into text data. The input is voice data, and the output is transcribed text data. Specifically, the server sends a voice recognition request, and the API analyzes the voice data and converts it into text.

[1042] Step 3:

[1043] The server passes the obtained text data to an emotion recognition engine (for example, the Text Analytics API from Microsoft Azure Cognitive Services) to analyze the emotions in the text data. The emotion recognition engine detects the emotions corresponding to each part of the text and adds emotion tags. The input is text data, and the output is emotion-tagged text data. Specifically, the server sends an emotion analysis request, and the API analyzes the text and adds emotion tags.

[1044] Step 4:

[1045] The server passes emotion-tagged text data to a natural language processing engine (e.g., OpenAI GPT-3) to extract key information and generate a summary. The input is emotion-tagged text data, and the output is the summary text. Specifically, a summary generation request is sent, and the generative AI model analyzes the text and creates a summary.

[1046] Step 5:

[1047] The server passes the summarized text and emotion tags to a template engine to format it into a predefined minutes format. The input is the summary text and emotion-tagged text data, and the output is text formatted in the minutes format. Specifically, the template engine rearranges the text data into a predetermined format and performs additional processing to visually display the emotion tag information.

[1048] Step 6:

[1049] The server converts the formatted minutes into PDF format and sends it to the user's terminal. The input is text formatted in the minutes format, and the output is a PDF minutes file. Specifically, the server converts the text data into a PDF file using PDF generation software (e.g., pdfkit), and then sends the PDF file by email or uploads it to cloud storage.

[1050] These are the processing steps of the factory robot operation meeting support app, which efficiently transcribes the audio data of meetings, automatically generates and distributes meeting minutes that have been summarized and analyzed for sentiment.

[1051] 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.

[1052] 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.

[1053] 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.

[1054] [Fourth embodiment]

[1055] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1056] 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.

[1057] 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).

[1058] 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.

[1059] 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.

[1060] 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).

[1061] 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.

[1062] 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.

[1063] 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.

[1064] 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.

[1065] 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.

[1066] 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.

[1067] 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."

[1068] In accordance with an embodiment of the present invention, there is provided a system for efficiently transcribing and summarizing audio data from a conference to create and distribute minutes, which uses a server, user terminals, and a data communication network.

[1069] First, after the conference ends, the user uploads the conference audio data to the server using a dedicated application. The dedicated application is installed on the user's device (PC, smartphone, tablet, etc.) and has the function to specify the audio file and send it to the server.

[1070] Server Operation

[1071] Server: When voice data is received, it is first temporarily stored. Next, the voice data is passed to an AI voice recognition engine to convert it into text data. The AI ​​voice recognition engine used here uses a commercial or open source API (e.g., a general-purpose voice recognition API) that provides highly accurate voice recognition.

[1072] The resulting text data is passed to a natural language processing engine, which then generates a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. The technologies used in this process apply natural language processing algorithms such as topic modeling and key phrase extraction.

[1073] The server then formats the summarized text into minutes, using a template engine to fit the summary text into a predefined minutes format, generating minutes in a standard format that users can easily review and share.

[1074] Finally, the server sends the generated minutes to the user. This process includes sending an email and sending a notification via a dedicated application. A notification is displayed on the user's device, and the generated minutes file (e.g., PDF or Word format) can be downloaded.

[1075] Specific examples

[1076] For example, in the case of a project meeting in which audio data "regarding the progress of development of the next product" is recorded, the following operations are performed.

[1077] 1. User: Uses a dedicated application to upload the meeting recording data to the server.

[1078] 2. Server: Receives the voice data and transcribes it using a commercial speech recognition API. For example, the server can obtain text data such as, "Development of our next product is progressing smoothly. We are currently in the prototype stage, and testing is scheduled to begin next month."

[1079] 3. Server: The obtained text data is passed to a natural language processing engine to generate a summary. The summary result is something like, "Development of the next product is progressing smoothly, and testing is scheduled to begin next month."

[1080] 4. Server: Format the summarized text into a minutes format. For example, it can be formatted into the following format using a template engine.

[1081] ★Agenda: Progress of next product development

[1082] Development is progressing smoothly

[1083] Testing is scheduled to begin next month

[1084] 5. Server: Converts the completed minutes into PDF format and sends them to the user's device. The user receives a notification and can download and view the minutes.

[1085] In this way, the present invention improves the efficiency and accuracy of creating meeting minutes.

[1086] The processing flow will be explained below.

[1087] Step 1:

[1088] User: After the conference ends, the user launches the dedicated application, selects the conference audio data file, and uploads it to the server. When the user clicks the "Upload" button, the audio data is sent to the server as an HTTP request.

[1089] Step 2:

[1090] Server: The server receives the HTTP request and temporarily stores the audio data in storage. Meta information such as the storage location and file name is also recorded in a database.

[1091] Step 3:

[1092] Server: The server passes the saved voice data to the AI ​​voice recognition engine. At this time, it generates an API request and calls the voice recognition API by specifying the file path of the voice data.

[1093] Step 4:

[1094] Server: Receives the transcription results returned from the speech recognition API and saves them as text data. For example, the resulting text data might be something like, "Let's start the meeting. We'll discuss the next project."

[1095] Step 5:

[1096] Server: Passes the transcribed text data to the natural language processing engine and makes an API request to generate a summary, again passing the text data as input.

[1097] Step 6:

[1098] Server: Receives the summary results returned by the natural language processing engine and saves them as text. For example, a summary such as "Meeting begins. Project discussion."

[1099] Step 7:

[1100] Server: The summary text is formatted into a minutes format using a template engine. The summary text is embedded in a pre-prepared template to generate a highly readable format.

[1101] Step 8:

[1102] Server: Converts the formatted minutes into PDF or Word format, creating a file format that users can easily view and share.

[1103] Step 9:

[1104] Server: Sends the generated minutes file to the user's device. A download link is provided to the user via email or a notification function in a dedicated application.

[1105] Step 10:

[1106] User: The user receives a notification on their device and can download and review the generated minutes via the app or email. If necessary, they can make corrections or add additional comments.

[1107] These are the specific steps for transcribing, summarizing, creating, and distributing meeting minutes based on audio data from a meeting. This process allows for efficient and accurate minutes to be created.

[1108] Example 1

[1109] 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."

[1110] Efficiently transcribing and summarizing meeting audio data to create and distribute meeting minutes requires time and effort, and there are issues with the accuracy of the results. It is also often difficult to standardize the format of the generated minutes.

[1111] 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.

[1112] In this invention, the server includes means for receiving conference audio data, means for converting the audio data into text data using highly accurate speech recognition technology, means for summarizing the text data using natural language processing technology, means for formatting the summarized text into a minutes format using a template engine, and means for delivering the formatted minutes to user terminals. This makes it possible to automatically generate minutes from conference audio data efficiently and with high accuracy and deliver them in a unified format.

[1113] "Conference audio data" refers to data in which audio information generated during a conference is recorded in digital format.

[1114] "Speech recognition technology" is a technology for converting speech information into text data, and is a technology that extracts text information from speech using highly accurate algorithms.

[1115] "Natural language processing technology" is a technology for analyzing text data to extract summaries and important information, and uses generative AI models and other algorithms.

[1116] A "template engine" is a software technology for fitting data into a standardized format, and is a technology for formatting any generated text into a pre-specified format.

[1117] A "user terminal" is an electronic device used by a user, such as a computer or smartphone, that can communicate with a server via a network.

[1118] The present invention relates to a system for efficiently transcribing and summarizing audio data from a meeting to create and distribute minutes, which is implemented using a server, user terminals, and a data communication network.

[1119] Hardware and software used

[1120] This system uses the following hardware and software:

[1121] 1. User terminal: Refers to an electronic device that can communicate with the server, such as a computer, smartphone, or tablet.

[1122] 2. Server: A server that contains the hardware required to run speech recognition and natural language processing technologies.

[1123] 3. Data communication network: Internet communication environment for uploading audio data and distributing minutes.

[1124] Specific software used includes:

[1125] 1. Speech Recognition API: Use Google Speech-to-Text API or other speech recognition APIs.

[1126] 2. Natural Language Processing Engine: Uses OpenAI GPT-3 and other natural language processing technologies.

[1127] 3. Template engine: Use a template engine such as Jinja2.

[1128] System action

[1129] The processing of this system is carried out as follows.

[1130] 1. After the conference ends, the user uploads the conference audio data to the server using a dedicated application.

[1131] Using the dedicated software installed on the user's terminal, the user specifies the audio file and clicks the "Upload" button.

[1132] 2. The server receives the audio data and temporarily stores it.

[1133] The server receives the audio files over the network and stores them securely in storage.

[1134] 3. The server uses a speech recognition API to convert the voice data into text data.

[1135] For example, pass an audio file to the Google Speech-to-Text API to generate text data.

[1136] 4. The server passes the text data to a natural language processing engine to generate a summary.

[1137] A generative AI model such as OpenAI GPT-3 is given the prompt "Summarize the key points of this text" to generate a summary.

[1138] 5. The server formats the summarized text into minutes format using a template engine.

[1139] The summary is formatted into a standard format for minutes using the Jinja2 template engine.

[1140] 6. The server distributes the generated minutes to the user terminal.

[1141] The generated minutes (e.g., PDF format) are sent to the user using the email sending function or notification function via a dedicated application.

[1142] Specific examples

[1143] For example, if audio data "regarding the development progress of the next product" is recorded at a project meeting, the process proceeds as follows.

[1144] 1. The user uploads an audio data file (e.g., progress_meeting.wav) using a dedicated application.

[1145] 2. The server receives this audio data and performs character recognition using the Google Speech-to-Text API.

[1146] Example of text data obtained: "Development of our next product is progressing smoothly. It is currently in the prototype stage, and testing is scheduled to begin next month."

[1147] 3. The server passes this text data to OpenAI GPT-3 to generate a summary.

[1148] Example prompt: "Summarize the key points of this passage."

[1149] Generated summary: "Development of next product is progressing well, testing will begin next month."

[1150] 4. The server formats the summary into minutes format using the Jinja2 template engine.

[1151] Example format:

[1152] ★Agenda: Progress of next product development

[1153] Development is progressing smoothly

[1154] Testing is scheduled to begin next month

[1155] 5. The server converts the formatted minutes into PDF format and sends them to the user.

[1156] A notification will be sent to the user's device, and the user can download the PDF minutes and check the contents.

[1157] This makes it possible to automatically generate minutes from meeting audio data efficiently and with high accuracy, and distribute them in a unified format.

[1158] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1159] Step 1:

[1160] After the conference ends, the user uploads the conference audio data to the server using a dedicated application. The user selects the audio data file (e.g., "meeting_record.wav") on the application screen and clicks the "Upload" button.

[1161] Input: Specify the audio data and give instructions to upload it on the user's device.

[1162] Output: The specified audio data is sent to the server via a data communication network.

[1163] Step 2:

[1164] The server temporarily stores the received audio data. The server saves the received file in a specific directory and records the saved path and file name in the database.

[1165] Input: Audio data sent from the user device.

[1166] Output: The audio data is saved to file storage and the information is recorded in a database.

[1167] Step 3:

[1168] The server passes the saved voice data to a voice recognition API and converts it into text data. The server then sends a request to the voice recognition API (e.g., Google Speech-to-Text API) with the path to the audio file as an argument. The API analyzes the voice data and returns the corresponding text data.

[1169] Input: The file path of the saved audio data and the request to the speech recognition API.

[1170] Output: Text data generated from the audio data.

[1171] Step 4:

[1172] The server receives the text data returned by the speech recognition API and stores it in a database. The server inserts this text data into the appropriate tables and records it in a log file.

[1173] Input: Text data returned from the speech recognition API.

[1174] Output: Text data is saved to a database and logged.

[1175] Step 5:

[1176] The server passes the text data stored in the database to a natural language processing engine to generate a summary. The server extracts the text data and passes it to a generative AI model (e.g., OpenAI GPT-3) along with the prompt, "Please summarize the key points of this passage."

[1177] Input: Text data stored in the database and prompt statements.

[1178] Output: A summary returned by the generative AI model.

[1179] Step 6:

[1180] The server receives the generated summary and stores it in a database. The server inserts the summary into the database and records it in a log file.

[1181] Input: A summary returned by the generative AI model.

[1182] Output: The summary sentences are stored in the database and logged.

[1183] Step 7:

[1184] The server passes the abstract to a template engine, which formats it into minutes format. The server passes the abstract to a template engine (e.g., Jinja2) as an argument, which formats it into minutes format.

[1185] Input: Generated abstract and minutes format template.

[1186] Output: Formatted meeting minutes text.

[1187] Step 8:

[1188] The server converts the formatted minutes into PDF format and saves it in file storage. The server passes the text obtained from the template engine to the PDF generation library to generate a PDF file.

[1189] Input: Formatted meeting minutes text.

[1190] Output: Meeting minutes file in PDF format.

[1191] Step 9:

[1192] The server delivers the PDF minutes to the user's device. The server then sends an email with the PDF file attached, and notifies the user through a dedicated application. The user's device receives the notification and displays a download link to the user.

[1193] Input: PDF file of meeting minutes and user contact information.

[1194] Output: A notification will be sent to the user's device, and the minutes will be available for download.

[1195] (Application example 1)

[1196] 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."

[1197] There is a need for a method to efficiently collect and summarize conversations and evaluations during operation of autonomous vehicles, and quickly generate and distribute them as operational reports. Currently, manual recording of conversations during operation and subsequent manual summarization are required, which is extremely time-consuming and labor-intensive. In addition, it is difficult for operation managers to grasp the operational status of multiple vehicles in a timely manner. Therefore, a highly efficient and automated system for recording and summarizing operational conversations is needed.

[1198] 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.

[1199] In this invention, the server includes means for receiving voice data, means for converting the voice data into text data, means for summarizing the text data using a natural language processing engine and formatting it into a standard operation report format, and means for converting the formatted report into PDF format and sending it by email. This makes it possible to automatically collect voice data from inside an autonomous vehicle and quickly generate and distribute summarized operation reports.

[1200] "Audio data" refers to data that records conversations, sounds, etc. in digital format.

[1201] "Text data" is digital data that is obtained by converting voice data into character information.

[1202] A "summary" refers to the extraction of important information from text data and reconstructing it in a short form.

[1203] "Format" refers to the form and layout of a document or data, which is formatted according to specific standards or rules.

[1204] A "report" is a document that summarizes the results and progress of a specific event or activity.

[1205] "Mounted vehicle device" refers to a device installed in an autonomous vehicle that records and uploads voice data.

[1206] A "cloud server" is a remote server provided over the Internet for storing and processing data.

[1207] A "natural language processing engine" is a software engine for analyzing text data and performing high-level processing such as summarization and classification.

[1208] "PDF format" is an abbreviation for Portable Document Format, a document format that can be displayed and printed across platforms.

[1209] "Email" refers to a means of exchanging messages electronically over the Internet.

[1210] In one embodiment of the present invention, a system is provided for automatically recording and summarizing conversations and evaluations during the operation of an autonomous vehicle to generate and distribute operational reports, which uses a cloud server, an in-vehicle device, a user terminal, and a data communication network.

[1211] System Configuration

[1212] 1. Vehicle-mounted equipment:

[1213] The vehicle will be equipped with a device that records voice data inside the vehicle and automatically uploads the audio files to a cloud server, collecting conversations and evaluations that occur during operation in real time.

[1214] 2. Cloud Server:

[1215] When voice data is received, it is first temporarily stored, then passed to an AI voice recognition engine (such as Google Cloud Speech-to-Text) to convert it into text data.

[1216] The resulting text data is passed to a natural language processing engine (such as OpenAI GPT or Google Cloud Natural Language API) to generate a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. The technologies used in this process apply natural language processing algorithms such as topic modeling and key phrase extraction.

[1217] The server then formats the summarized text into an operational report format, using a template engine to fit the summary text into a predefined operational report format, generating a standardized report that can be easily viewed and shared by users.

[1218] Finally, the server converts the generated operation report into PDF format and sends it to the user via email. A notification is sent to the user's terminal, and the generated report file becomes available for download.

[1219] Specific examples

[1220] For example, if audio data is recorded while an autonomous vehicle is operating that says, "Today's operation went very smoothly. Thanks to the new sensor system, obstacle avoidance is faster and more accurate. Fuel efficiency has also improved, resulting in lower fuel costs than expected," the following actions will be taken.

[1221] 1. Vehicle-mounted device: Audio data recorded inside the vehicle is automatically uploaded to a cloud server.

[1222] 2. Cloud server: Receives the voice data and transcribes it using Google Cloud Speech-to-Text. For example, the resulting text might be something like, "Today's operation went very smoothly. Thanks to the new sensor system, obstacle avoidance is now faster and more accurate. Fuel efficiency has also improved, resulting in lower fuel costs than expected."

[1223] 3. Cloud server: The obtained text data is passed to a natural language processing engine such as OpenAI GPT, which generates a summary. The summary result will be something like, "The vehicle travels smoothly, and the sensor system allows for fast and accurate obstacle avoidance. Fuel efficiency has improved, reducing fuel costs."

[1224] 4. Cloud server: The summarized text is formatted into an operational report format. For example, it is formatted into the following format using a template engine.

[1225] ★Operation report:

[1226] The operation is smooth

[1227] Sensor system enables fast and accurate obstacle avoidance

[1228] Improved fuel efficiency and reduced fuel costs

[1229] 5. Cloud server: The completed operation report is converted to PDF format and sent to the user via email. A notification is sent to the user's device, and the user can download and view the report.

[1230] This system will enable automated collection and summarization of conversations and evaluations that occur in real time while autonomous vehicles are operating, and will enable efficient creation and distribution of operational reports.

[1231] Example prompts to input to a generative AI model:

[1232] "Today's operation went very smoothly. Thanks to the new sensor system, obstacle avoidance is faster and more accurate. Fuel efficiency has also improved, resulting in lower fuel costs than expected."

[1233] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1234] Step 1:

[1235] An in-vehicle device collects audio data recorded inside the vehicle, which is then automatically uploaded to a cloud server. The input is the audio inside the vehicle, and the output is the upload of the audio file to the cloud server.

[1236] Step 2:

[1237] The server receives voice data from the vehicle's onboard device. The received voice data is temporarily stored and passed to an AI voice recognition engine for conversion into text data. The input is an audio file, and the output is text data. Specifically, the voice data is converted into text format using a service such as Google Cloud Speech-to-Text.

[1238] Step 3:

[1239] The server passes the converted text data to a natural language processing engine to generate a summary. The input is text data, and the output is summarized text. Specifically, OpenAI GPT and Google Cloud Natural Language API are used to analyze the text data, extract important information, and summarize it.

[1240] Step 4:

[1241] The server formats the summarized text into an operational report format. A template engine is used to fit the summary text into a predefined operational report format. The input is the summarized text, and the output is a formatted operational report. Specifically, the server generates an operational report by inserting the text into a format defined by a tool such as Tex or ReportLab.

[1242] Step 5:

[1243] The server converts the formatted operation report into PDF format and sends it to the user via email. The input is the formatted operation report, and the output is the sent PDF file. Specifically, the PDF file is generated using ReportLab and sent via email using SendGrid or Amazon SES. The user receives a notification and can download and view the generated report file.

[1244] 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.

[1245] As an embodiment of the present invention, we provide a system for efficiently transcribing and summarizing audio data from a meeting, and creating and distributing minutes that incorporate sentiment analysis. This system uses a server, user terminals, and a data communication network. Furthermore, by incorporating an emotion recognition engine, the emotions of the meeting participants can also be reflected in the minutes.

[1246] First, after the conference ends, the user uploads the conference audio data to the server using a dedicated application. The dedicated application is installed on the user's device (PC, smartphone, tablet, etc.) and has the function to specify the audio file and send it to the server.

[1247] Server Operation

[1248] Server: When voice data is received, it is first temporarily stored. Next, the voice data is passed to an AI voice recognition engine to convert it into text data. The AI ​​voice recognition engine used here uses a commercial or open source API (e.g., a general-purpose voice recognition API) that provides highly accurate voice recognition.

[1249] The obtained text data is also passed to the emotion engine, which analyzes the emotions of the meeting participants. The emotion engine recognizes the emotion corresponding to each part of the text data and adds an emotion tag. The technology used in this process applies natural language processing algorithms such as voice emotion recognition and text emotion analysis.

[1250] The server then passes the emotion-tagged text data to a natural language processing engine to generate a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. During this process, the importance and summary content are adjusted based on the emotion data.

[1251] The server formats the summarized text into a minutes format. Using a template engine, the summary text is formatted into a predefined minutes format. Furthermore, sentiment tags and contextual information are added to the text, allowing for emotional changes to be detected.

[1252] Finally, the server sends the generated minutes to the user. This process includes sending an email and sending a notification via a dedicated application. A notification is displayed on the user's device, and the generated minutes file (e.g., PDF or Word format) can be downloaded.

[1253] Specific examples

[1254] For example, in the case of a project meeting in which audio data "regarding the progress of development of the next product" is recorded, the following operations are performed.

[1255] 1. User: Uses a dedicated application to upload the meeting recording data to the server.

[1256] 2. Server: Receives the voice data and transcribes it using a commercial speech recognition API. For example, the server can obtain text data such as, "Development of our next product is progressing smoothly. We are currently in the prototype stage, and testing is scheduled to begin next month."

[1257] 3. Server: The emotion engine is used to analyze emotions from the text data and add tags. For example, the server can obtain emotion-tagged text such as, "Development of the next product is progressing smoothly [Enjoyed]. It is currently in the prototype stage [Neutral], and testing is scheduled to begin next month [Expected]."

[1258] 4. Server: The emotion-tagged text data is passed to a natural language processing engine, which generates a summary. The summary result is something like, "Development of the next product is progressing smoothly, and testing is scheduled to begin next month."

[1259] 5. Server: Format the summarized text into a minutes format. For example, it can be formatted into the following format using a template engine.

[1260] ★Agenda: Progress of next product development

[1261] Development is going smoothly [Joy]

[1262] Testing is expected to begin next month [expected]

[1263] 6. Server: Converts the completed minutes into PDF format and sends them to the user's device. The user receives a notification and can download and view the minutes.

[1264] In this way, the present invention not only improves the efficiency and accuracy of creating meeting minutes, but also provides richer information by reflecting the emotions of the participants.

[1265] The processing flow will be explained below.

[1266] Step 1:

[1267] User: After the conference ends, the user launches the dedicated application, selects the conference audio data file, and uploads it to the server. When the user clicks the "Upload" button, the audio data is sent to the server as an HTTP request.

[1268] Step 2:

[1269] Server: The server receives the HTTP request and temporarily stores the audio data in storage. Meta information such as the storage location and file name is also recorded in a database.

[1270] Step 3:

[1271] Server: The server passes the saved voice data to the AI ​​voice recognition engine. At this time, it generates an API request and calls the voice recognition API by specifying the file path of the voice data.

[1272] Step 4:

[1273] Server: Receives the transcription results returned from the speech recognition API and saves them as text data. For example, the resulting text data might be something like, "Let's start the meeting. We'll discuss the next project."

[1274] Step 5:

[1275] Server: The transcribed text data is passed to the emotion engine for sentiment analysis. The emotion engine recognizes the sentiment associated with each part of the text data and adds emotion tags. For example, emotion-tagged text data such as "Let's start the meeting [Neutral]. We will discuss the next project [Expected]" is obtained.

[1276] Step 6:

[1277] Server: The emotion-tagged text data is passed to a natural language processing engine to generate a summary. The importance of the summary is adjusted based on the emotion data. For example, a summary such as "Project discussion [expectations]" is generated.

[1278] Step 7:

[1279] Server: The summarized text is formatted into minutes format using a template engine. Sentiment tags are also embedded appropriately.

[1280] Step 8:

[1281] Server: Convert the formatted minutes into PDF or Word format. During the conversion, the emotional information is displayed visually in an easy-to-understand manner.

[1282] Step 9:

[1283] Server: Sends the generated minutes file to the user's device. A download link is provided to the user via email or a notification function in a dedicated application.

[1284] Step 10:

[1285] User: The user receives a notification on their device and can download and review the generated minutes via the app or email. If necessary, they can make corrections or add additional comments.

[1286] These are the specific processing steps of the system that combines the emotion engine. This not only improves the efficiency of creating meeting minutes, but also provides rich information that reflects the emotions of the participants.

[1287] Example 2

[1288] 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."

[1289] There is a need to efficiently transcribe and summarize meeting audio data and generate minutes that reflect the emotions of meeting participants. However, there are currently no meeting minutes creation systems that include a means to add emotion tags. As a result, it is difficult to provide rich information that incorporates the emotions of participants rather than simply transcribing the content of meetings.

[1290] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1291] In this invention, the server includes means for receiving conference audio data, means for converting the audio data into text data, means for attaching emotion tags to the text data, means for summarizing the emotion-tagged text data, means for formatting the summarized text into minutes format, and means for transmitting the formatted minutes. This makes it possible to accurately and efficiently record the contents of the conference as minutes and provide rich information that also reflects the emotions of the participants.

[1292] "Conference audio data" means electronic data containing speech and other utterances recorded during a conference.

[1293] "Means for receiving" refers to the function or method for transferring voice data from the user terminal to the server.

[1294] The "means for converting voice data into text data" refers to a method or device that includes voice recognition technology for analyzing voice and converting it into text information.

[1295] "Means for assigning emotion tags to text data" refers to a technology or method that analyzes the emotion of speech in recognized text information and adds appropriate emotion labels.

[1296] A "means for summarizing emotion-tagged text data" is a technology or algorithm that analyzes emotion-tagged text information, extracts important content, and summarizes it concisely.

[1297] "Means for formatting the summarized text into minutes format" refers to a template engine or other formatting method for organizing the summarized content into the prescribed minutes format.

[1298] "Means for sending formatted minutes" includes communication means and processes for delivering the generated minutes to the user via email or a dedicated application.

[1299] A "server" is a computer system that receives, processes, stores, analyzes, and distributes audio data.

[1300] A "user terminal" is a device such as a personal computer, smartphone, or tablet that is used to record and upload audio data to a server.

[1301] A "natural language processing engine" is software that includes artificial intelligence technology that analyzes text data and performs advanced processing such as generating summaries and analyzing sentiment.

[1302] This invention provides a system that efficiently transcribes and summarizes audio data from a conference, and generates and distributes minutes that reflect the emotions of the conference participants. This system uses a server, user terminals, and a data communication network.

[1303] The system operates using the following hardware and software: User terminals include devices such as personal computers, smartphones, and tablets. The server uses commercial or open-source AI speech recognition engines (e.g., general-purpose speech recognition APIs), emotion recognition engines (e.g., emotion analysis APIs using natural language processing technology), natural language processing engines (e.g., generative AI models), and template engines (e.g., Jinja2).

[1304] After the conference ends, users use a dedicated application to upload the conference audio data to the server. The dedicated application is installed on the user's device (PC, smartphone, tablet, etc.) and has the function of selecting an audio file and sending it to a specified server.

[1305] When the server receives the voice data, it first temporarily stores it. Next, it passes the voice data to an AI voice recognition engine to convert it into text data. The resulting text data is also passed to an emotion engine, which analyzes the emotions of the meeting participants. The emotion engine recognizes the emotion corresponding to each part of the text data and adds an emotion tag. The technology used in this process applies natural language processing algorithms such as voice emotion recognition and text emotion analysis.

[1306] The server then inputs the emotion-tagged text data into a natural language processing engine to generate a summary. The natural language processing engine analyzes the input text data, extracts important information, and creates a summary. During this process, the importance and summary content are adjusted based on the emotion data. The server then formats the summarized text into a minutes format. Using a template engine, the summary text is fitted into a predefined minutes format. Emotion tags and contextual information are also added, and the text is formatted to show changes in emotion.

[1307] Finally, the server sends the generated minutes to the user. This process includes sending an email and sending a notification via a dedicated application. A notification is displayed on the user's device, and the generated minutes file (e.g., PDF or Word format) can be downloaded.

[1308] Specific examples

[1309] For example, in a project meeting in which audio data on the "progress of development of the next product" is recorded, the following process takes place: The user uses a dedicated application to upload the recording of the meeting (file name: meeting_audio.wav) to the server. The server receives the audio data and transcribes it using a commercial speech recognition API. For example, text data such as "Development of the next product is progressing smoothly. It is currently in the prototype stage, and testing is scheduled to begin next month" is obtained. The server then uses an emotion engine to analyze the emotions from the text data and add tags. For example, emotion-tagged text such as "Development of the next product is progressing smoothly [Joy]. It is currently in the prototype stage [Neutral], and testing is scheduled to begin next month [Expectation]" is obtained. The server then passes the emotion-tagged text data to a natural language processing engine to generate a summary. The summary result is "Development of the next product is progressing smoothly, and testing is scheduled to begin next month." The server then formats the summarized text into meeting minutes format. For example, a template engine can be used to format it into the following format:

[1310] ★Agenda: Progress of next product development

[1311] Development is going smoothly [Joy]

[1312] Testing is expected to begin next month [expected]

[1313] Finally, the server converts the completed minutes into PDF format and sends them to the user's device. The user receives a notification and can download and view the minutes.

[1314] Prompt Sentence Examples

[1315] "Please generate minutes that include the emotions of the participants from audio data of a meeting regarding the development progress of the next product."

[1316] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1317] Step 1:

[1318] The user opens the dedicated application. The user selects the conference audio data file (e.g., meeting_audio.wav) and clicks the upload button. The device sends the selected audio data to the server. The input is the conference audio data, and the output is the audio data sent to the server.

[1319] Step 2:

[1320] The server receives the voice data sent from the device and temporarily stores it. The storage destination is data storage (e.g., storage service), the input is the sent voice data, and the output is the voice data stored in the data storage. Specifically, the server receives an HTTP request and writes binary data to storage.

[1321] Step 3:

[1322] The server passes the saved voice data to the AI ​​voice recognition engine. Specifically, this is the process of sending the voice data to a voice recognition API and converting it into text data. The input is the saved voice data, and the output is text data (e.g., "Development of the next product is going well..."). The server makes an API request and receives text data as a response.

[1323] Step 4:

[1324] The server inputs the acquired text data into the emotion engine for sentiment analysis. The emotion engine recognizes the emotion corresponding to each part of the text and assigns an emotion tag. The input is text data, and the output is emotion-tagged text data (e.g., "Development of the next product is going smoothly [joy]..."). Specifically, it calls the sentiment analysis API and reflects the results in the text.

[1325] Step 5:

[1326] The server passes the emotion-tagged text data to a natural language processing engine, which generates a summary of the text. The natural language processing engine extracts important information based on the emotion tags and creates a summary text. The input is emotion-tagged text data, and the output is summary text (e.g., "Development of the next product is progressing smoothly, and testing is scheduled to begin next month"). Specifically, the server generates a summary using a natural language processing model.

[1327] Step 6:

[1328] The server passes the summarized text to a template engine to format it into minutes of meeting format. A template engine (e.g., Jinja2) is used to fit the summary text into a predefined format. The input is the summary text, and the output is text formatted in minutes of meeting format (e.g., "Development progress of the next product; development is going well [joy]..."). Specifically, the text is input into the template engine, and formatted text is generated.

[1329] Step 7:

[1330] The server converts the formatted minutes into PDF format. A PDF generation program (e.g., ReportLab) is used to create a PDF from text data. The input is text formatted in minutes format, and the output is a PDF minutes file. Specifically, the server converts the text into PDF using a PDF generation library.

[1331] Step 8:

[1332] The server sends the generated PDF to the user. The server uses the email sending function or the notification function of a dedicated application. The input is a PDF format minutes file, and the output is a notification or email to the user's device. Specifically, the file is sent using the email sending API or push notification function.

[1333] Step 9:

[1334] The terminal receives the notification from the server and displays it to the user. The user checks the notification, clicks the provided link to download the minutes, and checks the contents. The input is the notification sent from the server, and the output is the minutes file downloaded to the user. Specifically, the terminal displays the notification message and allows the user to operate the link.

[1335] (Application example 2)

[1336] 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."

[1337] Previously, creating meeting minutes was often done manually, which was inefficient and time-consuming. It was also difficult to create minutes that reflected the emotions of the meeting participants, making it difficult to accurately capture the atmosphere and key points of the meeting. This made it difficult to share information in real time during operational meetings at factories and other locations, which sometimes hindered efficient decision-making.

[1338] 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 conference audio data, means for converting the audio data into text data, means for summarizing the text data, means for formatting the summarized text into a minutes format, means for performing sentiment analysis, means for adding sentiment tags, and means for transmitting the formatted minutes. This enables the automated and efficient creation of conference minutes, and further enables the creation of minutes that reflect the emotions of the conference participants. This enables rapid and accurate information sharing and promotes efficient decision-making.

[1339] "Meeting audio data" refers to audio files that record statements and discussions made during a meeting.

[1340] A "means for receiving" refers to a system or method for obtaining data from another device or network.

[1341] "Text data" refers to character string data converted from voice data by a voice recognition system.

[1342] A "summarization means" is a method or system for extracting important information from given text data and summarizing it concisely.

[1343] A "minutes format" is a predefined format for organizing and recording the contents of a meeting.

[1344] A "formatting means" is a method or system for converting and preparing data or information into a particular form or format.

[1345] A "transmitting means" is a system or method for sending data to another device or network.

[1346] "Sentiment analysis" is a technology that detects emotions from text data or audio data and evaluates the type and intensity of those emotions.

[1347] An "emotion tag" is a label or mark that indicates the emotion contained within a piece of text.

[1348] As an embodiment of this invention, we will explain an example of an "operational meeting support app for factory robots" installed on factory robots. The following explanation is based on the process of efficiently collecting meeting audio data, transcribing it, summarizing it, and analyzing its sentiment, and finally generating and distributing meeting minutes with sentiment tags.

[1349] System Configuration

[1350] The system uses factory robots, servers, and a data communication network. The factory robots are equipped with built-in microphones that collect audio data from meetings. The server processes the audio data and uses a highly accurate speech recognition API, emotion recognition engine, and natural language processing engine. The main software and hardware components are described below.

[1351] Hardware

[1352] Factory robots: Collect audio data using built-in microphones.

[1353] Server: A high-performance server used for data processing and storage.

[1354] software

[1355] Speech Recognition API: A commercial or open-source API for high-accuracy speech recognition (e.g., Google Speech-to-Text API).

[1356] Emotion recognition engine: Technology for analyzing emotions in text data and adding tags (e.g., Microsoft Azure Cognitive Services' Text Analytics).

[1357] Natural language processing engine: A technology for summarizing text data (e.g., OpenAI GPT-3).

[1358] Template engine: Software for formatting summary text into minutes format.

[1359] PDF generation software: Software to convert minutes into PDF format (e.g. pdfkit).

[1360] Processing flow

[1361] 1. Collection of audio data

[1362] The factory robots are equipped with built-in microphones to collect audio data from meetings in real time, and the audio files are temporarily stored inside the robots and then sent to a server via a network.

[1363] 2. Transcription

[1364] When the server receives the audio data, it converts it into text using the Google Cloud Speech-to-Text API, allowing the contents of the meeting to be recorded as text.

[1365] 3. Sentiment analysis

[1366] The resulting text data is passed to the Microsoft Azure Cognitive Services Text Analytics API for emotional tagging, which recognizes the emotions (happiness, sadness, anticipation, etc.) that correspond to each part of the text data and adds emotional tags.

[1367] 4. Summary Generation

[1368] The sentiment-tagged text data is then summarised using OpenAI GPT-3, which performs text analysis to extract and summarise key information concisely.

[1369] 5. Formatting of minutes

[1370] The generated summary and emotion tags are then formatted into a predefined minutes format using a template engine, allowing participants to understand the contents of the meeting, including changes in emotion, at a glance.

[1371] 6. Results Distribution

[1372] The final minutes are converted to PDF format using PDF generation software, then uploaded to email or cloud storage and notified to the relevant parties.

[1373] Specific examples

[1374] For example, if the following audio data is collected:

[1375] "Development of our next product is progressing smoothly and is currently in the prototype stage. Testing is scheduled to begin next month. All team members are highly motivated."

[1376] An example prompt based on this is:

[1377] Example prompt sentence:

[1378] Generate a summary for the following text: "Development of our next product is progressing smoothly and is currently in the prototype stage. Testing is scheduled to begin next month. All team members are highly motivated."

[1379] The final summary results are as follows:

[1380] "Development of the next product is progressing smoothly, and testing is scheduled to begin next month. The entire team is highly motivated."

[1381] As described above, the present invention automates and improves the efficiency of creating meeting minutes, and also provides information that reflects the emotions of the participants, thereby realizing richer information sharing.

[1382] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1383] Step 1:

[1384] The user collects audio data from meetings in real time using the factory robot's built-in microphone. The collected audio data is temporarily stored inside the robot. The audio data is then sent to a server via a network. The input is audio data, and the output is an audio file sent to the server.

[1385] Step 2:

[1386] The server receives the voice data and temporarily stores it. The received voice data is passed to a high-precision voice recognition API (e.g., Google Speech-to-Text API), which converts the voice data into text data. The input is voice data, and the output is transcribed text data. Specifically, the server sends a voice recognition request, and the API analyzes the voice data and converts it into text.

[1387] Step 3:

[1388] The server passes the obtained text data to an emotion recognition engine (for example, the Text Analytics API from Microsoft Azure Cognitive Services) to analyze the emotions in the text data. The emotion recognition engine detects the emotions corresponding to each part of the text and adds emotion tags. The input is text data, and the output is emotion-tagged text data. Specifically, the server sends an emotion analysis request, and the API analyzes the text and adds emotion tags.

[1389] Step 4:

[1390] The server passes emotion-tagged text data to a natural language processing engine (e.g., OpenAI GPT-3) to extract key information and generate a summary. The input is emotion-tagged text data, and the output is the summary text. Specifically, a summary generation request is sent, and the generative AI model analyzes the text and creates a summary.

[1391] Step 5:

[1392] The server passes the summarized text and emotion tags to a template engine to format it into a predefined minutes format. The input is the summary text and emotion-tagged text data, and the output is text formatted in the minutes format. Specifically, the template engine rearranges the text data into a predetermined format and performs additional processing to visually display the emotion tag information.

[1393] Step 6:

[1394] The server converts the formatted minutes into PDF format and sends it to the user's terminal. The input is text formatted in the minutes format, and the output is a PDF minutes file. Specifically, the server converts the text data into a PDF file using PDF generation software (e.g., pdfkit), and then sends the PDF file by email or uploads it to cloud storage.

[1395] These are the processing steps of the factory robot operation meeting support app, which efficiently transcribes the audio data of meetings, automatically generates and distributes meeting minutes that have been summarized and analyzed for sentiment.

[1396] 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.

[1397] 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.

[1398] 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.

[1399] 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.

[1400] FIG. 9 is a diagram illustrating 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 actions 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.

[1401] 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.

[1402] 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).

[1403] 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.

[1404] 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."

[1405] 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.

[1406] 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).

[1407] 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.

[1408] 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.

[1409] 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.

[1410] 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.

[1411] The hardware resource for executing a specific process can be any of the following processors: An example of a processor 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 example of a 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.

[1412] 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.

[1413] 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.

[1414] 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.

[1415] 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.

[1416] 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.

[1417] The following is further disclosed regarding the above embodiment.

[1418] (Claim 1)

[1419] means for receiving audio data of the conference;

[1420] means for converting voice data into text data;

[1421] a means for summarizing text data;

[1422] a means for formatting the summarized text into minutes format;

[1423] The system includes a means for transmitting formatted minutes.

[1424] (Claim 2)

[1425] 2. The system according to claim 1, wherein audio data of a conference is uploaded from a user terminal to a server.

[1426] (Claim 3)

[1427] 2. The system according to claim 1, wherein the text data is summarized using a natural language processing engine.

[1428] "Example 1"

[1429] (Claim 1)

[1430] means for receiving audio data of the conference;

[1431] A means for converting voice data into text data using highly accurate voice recognition technology;

[1432] A means for summarizing text data using natural language processing technology;

[1433] A means for formatting the summarized text into a minutes format using a template engine;

[1434] A system including means for delivering formatted minutes to a user terminal.

[1435] (Claim 2)

[1436] 2. The system according to claim 1, wherein audio data of a conference is uploaded from a user terminal to a server.

[1437] (Claim 3)

[1438] The system of claim 1, wherein the text data is summarized using a generative AI model and a prompt sentence.

[1439] "Application Example 1"

[1440] (Claim 1)

[1441] means for receiving audio data of the conference;

[1442] means for converting voice data into text data;

[1443] a means for summarizing text data;

[1444] a means for formatting the summarized text;

[1445] means for transmitting the formatted report;

[1446] A device installed in a car that records audio data from the meeting and automatically uploads the audio file;

[1447] A cloud server that receives audio files and transcribes them;

[1448] A means to summarize the text data using a natural language processing engine and format it into a standard operational report format;

[1449] A system that includes a means for converting formatted operational reports into PDF format and sending them by email.

[1450] (Claim 2)

[1451] 2. The system according to claim 1, wherein conference audio data is uploaded to the server from a user terminal or a vehicle-mounted device.

[1452] (Claim 3)

[1453] 2. The system according to claim 1, wherein the text data is summarized using a natural language processing engine.

[1454] "Example 2: Combining Emotion Engines"

[1455] (Claim 1)

[1456] means for receiving audio data of the conference;

[1457] means for converting voice data into text data;

[1458] A means for attaching emotion tags to text data;

[1459] a means for summarizing emotion-tagged text data;

[1460] a means for formatting the summarized text into minutes format;

[1461] The system includes a means for transmitting formatted minutes.

[1462] (Claim 2)

[1463] 2. The system according to claim 1, wherein audio data of a conference is uploaded from a user terminal to a server.

[1464] (Claim 3)

[1465] 2. The system according to claim 1, wherein the text data is summarized using a natural language processing engine.

[1466] "Application example 2 when combining emotion engines"

[1467] (Claim 1)

[1468] means for receiving audio data of the conference;

[1469] means for converting voice data into text data;

[1470] a means for summarizing text data;

[1471] a means for formatting the summarized text into minutes format;

[1472] a means for performing sentiment analysis;

[1473] a means for adding an emotion tag;

[1474] The system includes a means for transmitting formatted minutes.

[1475] (Claim 2)

[1476] 2. The system according to claim 1, wherein audio data of a conference is uploaded from a user terminal to a server.

[1477] (Claim 3)

[1478] 2. The system according to claim 1, wherein the text data is summarized using a natural language processing engine. [Explanation of symbols]

[1479] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for receiving audio data of the conference; means for converting voice data into text data; a means for summarizing text data; a means for formatting the summarized text into minutes format; The system includes a means for transmitting formatted minutes.

2. 2. The system according to claim 1, wherein conference audio data is uploaded from the user terminal to the server.

3. 2. The system according to claim 1, wherein the text data is summarized using a natural language processing engine.

Images