system

Abstract

Provide a system. The method includes: a means for recording a user's conversation in real time; A means for uploading the recorded audio files to a central server; A means for converting the voice files stored in the central server into text using a voice recognition system; A means of analyzing the text data using natural language processing technology and converting it into minutes format; A means for searching past minutes from a database in response to a search request from a user; When search results are incomplete, generative AI can be used to fill in the gaps. The system includes a means for displaying the minutes including the completed information so that the user can review them.

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 modern society, the number of business and personal conversations and discussions is increasing, making it necessary to accurately record important conversations and access them later. However, manually recording conversations is time-consuming and laborious, and human memory is imperfect, so there is a high chance of missing important information. Therefore, there is a need for a system that can automatically record and analyze conversations and easily search them. [Means for solving the problem]

[0005] The present invention solves the above-mentioned problems with a system that includes: means for recording user conversations in real time; means for uploading the recorded audio files to a central server; means for converting the audio files stored on the central server into text using a voice recognition system; means for analyzing the converted text using natural language processing technology and converting it into minutes format; means for searching a database for past minutes in response to a search request from a user; means for using generative AI to fill in missing parts if the search results are incomplete; and means for displaying the minutes including the filled information so that the user can review them. The present invention allows users to easily search and refer to past conversations and minutes, thereby filling in incomplete memories and improving productivity.

[0006] "User" refers to an individual or organization that uses the system to record conversations and later search and reference those conversations.

[0007] "Real-time recording" refers to the process of using a recording device to digitally capture audio at the exact moment a conversation is taking place.

[0008] "Server" means a central control unit or system for receiving, storing, and analyzing recorded audio files.

[0009] "Audio file" refers to an audio data file that digitally records the contents of a conversation.

[0010] A "speech recognition system" refers to a software or hardware system that uses machine learning algorithms to analyze input voice data and convert it into text data.

[0011] "Text conversion" refers to the process of converting audio data into text data.

[0012] "Natural language processing technology" refers to technology that enables computers to understand and analyze human language and extract and process information.

[0013] "Converting into minutes format" refers to the process of analyzing text data, summarizing and organizing the contents of meetings and conversations, and reconstructing them into a document format that is easy to remember.

[0014] "Search Request" means an instruction or query sent by a User to the System to search for a particular conversation or transcript.

[0015] A "database" refers to a digital storage system for systematically storing and managing information such as audio files, text data, and minutes.

[0016] "Generative AI" refers to artificial intelligence techniques used to fill in incomplete information using existing data.

[0017] "Completing" refers to the process of adding or inferring missing information to complete a picture.

[0018] "Minutes" refers to a document that records the contents of a meeting or conversation.

[0019] "Displaying" refers to the process of visually presenting information to a user through a user interface. [Brief explanation of the drawings]

[0020] [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

[0021] 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.

[0022] First, the terms used in the following description will be explained.

[0023] 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).

[0024] 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.

[0025] 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.

[0026] 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.

[0027] 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."

[0028] [First embodiment]

[0029] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0030] 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.

[0031] 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).

[0032] 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.

[0033] 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.

[0034] 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.

[0035] 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.

[0036] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0037] 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.

[0038] 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.

[0039] 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.

[0040] 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."

[0041] The present invention relates to a system for electronically recording user conversations, analyzing and supplementing the content of the recorded conversations, and automatically generating minutes of the conversations. Hereinafter, an embodiment of the present invention will be described in detail.

[0042] System Overview

[0043] 1. Transcript of conversation:

[0044] (User) The user launches the dedicated app and gives the command to "start recording."

[0045] (Device) The device uses a microphone to record the user's conversation in real time and stores the recorded data in temporary storage.

[0046] 2. Uploading audio data:

[0047] (Device) When recording is complete, an audio file is generated and metadata such as the date, time, location, and participants is added.

[0048] (Device) Uploads the recording file to a central server in accordance with security protocols.

[0049] 3. Central server for data storage and speech recognition:

[0050] (Server) Receives the uploaded audio file and stores it in the database.

[0051] (Server) The saved audio file is passed to a speech recognition system and converted into text data.

[0052] 4. Generating meeting minutes using natural language processing:

[0053] (Server) The textual data is input into a natural language processing (NLP) system and analysis is performed.

[0054] (Server) The NLP system extracts important information and conversational highlights from the text data and converts it into minutes format.

[0055] (Server) The generated minutes are saved in a database, and users can refer to them later.

[0056] 5. Search and memory completion functions:

[0057] (User) Using a dedicated app, users can search for past minutes by specifying specific keywords or dates and times.

[0058] (Terminal) The terminal that receives the user's search request sends it to the central server.

[0059] (Server) The server performs a search based on the specified keywords in the database and extracts relevant minutes.

[0060] (Server) If the search results are incomplete, generative AI is used to infer and complete the missing parts.

[0061] (Terminal) The minutes including the completed information are presented to the user and displayed visually.

[0062] Specific examples

[0063] Example 1: Meeting recording

[0064] (User) "I'll record the meeting starting at 10 o'clock today," and opens the dedicated app.

[0065] (Device) The app launches and you press the record button to record the start of the meeting.

[0066] (Device) The device records all conversations, stops recording when the meeting ends, and generates an audio file.

[0067] (Device) Add metadata such as "10:00 Meeting October 20, 2023" to the audio file and upload it to the central server.

[0068] Example 2: Search and complete past conversations

[0069] (User) "I want to check what was decided at last year's project meeting," so he enters the search keyword "Project 2022" into the app.

[0070] (Terminal) Receives the user's search request and sends it to the central server.

[0071] (Server) Search the database for relevant minutes and extract those that match the specified keywords.

[0072] (Server) If the results are incomplete, the generative AI will complete the relevant information and add details.

[0073] (Terminal) The completed minutes are displayed to the user, who then confirms the contents.

[0074] By going through these steps, users can automatically record, complete, and search the contents of important conversations and meetings, significantly improving the efficiency of their daily work. This invention will contribute to improving productivity, especially in the business world.

[0075] The processing flow will be explained below.

[0076] Step 1:

[0077] (User) The user launches the dedicated app and gives the command to "start recording."

[0078] (Device) The app will launch a voice recording program via voice recognition or tap.

[0079] (Device) The device will begin recording the conversation in real time using its microphone.

[0080] Step 2:

[0081] (Device) During recording, audio data is temporarily stored in a buffer.

[0082] (Device) Converts the recorded data into a file format (e.g. WAV, MP3) at regular intervals.

[0083] (Device) Adds metadata (date, time, location, participants) to the generated audio file.

[0084] Step 3:

[0085] (Device) Once recording is complete, prepare the audio file for uploading to the central server.

[0086] (Device) Sends the audio file to a central server using a secure protocol (e.g., HTTPS).

[0087] (Server) Receives the uploaded audio file.

[0088] Step 4:

[0089] (Server) Save the received audio file in the database.

[0090] (Server) Prepares the saved audio file for passing to the speech recognition system.

[0091] Step 5:

[0092] (Server) The speech recognition system analyzes the audio file and converts the audio data into text data.

[0093] (Server) Obtain the converted text data and proceed to the next processing step.

[0094] Step 6:

[0095] (Server) Inputs text data into a natural language processing (NLP) system.

[0096] (Server) NLP system analyzes the text data and extracts important information (e.g., decisions, action items).

[0097] Step 7:

[0098] (Server) Convert the extracted information into minutes format.

[0099] (Server) Save the generated minutes in the database and prepare to notify the user.

[0100] Step 8:

[0101] (User) When you want to look back on a past conversation, you can enter search keywords using a dedicated app.

[0102] (Terminal) Receives the user's search request and sends it to the central server.

[0103] Step 9:

[0104] (Server) Search for minutes based on specified keywords in the database.

[0105] (Server) If the search results are incomplete, prepare to fill in the missing parts using generative AI.

[0106] Step 10:

[0107] (Server) The generative AI predicts the missing parts and generates new minutes based on the supplemented information.

[0108] (Server) Sends the completed minutes to the user's terminal.

[0109] Step 11:

[0110] (Terminal) The minutes sent are displayed on the user interface.

[0111] (User) Check the completed results and obtain the required information.

[0112] Example 1

[0113] 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."

[0114] In today's business environment, many meetings and discussions are held, and it is necessary to record their contents and organize them for later reference. However, manual recording and minutes creation is time-consuming and prone to human error. Furthermore, it is difficult to efficiently search and reference past minutes, resulting in reduced work efficiency. It is also difficult to supplement information based on incomplete records. To solve these problems, a system is needed that can automatically create minutes using real-time conversation recording, speech recognition, and natural language processing, as well as highly accurate search and completion functions.

[0115] 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.

[0116] In this invention, the server includes a means for recording user conversations in real time, a means for adding metadata such as date, time, location, and participants to the recorded audio files and uploading them to a central server, and a means for sequentially converting the audio files stored on the central server into text using voice recognition technology. This allows users to efficiently record and manage conversation content and automatically generate important conversation content in the form of minutes. Users can also search past minutes based on specific keywords or dates and fill in missing parts using generative artificial intelligence, allowing them to quickly and accurately refer to the information they need.

[0117] "User" refers to a person who uses the system to record, upload, search, and check minutes of meetings.

[0118] A "terminal" is a device operated by a user, and is an electronic device that has a microphone, speaker, display, and Internet connection capabilities.

[0119] "Server" refers to a central computer system that stores audio files, processes speech recognition, natural language processing, generates and stores minutes, and performs search and completion processing.

[0120] "Conversation recording" refers to the act of collecting a user's speech as digital audio data using a microphone.

[0121] "Central Server" means a computer system located at the center of the system, which stores, analyzes, generates, searches, and completes voice data.

[0122] "Speech recognition technology" is a technology for converting voice data into text data, and is the process of analyzing voice signals and extracting their contents as text information.

[0123] "Natural language processing technology" is a technology that analyzes text data and performs processes such as semantic understanding, information extraction, and summary generation.

[0124] A "minutes format" is a document that organizes and records the contents of a conversation or meeting, and includes the date, time, location, participants, and what was said.

[0125] A "search request" is a request made by a user to locate past minutes based on specific keywords or dates.

[0126] "Data storage" refers to a memory area on a server for storing voice data, text data, minutes data, etc.

[0127] "Generative AI" is an AI technology that complements and generates new data based on existing information.

[0128] "Metadata" is auxiliary information added to an audio file, and includes attribute information such as date, time, location, and participants.

[0129] A "user interface" is an interface that allows users to operate the system, and provides functions such as voice recording, search, and checking minutes.

[0130] The present invention relates to a system for electronically recording user conversations, analyzing and supplementing the content of the recorded conversations, and automatically generating minutes of the conversations. Hereinafter, an embodiment of the present invention will be described in detail.

[0131] Conversation Record

[0132] (User) The user launches the dedicated app and instructs it to "start recording." Specifically, the user presses the "start recording button" using the smartphone app, which starts the recording process.

[0133] (Device) The device (such as a smartphone or tablet) uses a microphone to record the user's conversation in real time and saves the recorded data in temporary storage.

[0134] Uploading audio data

[0135] (Device) When the recording is complete, an audio file is generated and metadata such as the date, time, location, and participants is added. For example, metadata is added in the format "Meeting held at 10:00 on October 20, 2023."

[0136] (Device) Uploads the recording file to a central server following a security protocol. Specifically, the device sends the audio file to the server using an HTTPS connection.

[0137] Central server for data storage and voice recognition

[0138] (Server) Receives the uploaded audio files and stores them in a database. The server database has a table called "received_audio_files" that stores audio files and metadata.

[0139] (Server) The saved audio file is converted into text data using speech recognition technology (e.g., Google® Cloud Speech-to-Text API). For example, an audio file of "10 o'clock meeting" is converted into text such as "The agenda for the meeting is to review the budget."

[0140] Generating meeting minutes using natural language processing

[0141] (Server) The text data is input into natural language processing technology (e.g., Hugging Face's Transformers library) and analyzed. The text data is analyzed into a format such as "Important matters: budget, person in charge: Tanaka."

[0142] (Server) Extracts important information and key points from the text data and converts them into minutes format. For example, minutes may be converted into a format such as "Meeting name: 10:00 Meeting, Date and time: October 20, 2023, Content: Budget review will be on the agenda."

[0143] (Server) The generated minutes are saved in the database. The minutes data is stored in a table called "meeting_minutes".

[0144] Search and memory completion functions

[0145] (User) Search for past minutes by specifying specific keywords or dates using the dedicated app. For example, enter "Project 2022" in the app and press the search button.

[0146] (Device) The device receives the user's search request and sends it to the central server. The search query is sent to the server's API.

[0147] (Server) Perform a search based on the specified keywords in the database to extract relevant minutes, for example, filtering minutes that match the keywords "project" and "2022".

[0148] (Server) If the search results are incomplete, generative artificial intelligence (e.g., OpenAI (registered trademark) GPT-4 (registered trademark)) is used to infer and complete the missing parts. For example, GPT-4 completes the missing part about "project progress" and adds more detailed information.

[0149] (Device) The minutes containing the supplemented information are presented to the user, who confirms the contents. The supplemented minutes are displayed on the app screen.

[0150] Specific examples

[0151] Example 1: Meeting recording

[0152] (User) Opens the dedicated app and instructs, "I want to record the meeting starting at 10:00 today."

[0153] (Device) The app launches and you press the record button to record the start of the meeting.

[0154] (Device) The device records all conversations, stops recording when the meeting ends, and generates an audio file.

[0155] (Device) Add metadata such as "10:00 Meeting October 20, 2023" to the audio file and upload it to the central server.

[0156] Example 2: Search and complete past conversations

[0157] (User) In the app, he / she thinks, "I want to check what was decided at last year's project meeting," and enters the search keyword "Project 2022."

[0158] (Terminal) Receives the user's search request and sends it to the central server.

[0159] (Server) Search the database for relevant minutes and extract those that match the specified keywords.

[0160] (Server) If the results are incomplete, generative artificial intelligence (e.g., OpenAI GPT-4) will complete the relevant information and add details.

[0161] (Terminal) The completed minutes are displayed to the user, who then confirms the contents.

[0162] As described above, the present invention is a system that efficiently records user conversations and automatically converts important content into minutes format, thereby improving business efficiency.

[0163] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0164] Step 1:

[0165] Recording conversations

[0166] (User) Launches the dedicated app and instructs it to "start recording." This causes the device to receive as input the time the user started recording and the corresponding action.

[0167] (Device) Records user conversation in real time using a microphone. Audio data is collected sequentially and saved in temporary storage. An audio data file is generated as output.

[0168] Step 2:

[0169] Uploading audio data

[0170] (Device) When recording is complete, an audio file is generated and metadata such as the date, time, location, and participants are added. The command to stop recording and the metadata are given as input. This saves the audio file and metadata together.

[0171] (Device) Uploads the audio file to the central server using the HTTPS security protocol. The output is the audio file and metadata sent to the central server.

[0172] Step 3:

[0173] Central server for data storage and voice recognition

[0174] (Server) Receives uploaded audio files and stores them in a database. Receives audio files and metadata sent from the device as input. Stores audio files and metadata in a database as output.

[0175] (Server) The saved audio file is passed to speech recognition technology and converted into text data. Specifically, the Google Cloud Speech-to-Text API is used. The audio file is given as input, and the audio content is converted into text data as output.

[0176] Step 4:

[0177] Generating meeting minutes using natural language processing

[0178] (Server) The text data is input into natural language processing technology (Hugging Face's Transformers library) and analyzed. Text data is given as input, and text data with key points extracted is obtained as output.

[0179] (Server) The NLP system converts important information and conversational points from text data into a transcript format. The parsed text data is given as input, and formatted transcripts are generated as output.

[0180] (Server) The generated minutes are saved in a database. The minutes data is given as input, and saved in a database as output.

[0181] Step 5:

[0182] Search and memory completion functions

[0183] (User) uses a dedicated app to input specific keywords and dates to search for past minutes. The search keywords and dates are given as input, and the device sends a search request to the central server.

[0184] (Terminal) Receives the user's search request and sends it to the central server. The user's search keywords and date and time are given as input, and the search request is sent to the server as output.

[0185] (Server) Performs a search based on the specified keywords in the database and extracts relevant minutes. The search query is given as input, and the relevant minutes are extracted as output.

[0186] (Server) If the search results are incomplete, generative artificial intelligence (OpenAI GPT-4) is used to infer and complete the missing parts. The extracted incomplete minutes data is given as input, and the completed minutes are obtained as output.

[0187] (Terminal) Present the minutes containing the completed information to the user. The completed minutes data is given as input, and is displayed on the user's terminal as output.

[0188] As described above, the specific actions, data processing, and data calculations performed at each step form a series of flows, which realize efficient recording of conversations, generation of minutes, and search and completion functions.

[0189] (Application example 1)

[0190] 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."

[0191] Conversations held inside autonomous vehicles are likely to contain important information, but there is no appropriate means to efficiently record the content and use it as minutes. There is also a need for a simple and effective process for uploading and searching audio files, as well as for completing incomplete minutes. This invention solves these problems and provides a system for efficiently recording, searching, and completing conversations inside autonomous vehicles and using them as minutes.

[0192] 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.

[0193] In this invention, the server includes means for recording user conversations in real time, means for uploading recorded audio files to a central server, means for converting the audio files stored in the central server into text using a voice recognition system, means for analyzing the converted text using natural language processing technology and converting it into minutes format, means for searching past minutes from a database in response to a search request from a user, means for completing missing parts using generative AI when the search results are incomplete, means for displaying the minutes including the completed information so that the user can check them, means for recording conversations inside an autonomous vehicle, adding metadata to the recorded conversations, means for creating minutes including specific dates, times, and locations based on the metadata, and means for processing search requests based on the vehicle's location and ID information. This makes it possible to efficiently record conversations inside an autonomous vehicle and increase the availability of minutes.

[0194] "Recording a user's conversation in real time" means recording the user's voice immediately on the spot using an electronic device.

[0195] "Uploading the recorded audio file to the central server" means transferring the audio data stored in the device to a central storage device via a network.

[0196] "Converting voice files stored on a central server into text using a voice recognition system" means converting voice data stored in a central storage device into text data using voice recognition technology.

[0197] "Analyzing the text data using natural language processing technology and converting it into minutes format" means analyzing the data converted from audio to text using natural language processing technology, extracting summaries and important points, and formatting it into an official record format.

[0198] "Searching for past minutes from a database in response to a search request from a user" means receiving a search request from a user based on specific conditions and searching for past records in a database in accordance with that request.

[0199] "Using generative AI to fill in missing parts when search results are incomplete" means that when the results obtained through a search are incomplete, generative AI is used to supplement and add missing information.

[0200] "Displaying the minutes including the supplemented information so that the user can check them" means visually presenting the final minutes to the user, including the information supplemented by the generative AI.

[0201] "Recording conversations inside an autonomous vehicle and uploading them with metadata" means recording voice conversations inside an autonomous vehicle and sending the audio files to a central server along with additional information such as date, time, and location.

[0202] "Creating minutes with specific dates, times, and locations based on metadata" means using additional information (metadata) to generate an official record based on the content of conversations that took place at specific times and locations.

[0203] "Processing search requests based on vehicle location and ID information" means accepting search requests from users based on the vehicle's current location and identification information, and locating and providing appropriate records.

[0204] The present invention relates to a system for efficiently recording conversations in an autonomous vehicle and using the recorded conversations as minutes. Detailed embodiments of the present invention will be described below.

[0205] System configuration

[0206] User device: A smartphone or in-car smart device used to start and stop recording conversations and upload audio data to a central server.

[0207] Central server: Responsible for storing recording data, speech recognition, natural language processing, and generating meeting transcripts.

[0208] Database: A storage system for storing audio files, text data, and generated transcripts.

[0209] Generative AI: Used to complete incomplete search results and generate detailed meeting transcripts.

[0210] Program processing

[0211] Record and upload conversations

[0212] 1. The user device launches the app inside the autonomous vehicle and selects "start recording."

[0213] 2. The device uses the vehicle's microphone to record the user's conversation in real time and temporarily stores the audio data.

[0214] 3. When the recording is complete, an audio file is generated and metadata such as the date and time, vehicle ID, and location information is added to the file. This metadata specifically includes GPS information and the vehicle's identification ID.

[0215] 4. The device uploads the audio file to a central server following security protocols.

[0216] Audio data processing

[0217] 5. The central server receives the uploaded audio files and stores them in a database.

[0218] 6. The server converts the voice data into text using a speech recognition system (e.g., Google Cloud Speech-To-Text API).

[0219] 7. The text data is fed into a natural language processing system based on Hugging Face's transformers library, which extracts and summarizes important information and key points of the conversation.

[0220] 8. The generated minutes are stored in a database for users to refer to later.

[0221] Search and complete meeting minutes

[0222] 9. The user uses the app to search for minutes by specifying specific keywords, date and time, and vehicle location information.

[0223] 10. The central server uses ElasticSearch® to search for the relevant minutes in the database and provides the results to the user.

[0224] 11. If the search results are incomplete, a generative AI (e.g., GPT-3 (registered trademark)) will fill in the missing parts and generate a completed transcript.

[0225] 12. The completed minutes will be displayed on the terminal for visual confirmation.

[0226] Specific examples

[0227] Meeting minutes

[0228] User: When starting a sales strategy meeting in an autonomous vehicle, the user enters into the app, "I will record today's sales strategy meeting. It will start at 10:00 and will be held at the Tokyo branch."

[0229] Server: Records conversations during meetings, adds metadata, and uploads them to a central server.

[0230] Server: The uploaded audio data is converted into text and minutes format, and stored in a database.

[0231] Search past minutes

[0232] User: Type into the app, "Find minutes from last December's marketing meeting."

[0233] Server: Searches the database using Elasticsearch and displays the relevant minutes on the user's terminal.

[0234] Server: If the search results are incomplete, the generative AI fills in the gaps and provides information.

[0235] This makes it possible to efficiently record conversations and discussions inside an autonomous vehicle and easily search for and supplement the necessary information.

[0236] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0237] Step 1:

[0238] The user launches the dedicated app in the autonomous vehicle and selects "start recording." This starts the recording function.

[0239] Step 2:

[0240] The device uses the vehicle's microphone to record the user's conversation in real time. The recorded data is temporarily stored in the device's storage. The input is the user's voice data, and the output is a temporary audio file.

[0241] Step 3:

[0242] When the user selects "stop recording," the device stops recording and generates an audio file. This audio file is assigned metadata such as the date and time, vehicle ID, and location information. The input is a temporary audio file and metadata, and the output is an audio file with metadata.

[0243] Step 4:

[0244] The device uploads the generated audio file to a central server according to a security protocol. The input is the audio file with metadata, and the output is the saved audio file on the central server.

[0245] Step 5:

[0246] The server receives the uploaded audio files and stores them in a database. The input is the audio file, and the output is the audio data stored in the database.

[0247] Step 6:

[0248] The server passes the audio file to a speech recognition system and converts it into text data, for example, using the Google Cloud Speech-To-Text API. The input is audio data, and the output is text data.

[0249] Step 7:

[0250] Text data is input into a natural language processing system to extract and analyze important information and key points of the conversation. This analysis uses the Hugging Face transformers library. The input is text data, and the output is summary data in the form of meeting minutes.

[0251] Step 8:

[0252] The server saves the generated minutes in a database so that users can refer to them later. The input is summary data in minutes format, and the output is minutes saved in the database.

[0253] Step 9:

[0254] The user searches for meeting minutes using a dedicated app by specifying specific keywords, date and time, and vehicle location information. This operation sends a search request to the server. The input is a search query, and the output is a search request.

[0255] Step 10:

[0256] The server uses Elasticsearch to search for the relevant minutes in the database and provide the results to the user. The input is the search request and the output is the search results.

[0257] Step 11:

[0258] If the search results are incomplete, we use a generative AI (e.g., GPT-3) to fill in the gaps and generate a complete transcript. The input is the incomplete search results, and the output is the completed transcript.

[0259] Step 12:

[0260] The completed minutes are displayed on the user's terminal, allowing the user to visually check them. The input is the completed minutes, and the output is the minutes displayed on the user's terminal.

[0261] 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.

[0262] The present invention relates to a system that electronically records user conversations, analyzes and supplements the content, and automatically formats it into minutes. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, emotional information is added to the conversation content, generating more detailed minutes.

[0263] System Overview

[0264] 1. Transcript of conversation:

[0265] (User) The user launches the dedicated app and gives the command to "start recording."

[0266] (Terminal) The terminal launches a voice recording program and begins recording the conversation in real time using the microphone.

[0267] (Device) At the same time, the emotion engine is also activated, analyzing emotions from the user's conversation in real time.

[0268] 2. Integration and upload of voice and emotion data:

[0269] (Device) During recording, audio data is temporarily stored in a buffer.

[0270] (Device) Along with the recorded data, emotion data recognized by the emotion engine is also recorded.

[0271] (Device) Convert the recorded data into a file format (e.g. WAV, MP3) at regular intervals and add emotional information as metadata.

[0272] (Device) Upload the audio file to a central server using a secure protocol (e.g., HTTPS).

[0273] 3. Central server for data storage and speech recognition:

[0274] (Server) Receives the uploaded audio file and emotion metadata and stores them in a database.

[0275] (Server) The saved audio file is passed to a speech recognition system and converted into text data.

[0276] 4. Generating meeting minutes using natural language processing:

[0277] (Server) The textual data is input into a natural language processing (NLP) system and analysis is performed.

[0278] (Server) The NLP system extracts important information and key points of the conversation from the text data and converts them into minutes format that reflects emotional information.

[0279] (Server) The generated minutes are saved in a database, and users can refer to them later.

[0280] 5. Search and memory completion functions:

[0281] (User) Using a dedicated app, users can search for past minutes by specifying specific keywords or dates and times.

[0282] (Terminal) The terminal that receives the user's search request sends it to the central server.

[0283] (Server) The server searches the minutes based on the specified keywords in the database and extracts relevant minutes.

[0284] (Server) If the search results are incomplete, prepare to fill in the missing parts using generative AI.

[0285] (Server) The generative AI predicts the missing parts and generates new minutes based on the supplemented information.

[0286] (Terminal) The minutes including the completed information are presented to the user and displayed visually.

[0287] Specific examples

[0288] Example 1: Meeting recording and sentiment analysis

[0289] (User) "I'll record the meeting starting at 10 o'clock today," and opens the dedicated app.

[0290] (Device) The app launches and you press the record button to record the start of the meeting.

[0291] (Device) The device records all conversations, and the emotion engine analyzes and records emotions from the conversation.

[0292] (Device) Stop recording when the meeting ends and generate an audio file.

[0293] (Device) The audio file is given metadata such as "10:00 Meeting, October 20, 2023," as well as emotional data, and then uploaded to a central server.

[0294] Example 2: Search and complete past conversations

[0295] (User) "I want to check what was decided at last year's project meeting," so he enters the search keyword "Project 2022" into the app.

[0296] (Terminal) Receives the user's search request and sends it to the central server.

[0297] (Server) Search the database for relevant minutes and extract those that match the specified keywords.

[0298] (Server) If the results are incomplete, the generative AI will complete the relevant information and add details.

[0299] (Terminal) The completed minutes are displayed to the user, who then confirms the contents.

[0300] The system allows users to obtain not just minutes, but detailed records that reflect the emotions expressed during the conversation, leading to a deeper understanding of the conversation and more reliable decision-making.

[0301] The processing flow will be explained below.

[0302] Step 1:

[0303] (User) The user launches the dedicated app and gives the command to "start recording."

[0304] (Device) The app will launch the voice recording program and emotion engine and begin recording your conversation in real time using the microphone.

[0305] (Device) The emotion engine simultaneously analyzes the user's tone of voice and vocabulary to generate emotion data.

[0306] Step 2:

[0307] (Device) During recording, the voice data and emotion data are temporarily stored in a buffer.

[0308] (Device) Convert the recorded data into a file format (e.g. WAV, MP3) at regular intervals and add emotional information as metadata.

[0309] Step 3:

[0310] (Device) Once the recording is complete, prepare the audio file and emotion data for uploading to the central server.

[0311] (Device) Sends the audio file to a central server using a secure protocol (e.g., HTTPS).

[0312] (Server) Receives the uploaded audio file and emotional metadata.

[0313] Step 4:

[0314] (Server) Store the received audio files and emotion data in a database.

[0315] (Server) The saved audio file is passed to a speech recognition system and converted into text data.

[0316] Step 5:

[0317] (Server) The speech recognition system converts the speech data into text data.

[0318] (Server) Emotion data is extracted along with the converted text data, and the process proceeds to the next step.

[0319] Step 6:

[0320] (Server) Input text data and emotion data into a natural language processing (NLP) system.

[0321] (Server) The NLP system analyzes the text data and extracts key information (e.g., decisions, action items) and sentiment changes.

[0322] Step 7:

[0323] (Server) Based on the extracted information, convert it into a minutes format that reflects changes in emotions.

[0324] (Server) Save the generated minutes in the database and prepare to notify the user.

[0325] Step 8:

[0326] (User) When you want to look back on a past conversation, you can use a dedicated app to enter search keywords and date and time.

[0327] (Terminal) The terminal that receives the user's search request sends it to the central server.

[0328] Step 9:

[0329] (Server) Search for minutes based on specified keywords in the database.

[0330] (Server) Search results also include related emotional information.

[0331] Step 10:

[0332] (Server) If the search results are incomplete, prepare to fill in the missing parts using generative AI.

[0333] Step 11:

[0334] (Server) The generative AI predicts the missing parts and generates new minutes based on the supplemented information.

[0335] (Server) Sends the completed minutes and emotion information to the user terminal.

[0336] Step 12:

[0337] (Terminal) The transmitted minutes and emotion information are displayed on the user interface.

[0338] (User) Check the completed results and sentiment information to obtain the required information.

[0339] Example 2

[0340] 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."

[0341] While conventional meeting minutes-taking systems have the ability to record and analyze the content of conversations, they are unable to reflect emotional information during the conversation, and therefore are unable to fully convey the nuances and atmosphere of the discussion.In addition, when search results are incomplete, they lack the functionality to supplement appropriate information, which can make it inconvenient for users to use past meeting minutes.

[0342] 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.

[0343] In this invention, the server includes means for recording user conversations in real time, means for analyzing and integrating emotional data into the recorded audio data in real time, means for uploading the audio files and emotional metadata to a central server, means for converting the audio files stored in the central server into text using a voice recognition system, means for analyzing the converted text using natural language processing technology and converting it into a minutes format including emotional information, means for searching a database for past minutes in response to a search request from a user, means for completing missing parts using generative AI if the search results are incomplete, and means for displaying the minutes including the completed information so that the user can review them. This makes it possible to add emotional information to the content of the conversation, creating more detailed and easy-to-understand minutes, and automatically completing missing information during searches, allowing users to efficiently utilize past minutes.

[0344] "User" refers to any person or entity that operates this system.

[0345] "Conversation" refers to the content of communication uttered by a user as voice.

[0346] "Recording in real time" refers to recording the user's speech as audio data in real time.

[0347] "Emotion data" refers to emotional information analyzed from the user's voice.

[0348] "Audio data" refers to audio files that record the user's conversation.

[0349] "Emotional metadata" refers to meta-information that includes emotional information added to audio data.

[0350] "Central Server" refers to a server for storing and processing audio data and emotional metadata.

[0351] "Speech recognition system" refers to technology or software for converting voice data into text data.

[0352] "Text conversion" refers to the analysis of audio data and conversion into text information.

[0353] "Natural language processing technology (NLP)" refers to technology for analyzing text data, understanding meaning and context, and extracting information.

[0354] "Minutes format" refers to a document format that organizes the contents of the conversation and summarizes the main points.

[0355] "Generative AI" refers to artificial intelligence that has reasoning techniques to complement incomplete data and missing information.

[0356] "Database" refers to a system for storing and managing audio data, text data, and associated metadata in a searchable manner.

[0357] A "search request" refers to a query made by a user to a system seeking specific information.

[0358] "Search Results" means the information provided in response to a search request by the system.

[0359] "Complementary information" refers to data that generative AI adds to missing or incomplete search results.

[0360] "Visually displaying" refers to displaying information on a screen in a form that can be viewed by a user.

[0361] This invention relates to a system that electronically records user conversations, analyzes and supplements the content, and automatically creates minutes. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, emotional information is added to the conversation content, generating more detailed minutes.

[0362] The user first launches the dedicated app on a device such as a smartphone and gives the command "start recording." This causes the device to launch a voice recording program and begin recording the conversation in real time using the microphone. At the same time, an emotion engine is also launched, which analyzes the emotions from the user's conversation in real time. This emotion engine may use, for example, an emotion analysis API.

[0363] During recording, the audio data is temporarily stored in a buffer memory. At the same time, emotional data analyzed by the emotion engine is also recorded. The recorded data and emotional data are converted into audio files (e.g., WAV, MP3) at regular intervals, and emotional information is added as metadata. These audio files are then uploaded to a central server using a security protocol (e.g., HTTPS).

[0364] The server receives the uploaded audio file and emotion metadata and stores them in a database. The saved audio file is converted into text data by a speech recognition system. For example, the Google Cloud Speech-to-Text API can be used as the speech recognition system.

[0365] The text data is then input into a natural language processing (NLP) system on the server for analysis. The NLP system extracts important information and key points from the text data and converts them into minutes that reflect emotional information. The generated minutes are saved in a database for users to refer to later.

[0366] Users can use a dedicated app to search for past minutes by specifying specific keywords or dates. The device that receives the user's search request sends it to a central server. The server searches the database for minutes based on the specified keywords and extracts relevant minutes.

[0367] If the search results are incomplete, the server uses a generative AI (e.g., a generative AI model) to fill in the missing parts. The generative AI infers the missing parts and generates new minutes based on the completed information. The minutes, including the completed information, are sent to the device and visually displayed to the user.

[0368] Specific examples

[0369] Example 1: Meeting recording and sentiment analysis

[0370] User: "I want to record the meeting starting at 10 o'clock today," and opens the dedicated app.

[0371] Device: The app launches and you press the record button to record the start of the meeting.

[0372] Device: The device records all conversations, and the emotion engine analyzes and records emotions from the conversation.

[0373] Terminal: Stop recording when the meeting ends and generate an audio file.

[0374] Terminal: The audio file is given metadata such as "10:00 Meeting, October 20, 2023," as well as emotional data, and uploaded to a central server.

[0375] Example 2: Search and complete past conversations

[0376] User: "I want to check what was decided at last year's project meeting," so enters the search keyword "Project 2022" into the app.

[0377] Terminal: Receives user search requests and sends them to the central server.

[0378] Server: Searches the database for relevant minutes and extracts those that match the specified keywords.

[0379] Server: If the results are incomplete, the generative AI completes the relevant information and adds details.

[0380] Terminal: The completed minutes are displayed to the user, who then confirms the contents.

[0381] This system allows users to easily obtain highly detailed and emotionally sensitive meeting minutes, leading to a deeper understanding of the meeting content and more efficient decision-making.

[0382] Prompt Sentence Examples

[0383] Please explain in detail the process by which a user records a meeting and then analyzes the audio and emotion data to generate a transcript.

[0384] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0385] Step 1:

[0386] Launch the app and start recording

[0387] (User) The user launches the dedicated app on their smartphone and commands it to "start recording." This launches the app, and the user can start recording by touch operation or voice command.

[0388] Input: "Start recording" command from user

[0389] Output: Launch an audio recording program

[0390] Step 2:

[0391] Real-time audio recording

[0392] (Device) The app will launch a voice recording program in the background and begin recording your conversation in real time using your device's microphone.

[0393] Input: Audio input from the device microphone

[0394] Output: Saving real-time audio data to a buffer

[0395] Step 3:

[0396] Emotion engine activation and analysis

[0397] (Device) The emotion engine is started at the same time as recording, and the emotion is analyzed in real time from the user's conversation. At this point, the emotion analysis API is used to generate emotion data. For example, the emotion is identified from the tone, speed, and wording of the conversation.

[0398] Input: Real-time input of audio data

[0399] Output: Real-time emotion data generation

[0400] Step 4:

[0401] Data storage and integration

[0402] (Device) During recording, the voice data is temporarily stored in buffer memory, and the analyzed emotional data is also recorded.

[0403] Input: Real-time voice data and emotion data

[0404] Output: Temporarily saved voice data and emotion data

[0405] Step 5:

[0406] File Creation and Conversion

[0407] (Device) At regular intervals (for example, every 5 minutes), the recorded data is converted into an audio file format such as WAV or MP3, and emotional information is added as metadata. During this process, the audio data is encoded and emotional information is added as metadata.

[0408] Input: Temporarily saved voice data and emotion data

[0409] Output: Audio file (WAV, MP3 format) and emotional metadata

[0410] Step 6:

[0411] Uploading data

[0412] (Device) Upload the audio file and emotional metadata to a central server using a secure protocol such as HTTPS.

[0413] Input: Audio file and emotional metadata

[0414] Output: Data transfer to the server

[0415] Step 7:

[0416] Receiving and saving audio files

[0417] (Server) The server receives the audio file and emotional metadata uploaded from the device and stores them in a database.

[0418] Input: Uploaded audio file and sentiment metadata

[0419] Output: Save to database

[0420] Step 8:

[0421] Executing voice recognition

[0422] (Server) The server passes the saved audio file to a speech recognition system and converts it into text data, for example, using the Google Cloud Speech-to-Text API.

[0423] Input: Saved audio file

[0424] Output: Text data

[0425] Step 9:

[0426] Text data analysis

[0427] (Server) The data converted to text by speech recognition is input into a natural language processing (NLP) system for analysis. The NLP system extracts important information and key points from the text data.

[0428] Input: Text data

[0429] Output: Key information and talking points

[0430] Step 10:

[0431] Generate meeting minutes

[0432] (Server) Based on the extracted information, the minutes are converted into a format that reflects emotional information. The minutes are saved in a structured document format.

[0433] Input: Key information, conversation highlights, sentiment data

[0434] Output: Minutes document

[0435] Step 11:

[0436] Submitting a search request

[0437] (User) The user searches for past minutes by entering specific keywords or dates and times in a dedicated app.

[0438] Input: Search keyword, date and time

[0439] Output: Generate a search request

[0440] Step 12:

[0441] Submitting a search request

[0442] (Terminal) The terminal that receives the user's search request sends it to the central server.

[0443] Input: Search request

[0444] Output: Forwarding the request to the server

[0445] Step 13:

[0446] Database search

[0447] (Server) The server searches the minutes based on the specified keywords in the database and extracts relevant minutes.

[0448] Input: Search request, database

[0449] Output: Search results

[0450] Step 14:

[0451] Complementation by generative AI

[0452] (Server) If the search results are incomplete, the missing parts are filled in using a generative AI (e.g., a generative AI model). The generative AI infers the missing parts and generates complementary information.

[0453] Input: Incomplete search results

[0454] Output: Completed minutes information

[0455] Step 15:

[0456] Displaying search results

[0457] (Terminal) The minutes containing the supplemented information are presented to the user and displayed visually, allowing the user to confirm the necessary information.

[0458] Input: Completed minutes information

[0459] Output: Displaying information to the user

[0460] The above is a detailed description of the specific processing steps of the system.

[0461] (Application example 2)

[0462] 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."

[0463] In recent years, with the increasing popularity of self-driving vehicles, there is a growing need to record and analyze conversations and instructions given while driving. However, there is no system that can record and search detailed minutes that reflect the content and emotions of conversations, rather than simply recording audio. There is also a need for a system that can extract important information from conversations while driving and make it available for later review, thereby supporting driver decision-making.

[0464] The identification process by the identification 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 recording user conversations in real time; means for uploading recorded audio files to a central server; means for converting audio files stored in the central server into text using a voice recognition system; means for analyzing the converted text using natural language processing technology and converting it into a minutes format; means for searching past minutes from a database in response to a search request from a user; means for complementing missing parts using generative AI if the search results are incomplete; means for displaying the minutes including the complemented information so that the user can check them; means for recording conversations while driving using an on-board computer; and means for using an emotion analysis engine to analyze emotions from recorded audio data. This allows for detailed recording of conversations while driving and reflecting emotional information, thereby supporting the driver's decision-making and making it possible to easily search for and check important information later.

[0465] "Means for recording user conversations in real time" refers to devices or software that instantly record the user's conversations as audio on the spot.

[0466] "Means for uploading recorded audio files to a central server" refers to devices or software that transfer locally recorded audio data over a network to a remote server for storage.

[0467] "Means for converting audio files stored on a central server into text using a voice recognition system" refers to a system or software for analyzing audio data stored on a remote server and converting it into text data.

[0468] "Means of analyzing text data using natural language processing technology and converting it into minutes format" refers to devices or software that use natural language processing technology to analyze text data, extract the key points of the conversation, and format it into minutes.

[0469] "Means for searching a database for past minutes in response to a search request from a user" refers to a system or software for searching a database for minutes generated in the past based on keywords or conditions specified by the user.

[0470] "Means of using generative AI to fill in gaps when search results are incomplete" refers to devices or software that use generative AI technology to fill in gaps when search results do not contain sufficient information.

[0471] "Means for displaying the minutes including the completed information so that the user can check them" refers to a device or interface for displaying the minutes with the missing information completed so that the user can visually check them.

[0472] "Means for recording conversations while driving using an on-board computer" refers to devices or software that use a computer system installed in a vehicle to record conversations while driving.

[0473] "Means using an emotion analysis engine that analyzes emotions from recorded voice data" refers to an engine or software that analyzes recorded voice data and identifies and classifies emotions in conversations.

[0474] The present invention is a system that records user conversations in real time and analyzes and processes the recorded data. In particular, it records conversations while driving an autonomous vehicle and generates minutes that incorporate emotion analysis. Specific embodiments of the present invention are described below.

[0475] System Overview

[0476] The system consists of the following main elements:

[0477] A means for users to record conversations in real time

[0478] A means to upload recorded audio files to a central server

[0479] A means of converting audio files stored on a central server into text using a voice recognition system

[0480] A method for analyzing text data using natural language processing technology and converting it into minutes format

[0481] A means of searching past minutes from a database in response to a search request from a user

[0482] A means of using generative AI to fill in gaps in incomplete search results

[0483] A means of displaying the minutes with the completed information so that the user can review them

[0484] Means for recording conversations while driving using an on-board computer

[0485] A method using an emotion analysis engine to analyze emotions from recorded voice data

[0486] Hardware and software used

[0487] The following hardware and software is used:

[0488] Hardware:

[0489] Microphones in cars

[0490] In-vehicle computer

[0491] Internet connection device

[0492] software:

[0493] Audio recording software (e.g. PyAudio)

[0494] Sentiment analysis engine (e.g., EmotionEngine)

[0495] Speech recognition systems (e.g., Google Cloud Speech-to-Text)

[0496] Natural language processing systems (e.g., OpenAI's GPT-4)

[0497] Database management systems (e.g., MySQL, MongoDB)

[0498] Program processing overview

[0499] The overall system process is as follows:

[0500] 1. Real-time recording:

[0501] The terminal (on-board computer) uses the microphone inside the car to record the user's conversation in real time. The recorded voice is temporarily stored in a buffer, and the emotion analysis engine analyzes and records the emotional data in parallel.

[0502] 2. Generate and upload the audio file:

[0503] At regular intervals, the audio data in the buffer is converted to WAV or MP3 format, and an audio file with emotional information added as metadata is generated, which is then uploaded to a central server via a secure protocol (e.g., HTTPS).

[0504] 3. Speech Recognition and Text Transcription:

[0505] The central server inputs the received voice file into a voice recognition system and converts it into text data, which is then stored in a database.

[0506] 4. Natural Language Processing and Meeting Transcription Generation:

[0507] The text data is passed to a natural language processing system, which extracts and analyzes the main points of the conversation, generating minutes with emotional information added and saving them in a database.

[0508] 5. Search and complete:

[0509] When a user searches for past minutes, the system extracts relevant information from the database and displays it to the user. If there are any missing parts, it uses generative AI to fill in the gaps and provides that information to the user.

[0510] Specific examples

[0511] For example, when a user searches for minutes using keywords such as "Project Meeting October 20, 2023," the system will search for relevant minutes from the database and fill in any missing parts using generative AI.

[0512] Prompt Sentence Examples

[0513] "Search for minutes related to 'Project Meeting October 20, 2023' in your database. If there are any gaps, fill them in using your generative AI model."

[0514] The above is a specific embodiment of the present invention. By using this system, users can record conversations while driving in detail, generate minutes that include emotion analysis information, and easily search and check them later.

[0515] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0516] Step 1:

[0517] The terminal (on-board computer) uses the vehicle's microphone to record the user's conversation in real time. The input is the user's voice, and the output is the voice data. The voice recording software is launched and the recorded voice data is temporarily stored in a buffer. In parallel, the emotion analysis engine receives the user's voice data, analyzes and records the emotional data. The output of the emotion analysis engine is emotional information data.

[0518] Step 2:

[0519] The device periodically converts the audio data in the buffer into WAV or MP3 format. The input is the audio data stored in the buffer, and the output is an audio file (WAV or MP3). After the conversion is complete, emotional information is added as metadata and saved as tag information in the audio file.

[0520] Step 3:

[0521] The device uploads the audio file to the central server using the HTTPS protocol. The input is the audio file and its metadata, and the output is the audio file stored on the central server. The device verifies the secure connection and sends the audio file. The central server stores the received audio file in a database.

[0522] Step 4:

[0523] The server inputs the saved voice file into a voice recognition system and converts it into text data. The input is a voice file and the output is text data. The voice recognition system converts the voice data into text and records the text in a database.

[0524] Step 5:

[0525] The server passes the converted data to a natural language processing system for analysis. The input is text data, and the output is summarized data in the form of minutes. The natural language processing system analyzes the text data, extracts important content from the conversation, and converts it into minutes format. Emotional information is also reflected here.

[0526] Step 6:

[0527] The user sends a request to search for past meeting minutes from the terminal. The input is the user's search keywords and conditions, and the output is a search request. The terminal sends the request to the server, which searches the database for past records.

[0528] Step 7:

[0529] The server extracts minutes from the database that match the search criteria. The input is the user's search request, and the output is the corresponding minutes data. If the search results are incomplete, generative AI is used to fill in the missing parts and generate the final minutes data.

[0530] Step 8:

[0531] The server sends the minutes including the completed information to the terminal. The input is the complete minutes data, and the output is the data for display. The terminal visually displays the received minutes to the user, allowing them to check them.

[0532] 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.

[0533] 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.

[0534] 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.

[0535] [Second embodiment]

[0536] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0537] 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.

[0538] 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).

[0539] 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.

[0540] 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.

[0541] 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).

[0542] 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.

[0543] 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.

[0544] 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.

[0545] 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.

[0546] 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.

[0547] 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."

[0548] The present invention relates to a system for electronically recording user conversations, analyzing and supplementing the content of the recorded conversations, and automatically generating minutes of the conversations. Hereinafter, an embodiment of the present invention will be described in detail.

[0549] System Overview

[0550] 1. Transcript of conversation:

[0551] (User) The user launches the dedicated app and gives the command to "start recording."

[0552] (Device) The device uses a microphone to record the user's conversation in real time and stores the recorded data in temporary storage.

[0553] 2. Uploading audio data:

[0554] (Device) When recording is complete, an audio file is generated and metadata such as the date, time, location, and participants is added.

[0555] (Device) Uploads the recording file to a central server in accordance with security protocols.

[0556] 3. Central server for data storage and speech recognition:

[0557] (Server) Receives the uploaded audio file and stores it in the database.

[0558] (Server) The saved audio file is passed to a speech recognition system and converted into text data.

[0559] 4. Generating meeting minutes using natural language processing:

[0560] (Server) The textual data is input into a natural language processing (NLP) system and analysis is performed.

[0561] (Server) The NLP system extracts important information and conversational highlights from the text data and converts it into minutes format.

[0562] (Server) The generated minutes are saved in a database, and users can refer to them later.

[0563] 5. Search and memory completion functions:

[0564] (User) Using a dedicated app, users can search for past minutes by specifying specific keywords or dates and times.

[0565] (Terminal) The terminal that receives the user's search request sends it to the central server.

[0566] (Server) The server performs a search based on the specified keywords in the database and extracts relevant minutes.

[0567] (Server) If the search results are incomplete, generative AI is used to infer and complete the missing parts.

[0568] (Terminal) The minutes including the completed information are presented to the user and displayed visually.

[0569] Specific examples

[0570] Example 1: Meeting recording

[0571] (User) "I'll record the meeting starting at 10 o'clock today," and opens the dedicated app.

[0572] (Device) The app launches and you press the record button to record the start of the meeting.

[0573] (Device) The device records all conversations, stops recording when the meeting ends, and generates an audio file.

[0574] (Device) Add metadata such as "10:00 Meeting October 20, 2023" to the audio file and upload it to the central server.

[0575] Example 2: Search and complete past conversations

[0576] (User) "I want to check what was decided at last year's project meeting," so he enters the search keyword "Project 2022" into the app.

[0577] (Terminal) Receives the user's search request and sends it to the central server.

[0578] (Server) Search the database for relevant minutes and extract those that match the specified keywords.

[0579] (Server) If the results are incomplete, the generative AI will complete the relevant information and add details.

[0580] (Terminal) The completed minutes are displayed to the user, who then confirms the contents.

[0581] By going through these steps, users can automatically record, complete, and search the contents of important conversations and meetings, significantly improving the efficiency of their daily work. This invention will contribute to improving productivity, especially in the business world.

[0582] The processing flow will be explained below.

[0583] Step 1:

[0584] (User) The user launches the dedicated app and gives the command to "start recording."

[0585] (Device) The app will launch a voice recording program via voice recognition or tap.

[0586] (Device) The device will begin recording the conversation in real time using its microphone.

[0587] Step 2:

[0588] (Device) During recording, audio data is temporarily stored in a buffer.

[0589] (Device) Converts the recorded data into a file format (e.g. WAV, MP3) at regular intervals.

[0590] (Device) Adds metadata (date, time, location, participants) to the generated audio file.

[0591] Step 3:

[0592] (Device) Once recording is complete, prepare the audio file for uploading to the central server.

[0593] (Device) Sends the audio file to a central server using a secure protocol (e.g., HTTPS).

[0594] (Server) Receives the uploaded audio file.

[0595] Step 4:

[0596] (Server) Save the received audio file in the database.

[0597] (Server) Prepares the saved audio file for passing to the speech recognition system.

[0598] Step 5:

[0599] (Server) The speech recognition system analyzes the audio file and converts the audio data into text data.

[0600] (Server) Obtain the converted text data and proceed to the next processing step.

[0601] Step 6:

[0602] (Server) Inputs text data into a natural language processing (NLP) system.

[0603] (Server) NLP system analyzes the text data and extracts important information (e.g., decisions, action items).

[0604] Step 7:

[0605] (Server) Convert the extracted information into minutes format.

[0606] (Server) Save the generated minutes in the database and prepare to notify the user.

[0607] Step 8:

[0608] (User) When you want to look back on a past conversation, you can enter search keywords using a dedicated app.

[0609] (Terminal) Receives the user's search request and sends it to the central server.

[0610] Step 9:

[0611] (Server) Search for minutes based on specified keywords in the database.

[0612] (Server) If the search results are incomplete, prepare to fill in the missing parts using generative AI.

[0613] Step 10:

[0614] (Server) The generative AI predicts the missing parts and generates new minutes based on the supplemented information.

[0615] (Server) Sends the completed minutes to the user's terminal.

[0616] Step 11:

[0617] (Terminal) The minutes sent are displayed on the user interface.

[0618] (User) Check the completed results and obtain the required information.

[0619] Example 1

[0620] 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."

[0621] In today's business environment, many meetings and discussions are held, and it is necessary to record their contents and organize them for later reference. However, manual recording and minutes creation is time-consuming and prone to human error. Furthermore, it is difficult to efficiently search and reference past minutes, resulting in reduced work efficiency. It is also difficult to supplement information based on incomplete records. To solve these problems, a system is needed that can automatically create minutes using real-time conversation recording, speech recognition, and natural language processing, as well as highly accurate search and completion functions.

[0622] 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.

[0623] In this invention, the server includes a means for recording user conversations in real time, a means for adding metadata such as date, time, location, and participants to the recorded audio files and uploading them to a central server, and a means for sequentially converting the audio files stored on the central server into text using voice recognition technology. This allows users to efficiently record and manage conversation content and automatically generate important conversation content in the form of minutes. Users can also search past minutes based on specific keywords or dates and fill in missing parts using generative artificial intelligence, allowing them to quickly and accurately refer to the information they need.

[0624] "User" refers to a person who uses the system to record, upload, search, and check minutes of meetings.

[0625] A "terminal" is a device operated by a user, and is an electronic device that has a microphone, speaker, display, and Internet connection capabilities.

[0626] "Server" refers to a central computer system that stores audio files, processes speech recognition, natural language processing, generates and stores minutes, and performs search and completion processing.

[0627] "Conversation recording" refers to the act of collecting a user's speech as digital audio data using a microphone.

[0628] "Central Server" means a computer system located at the center of the system, which stores, analyzes, generates, searches, and completes voice data.

[0629] "Speech recognition technology" is a technology for converting voice data into text data, and is the process of analyzing voice signals and extracting their contents as text information.

[0630] "Natural language processing technology" is a technology that analyzes text data and performs processes such as semantic understanding, information extraction, and summary generation.

[0631] A "minutes format" is a document that organizes and records the contents of a conversation or meeting, and includes the date, time, location, participants, and what was said.

[0632] A "search request" is a request made by a user to locate past minutes based on specific keywords or dates.

[0633] "Data storage" refers to a memory area on a server for storing voice data, text data, minutes data, etc.

[0634] "Generative AI" is an AI technology that complements and generates new data based on existing information.

[0635] "Metadata" is auxiliary information added to an audio file, and includes attribute information such as date, time, location, and participants.

[0636] A "user interface" is an interface that allows users to operate the system, and provides functions such as voice recording, search, and checking minutes.

[0637] The present invention relates to a system for electronically recording user conversations, analyzing and supplementing the content of the recorded conversations, and automatically generating minutes of the conversations. Hereinafter, an embodiment of the present invention will be described in detail.

[0638] Conversation Record

[0639] (User) The user launches the dedicated app and instructs it to "start recording." Specifically, the user presses the "start recording button" using the smartphone app, which starts the recording process.

[0640] (Device) The device (such as a smartphone or tablet) uses a microphone to record the user's conversation in real time and saves the recorded data in temporary storage.

[0641] Uploading audio data

[0642] (Device) When the recording is complete, an audio file is generated and metadata such as the date, time, location, and participants is added. For example, metadata is added in the format "Meeting held at 10:00 on October 20, 2023."

[0643] (Device) Uploads the recording file to a central server following a security protocol. Specifically, the device sends the audio file to the server using an HTTPS connection.

[0644] Central server for data storage and voice recognition

[0645] (Server) Receives the uploaded audio files and stores them in a database. The server database has a table called "received_audio_files" that stores audio files and metadata.

[0646] (Server) The saved audio file is converted into text data using speech recognition technology (e.g., Google Cloud Speech-to-Text API). For example, an audio file of "10 o'clock meeting" is converted into text such as "The agenda for the meeting is budget review."

[0647] Generating meeting minutes using natural language processing

[0648] (Server) The text data is input into natural language processing technology (e.g., Hugging Face's Transformers library) and analyzed. The text data is analyzed into a format such as "Important matters: budget, person in charge: Tanaka."

[0649] (Server) Extracts important information and key points from the text data and converts them into minutes format. For example, minutes may be converted into a format such as "Meeting name: 10:00 Meeting, Date and time: October 20, 2023, Content: Budget review will be on the agenda."

[0650] (Server) The generated minutes are saved in the database. The minutes data is stored in a table called "meeting_minutes".

[0651] Search and memory completion functions

[0652] (User) Search for past minutes by specifying specific keywords or dates using the dedicated app. For example, enter "Project 2022" in the app and press the search button.

[0653] (Device) The device receives the user's search request and sends it to the central server. The search query is sent to the server's API.

[0654] (Server) Perform a search based on the specified keywords in the database to extract relevant minutes, for example, filtering minutes that match the keywords "project" and "2022".

[0655] (Server) If the search results are incomplete, generative artificial intelligence (e.g., OpenAI GPT-4) is used to infer and complete the missing parts. For example, GPT-4 completes the missing part about "project progress" and adds more detailed information.

[0656] (Device) The minutes containing the supplemented information are presented to the user, who confirms the contents. The supplemented minutes are displayed on the app screen.

[0657] Specific examples

[0658] Example 1: Meeting recording

[0659] (User) Opens the dedicated app and instructs, "I want to record the meeting starting at 10:00 today."

[0660] (Device) The app launches and you press the record button to record the start of the meeting.

[0661] (Device) The device records all conversations, stops recording when the meeting ends, and generates an audio file.

[0662] (Device) Add metadata such as "10:00 Meeting October 20, 2023" to the audio file and upload it to the central server.

[0663] Example 2: Search and complete past conversations

[0664] (User) In the app, he / she thinks, "I want to check what was decided at last year's project meeting," and enters the search keyword "Project 2022."

[0665] (Terminal) Receives the user's search request and sends it to the central server.

[0666] (Server) Search the database for relevant minutes and extract those that match the specified keywords.

[0667] (Server) If the results are incomplete, generative artificial intelligence (e.g., OpenAI GPT-4) will complete the relevant information and add details.

[0668] (Terminal) The completed minutes are displayed to the user, who then confirms the contents.

[0669] As described above, the present invention is a system that efficiently records user conversations and automatically converts important content into minutes format, thereby improving business efficiency.

[0670] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0671] Step 1:

[0672] Recording conversations

[0673] (User) Launches the dedicated app and instructs it to "start recording." This causes the device to receive as input the time the user started recording and the corresponding action.

[0674] (Device) Records user conversation in real time using a microphone. Audio data is collected sequentially and saved in temporary storage. An audio data file is generated as output.

[0675] Step 2:

[0676] Uploading audio data

[0677] (Device) When recording is complete, an audio file is generated and metadata such as the date, time, location, and participants are added. The command to stop recording and the metadata are given as input. This saves the audio file and metadata together.

[0678] (Device) Uploads the audio file to the central server using the HTTPS security protocol. The output is the audio file and metadata sent to the central server.

[0679] Step 3:

[0680] Central server for data storage and voice recognition

[0681] (Server) Receives uploaded audio files and stores them in a database. Receives audio files and metadata sent from the device as input. Stores audio files and metadata in a database as output.

[0682] (Server) The saved audio file is passed to speech recognition technology and converted into text data. Specifically, the Google Cloud Speech-to-Text API is used. The audio file is given as input, and the audio content is converted into text data as output.

[0683] Step 4:

[0684] Generating meeting minutes using natural language processing

[0685] (Server) The text data is input into natural language processing technology (Hugging Face's Transformers library) and analyzed. Text data is given as input, and text data with key points extracted is obtained as output.

[0686] (Server) The NLP system converts important information and conversational points from text data into a transcript format. The parsed text data is given as input, and formatted transcripts are generated as output.

[0687] (Server) The generated minutes are saved in a database. The minutes data is given as input, and saved in a database as output.

[0688] Step 5:

[0689] Search and memory completion functions

[0690] (User) uses a dedicated app to input specific keywords and dates to search for past minutes. The search keywords and dates are given as input, and the device sends a search request to the central server.

[0691] (Terminal) Receives the user's search request and sends it to the central server. The user's search keywords and date and time are given as input, and the search request is sent to the server as output.

[0692] (Server) Performs a search based on the specified keywords in the database and extracts relevant minutes. The search query is given as input, and the relevant minutes are extracted as output.

[0693] (Server) If the search results are incomplete, generative artificial intelligence (OpenAI GPT-4) is used to infer and complete the missing parts. The extracted incomplete minutes data is given as input, and the completed minutes are obtained as output.

[0694] (Terminal) Present the minutes containing the completed information to the user. The completed minutes data is given as input, and is displayed on the user's terminal as output.

[0695] As described above, the specific actions, data processing, and data calculations performed at each step form a series of flows, which realize efficient recording of conversations, generation of minutes, and search and completion functions.

[0696] (Application example 1)

[0697] 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."

[0698] Conversations held inside autonomous vehicles are likely to contain important information, but there is no appropriate means to efficiently record the content and use it as minutes. There is also a need for a simple and effective process for uploading and searching audio files, as well as for completing incomplete minutes. This invention solves these problems and provides a system for efficiently recording, searching, and completing conversations inside autonomous vehicles and using them as minutes.

[0699] 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.

[0700] In this invention, the server includes means for recording user conversations in real time, means for uploading recorded audio files to a central server, means for converting the audio files stored in the central server into text using a voice recognition system, means for analyzing the converted text using natural language processing technology and converting it into minutes format, means for searching past minutes from a database in response to a search request from a user, means for completing missing parts using generative AI when the search results are incomplete, means for displaying the minutes including the completed information so that the user can check them, means for recording conversations inside an autonomous vehicle, adding metadata to the recorded conversations, means for creating minutes including specific dates, times, and locations based on the metadata, and means for processing search requests based on the vehicle's location and ID information. This makes it possible to efficiently record conversations inside an autonomous vehicle and increase the availability of minutes.

[0701] "Recording a user's conversation in real time" means recording the user's voice immediately on the spot using an electronic device.

[0702] "Uploading the recorded audio file to the central server" means transferring the audio data stored in the device to a central storage device via a network.

[0703] "Converting voice files stored on a central server into text using a voice recognition system" means converting voice data stored in a central storage device into text data using voice recognition technology.

[0704] "Analyzing the text data using natural language processing technology and converting it into minutes format" means analyzing the data converted from audio to text using natural language processing technology, extracting summaries and important points, and formatting it into an official record format.

[0705] "Searching for past minutes from a database in response to a search request from a user" means receiving a search request from a user based on specific conditions and searching for past records in a database in accordance with that request.

[0706] "Using generative AI to fill in missing parts when search results are incomplete" means that when the results obtained through a search are incomplete, generative AI is used to supplement and add missing information.

[0707] "Displaying the minutes including the supplemented information so that the user can check them" means visually presenting the final minutes to the user, including the information supplemented by the generative AI.

[0708] "Recording conversations inside an autonomous vehicle and uploading them with metadata" means recording voice conversations inside an autonomous vehicle and sending the audio files to a central server along with additional information such as date, time, and location.

[0709] "Creating minutes with specific dates, times, and locations based on metadata" means using additional information (metadata) to generate an official record based on the content of conversations that took place at specific times and locations.

[0710] "Processing search requests based on vehicle location and ID information" means accepting search requests from users based on the vehicle's current location and identification information, and locating and providing appropriate records.

[0711] The present invention relates to a system for efficiently recording conversations in an autonomous vehicle and using the recorded conversations as minutes. Detailed embodiments of the present invention will be described below.

[0712] System configuration

[0713] User device: A smartphone or in-car smart device used to start and stop recording conversations and upload audio data to a central server.

[0714] Central server: Responsible for storing recording data, speech recognition, natural language processing, and generating meeting transcripts.

[0715] Database: A storage system for storing audio files, text data, and generated transcripts.

[0716] Generative AI: Used to complete incomplete search results and generate detailed meeting transcripts.

[0717] Program processing

[0718] Record and upload conversations

[0719] 1. The user device launches the app inside the autonomous vehicle and selects "start recording."

[0720] 2. The device uses the vehicle's microphone to record the user's conversation in real time and temporarily stores the audio data.

[0721] 3. When the recording is complete, an audio file is generated and metadata such as the date and time, vehicle ID, and location information is added to the file. This metadata specifically includes GPS information and the vehicle's identification ID.

[0722] 4. The device uploads the audio file to a central server following security protocols.

[0723] Audio data processing

[0724] 5. The central server receives the uploaded audio files and stores them in a database.

[0725] 6. The server converts the voice data into text using a speech recognition system (e.g., Google Cloud Speech-To-Text API).

[0726] 7. The text data is fed into a natural language processing system based on Hugging Face's transformers library, which extracts and summarizes important information and key points of the conversation.

[0727] 8. The generated minutes are stored in a database for users to refer to later.

[0728] Search and complete meeting minutes

[0729] 9. The user uses the app to search for minutes by specifying specific keywords, date and time, and vehicle location information.

[0730] 10. The central server uses Elasticsearch to search for the relevant minutes in the database and provide the results to the user.

[0731] 11. If the search results are incomplete, generative AI (e.g., GPT-3) will fill in the missing parts and generate a completed transcript.

[0732] 12. The completed minutes will be displayed on the terminal for visual confirmation.

[0733] Specific examples

[0734] Meeting minutes

[0735] User: When starting a sales strategy meeting in an autonomous vehicle, the user enters into the app, "I will record today's sales strategy meeting. It will start at 10:00 and will be held at the Tokyo branch."

[0736] Server: Records conversations during meetings, adds metadata, and uploads them to a central server.

[0737] Server: The uploaded audio data is converted into text and minutes format, and stored in a database.

[0738] Search past minutes

[0739] User: Type into the app, "Find minutes from last December's marketing meeting."

[0740] Server: Searches the database using Elasticsearch and displays the relevant minutes on the user's terminal.

[0741] Server: If the search results are incomplete, the generative AI fills in the gaps and provides information.

[0742] This makes it possible to efficiently record conversations and discussions inside an autonomous vehicle and easily search for and supplement the necessary information.

[0743] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0744] Step 1:

[0745] The user launches the dedicated app in the autonomous vehicle and selects "start recording." This starts the recording function.

[0746] Step 2:

[0747] The device uses the vehicle's microphone to record the user's conversation in real time. The recorded data is temporarily stored in the device's storage. The input is the user's voice data, and the output is a temporary audio file.

[0748] Step 3:

[0749] When the user selects "stop recording," the device stops recording and generates an audio file. This audio file is assigned metadata such as the date and time, vehicle ID, and location information. The input is a temporary audio file and metadata, and the output is an audio file with metadata.

[0750] Step 4:

[0751] The device uploads the generated audio file to a central server according to a security protocol. The input is the audio file with metadata, and the output is the saved audio file on the central server.

[0752] Step 5:

[0753] The server receives the uploaded audio files and stores them in a database. The input is the audio file, and the output is the audio data stored in the database.

[0754] Step 6:

[0755] The server passes the audio file to a speech recognition system and converts it into text data, for example, using the Google Cloud Speech-To-Text API. The input is audio data, and the output is text data.

[0756] Step 7:

[0757] Text data is input into a natural language processing system to extract and analyze important information and key points of the conversation. This analysis uses the Hugging Face transformers library. The input is text data, and the output is summary data in the form of meeting minutes.

[0758] Step 8:

[0759] The server saves the generated minutes in a database so that users can refer to them later. The input is summary data in minutes format, and the output is minutes saved in the database.

[0760] Step 9:

[0761] The user searches for meeting minutes using a dedicated app by specifying specific keywords, date and time, and vehicle location information. This operation sends a search request to the server. The input is a search query, and the output is a search request.

[0762] Step 10:

[0763] The server uses Elasticsearch to search for the relevant minutes in the database and provide the results to the user. The input is the search request and the output is the search results.

[0764] Step 11:

[0765] If the search results are incomplete, we use a generative AI (e.g., GPT-3) to fill in the gaps and generate a complete transcript. The input is the incomplete search results, and the output is the completed transcript.

[0766] Step 12:

[0767] The completed minutes are displayed on the user's terminal, allowing the user to visually check them. The input is the completed minutes, and the output is the minutes displayed on the user's terminal.

[0768] 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.

[0769] The present invention relates to a system that electronically records user conversations, analyzes and supplements the content, and automatically formats it into minutes. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, emotional information is added to the conversation content, generating more detailed minutes.

[0770] System Overview

[0771] 1. Transcript of conversation:

[0772] (User) The user launches the dedicated app and gives the command to "start recording."

[0773] (Terminal) The terminal launches a voice recording program and begins recording the conversation in real time using the microphone.

[0774] (Device) At the same time, the emotion engine is also activated, analyzing emotions from the user's conversation in real time.

[0775] 2. Integration and upload of voice and emotion data:

[0776] (Device) During recording, audio data is temporarily stored in a buffer.

[0777] (Device) Along with the recorded data, emotion data recognized by the emotion engine is also recorded.

[0778] (Device) Convert the recorded data into a file format (e.g. WAV, MP3) at regular intervals and add emotional information as metadata.

[0779] (Device) Upload the audio file to a central server using a secure protocol (e.g., HTTPS).

[0780] 3. Central server for data storage and speech recognition:

[0781] (Server) Receives the uploaded audio file and emotion metadata and stores them in a database.

[0782] (Server) The saved audio file is passed to a speech recognition system and converted into text data.

[0783] 4. Generating meeting minutes using natural language processing:

[0784] (Server) The textual data is input into a natural language processing (NLP) system and analysis is performed.

[0785] (Server) The NLP system extracts important information and key points of the conversation from the text data and converts them into minutes format that reflects emotional information.

[0786] (Server) The generated minutes are saved in a database, and users can refer to them later.

[0787] 5. Search and memory completion functions:

[0788] (User) Using a dedicated app, users can search for past minutes by specifying specific keywords or dates and times.

[0789] (Terminal) The terminal that receives the user's search request sends it to the central server.

[0790] (Server) The server searches the minutes based on the specified keywords in the database and extracts relevant minutes.

[0791] (Server) If the search results are incomplete, prepare to fill in the missing parts using generative AI.

[0792] (Server) The generative AI predicts the missing parts and generates new minutes based on the supplemented information.

[0793] (Terminal) The minutes including the completed information are presented to the user and displayed visually.

[0794] Specific examples

[0795] Example 1: Meeting recording and sentiment analysis

[0796] (User) "I'll record the meeting starting at 10 o'clock today," and opens the dedicated app.

[0797] (Device) The app launches and you press the record button to record the start of the meeting.

[0798] (Device) The device records all conversations, and the emotion engine analyzes and records emotions from the conversation.

[0799] (Device) Stop recording when the meeting ends and generate an audio file.

[0800] (Device) The audio file is given metadata such as "10:00 Meeting, October 20, 2023," as well as emotional data, and then uploaded to a central server.

[0801] Example 2: Search and complete past conversations

[0802] (User) "I want to check what was decided at last year's project meeting," so he enters the search keyword "Project 2022" into the app.

[0803] (Terminal) Receives the user's search request and sends it to the central server.

[0804] (Server) Search the database for relevant minutes and extract those that match the specified keywords.

[0805] (Server) If the results are incomplete, the generative AI will complete the relevant information and add details.

[0806] (Terminal) The completed minutes are displayed to the user, who then confirms the contents.

[0807] The system allows users to obtain not just minutes, but detailed records that reflect the emotions expressed during the conversation, leading to a deeper understanding of the conversation and more reliable decision-making.

[0808] The processing flow will be explained below.

[0809] Step 1:

[0810] (User) The user launches the dedicated app and gives the command to "start recording."

[0811] (Device) The app will launch the voice recording program and emotion engine and begin recording your conversation in real time using the microphone.

[0812] (Device) The emotion engine simultaneously analyzes the user's tone of voice and vocabulary to generate emotion data.

[0813] Step 2:

[0814] (Device) During recording, the voice data and emotion data are temporarily stored in a buffer.

[0815] (Device) Convert the recorded data into a file format (e.g. WAV, MP3) at regular intervals and add emotional information as metadata.

[0816] Step 3:

[0817] (Device) Once the recording is complete, prepare the audio file and emotion data for uploading to the central server.

[0818] (Device) Sends the audio file to a central server using a secure protocol (e.g., HTTPS).

[0819] (Server) Receives the uploaded audio file and emotional metadata.

[0820] Step 4:

[0821] (Server) Store the received audio files and emotion data in a database.

[0822] (Server) The saved audio file is passed to a speech recognition system and converted into text data.

[0823] Step 5:

[0824] (Server) The speech recognition system converts the speech data into text data.

[0825] (Server) Emotion data is extracted along with the converted text data, and the process proceeds to the next step.

[0826] Step 6:

[0827] (Server) Input text data and emotion data into a natural language processing (NLP) system.

[0828] (Server) The NLP system analyzes the text data and extracts key information (e.g., decisions, action items) and sentiment changes.

[0829] Step 7:

[0830] (Server) Based on the extracted information, convert it into a minutes format that reflects changes in emotions.

[0831] (Server) Save the generated minutes in the database and prepare to notify the user.

[0832] Step 8:

[0833] (User) When you want to look back on a past conversation, you can use a dedicated app to enter search keywords and date and time.

[0834] (Terminal) The terminal that receives the user's search request sends it to the central server.

[0835] Step 9:

[0836] (Server) Search for minutes based on specified keywords in the database.

[0837] (Server) Search results also include related emotional information.

[0838] Step 10:

[0839] (Server) If the search results are incomplete, prepare to fill in the missing parts using generative AI.

[0840] Step 11:

[0841] (Server) The generative AI predicts the missing parts and generates new minutes based on the supplemented information.

[0842] (Server) Sends the completed minutes and emotion information to the user terminal.

[0843] Step 12:

[0844] (Terminal) The transmitted minutes and emotion information are displayed on the user interface.

[0845] (User) Check the completed results and sentiment information to obtain the required information.

[0846] Example 2

[0847] 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."

[0848] While conventional meeting minutes-taking systems have the ability to record and analyze the content of conversations, they are unable to reflect emotional information during the conversation, and therefore are unable to fully convey the nuances and atmosphere of the discussion.In addition, when search results are incomplete, they lack the functionality to supplement appropriate information, which can make it inconvenient for users to use past meeting minutes.

[0849] 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.

[0850] In this invention, the server includes means for recording user conversations in real time, means for analyzing and integrating emotional data into the recorded audio data in real time, means for uploading the audio files and emotional metadata to a central server, means for converting the audio files stored in the central server into text using a voice recognition system, means for analyzing the converted text using natural language processing technology and converting it into a minutes format including emotional information, means for searching a database for past minutes in response to a search request from a user, means for completing missing parts using generative AI if the search results are incomplete, and means for displaying the minutes including the completed information so that the user can review them. This makes it possible to add emotional information to the content of the conversation, creating more detailed and easy-to-understand minutes, and automatically completing missing information during searches, allowing users to efficiently utilize past minutes.

[0851] "User" refers to any person or entity that operates this system.

[0852] "Conversation" refers to the content of communication uttered by a user as voice.

[0853] "Recording in real time" refers to recording the user's speech as audio data in real time.

[0854] "Emotion data" refers to emotional information analyzed from the user's voice.

[0855] "Audio data" refers to audio files that record the user's conversation.

[0856] "Emotional metadata" refers to meta-information that includes emotional information added to audio data.

[0857] "Central Server" refers to a server for storing and processing audio data and emotional metadata.

[0858] "Speech recognition system" refers to technology or software for converting voice data into text data.

[0859] "Text conversion" refers to the analysis of audio data and conversion into text information.

[0860] "Natural language processing technology (NLP)" refers to technology for analyzing text data, understanding meaning and context, and extracting information.

[0861] "Minutes format" refers to a document format that organizes the contents of the conversation and summarizes the main points.

[0862] "Generative AI" refers to artificial intelligence that has reasoning techniques to complement incomplete data and missing information.

[0863] "Database" refers to a system for storing and managing audio data, text data, and associated metadata in a searchable manner.

[0864] A "search request" refers to a query made by a user to a system seeking specific information.

[0865] "Search Results" means the information provided in response to a search request by the system.

[0866] "Complementary information" refers to data that generative AI adds to missing or incomplete search results.

[0867] "Visually displaying" refers to displaying information on a screen in a form that can be viewed by a user.

[0868] This invention relates to a system that electronically records user conversations, analyzes and supplements the content, and automatically creates minutes. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, emotional information is added to the conversation content, generating more detailed minutes.

[0869] The user first launches the dedicated app on a device such as a smartphone and gives the command "start recording." This causes the device to launch a voice recording program and begin recording the conversation in real time using the microphone. At the same time, an emotion engine is also launched, which analyzes the emotions from the user's conversation in real time. This emotion engine may use, for example, an emotion analysis API.

[0870] During recording, the audio data is temporarily stored in a buffer memory. At the same time, emotional data analyzed by the emotion engine is also recorded. The recorded data and emotional data are converted into audio files (e.g., WAV, MP3) at regular intervals, and emotional information is added as metadata. These audio files are then uploaded to a central server using a security protocol (e.g., HTTPS).

[0871] The server receives the uploaded audio file and emotion metadata and stores them in a database. The saved audio file is converted into text data by a speech recognition system. For example, the Google Cloud Speech-to-Text API can be used as the speech recognition system.

[0872] The text data is then input into a natural language processing (NLP) system on the server for analysis. The NLP system extracts important information and key points from the text data and converts them into minutes that reflect emotional information. The generated minutes are saved in a database for users to refer to later.

[0873] Users can use a dedicated app to search for past minutes by specifying specific keywords or dates. The device that receives the user's search request sends it to a central server. The server searches the database for minutes based on the specified keywords and extracts relevant minutes.

[0874] If the search results are incomplete, the server uses a generative AI (e.g., a generative AI model) to fill in the missing parts. The generative AI infers the missing parts and generates new minutes based on the completed information. The minutes, including the completed information, are sent to the device and visually displayed to the user.

[0875] Specific examples

[0876] Example 1: Meeting recording and sentiment analysis

[0877] User: "I want to record the meeting starting at 10 o'clock today," and opens the dedicated app.

[0878] Device: The app launches and you press the record button to record the start of the meeting.

[0879] Device: The device records all conversations, and the emotion engine analyzes and records emotions from the conversation.

[0880] Terminal: Stop recording when the meeting ends and generate an audio file.

[0881] Terminal: The audio file is given metadata such as "10:00 Meeting, October 20, 2023," as well as emotional data, and uploaded to a central server.

[0882] Example 2: Search and complete past conversations

[0883] User: "I want to check what was decided at last year's project meeting," so enters the search keyword "Project 2022" into the app.

[0884] Terminal: Receives user search requests and sends them to the central server.

[0885] Server: Searches the database for relevant minutes and extracts those that match the specified keywords.

[0886] Server: If the results are incomplete, the generative AI completes the relevant information and adds details.

[0887] Terminal: The completed minutes are displayed to the user, who then confirms the contents.

[0888] This system allows users to easily obtain highly detailed and emotionally sensitive meeting minutes, leading to a deeper understanding of the meeting content and more efficient decision-making.

[0889] Prompt Sentence Examples

[0890] Please explain in detail the process by which a user records a meeting and then analyzes the audio and emotion data to generate a transcript.

[0891] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0892] Step 1:

[0893] Launch the app and start recording

[0894] (User) The user launches the dedicated app on their smartphone and commands it to "start recording." This launches the app, and the user can start recording by touch operation or voice command.

[0895] Input: "Start recording" command from user

[0896] Output: Launch an audio recording program

[0897] Step 2:

[0898] Real-time audio recording

[0899] (Device) The app will launch a voice recording program in the background and begin recording your conversation in real time using your device's microphone.

[0900] Input: Audio input from the device microphone

[0901] Output: Saving real-time audio data to a buffer

[0902] Step 3:

[0903] Emotion engine activation and analysis

[0904] (Device) The emotion engine is started at the same time as recording, and the emotion is analyzed in real time from the user's conversation. At this point, the emotion analysis API is used to generate emotion data. For example, the emotion is identified from the tone, speed, and wording of the conversation.

[0905] Input: Real-time input of audio data

[0906] Output: Real-time emotion data generation

[0907] Step 4:

[0908] Data storage and integration

[0909] (Device) During recording, the voice data is temporarily stored in buffer memory, and the analyzed emotional data is also recorded.

[0910] Input: Real-time voice data and emotion data

[0911] Output: Temporarily saved voice data and emotion data

[0912] Step 5:

[0913] File Creation and Conversion

[0914] (Device) At regular intervals (for example, every 5 minutes), the recorded data is converted into an audio file format such as WAV or MP3, and emotional information is added as metadata. During this process, the audio data is encoded and emotional information is added as metadata.

[0915] Input: Temporarily saved voice data and emotion data

[0916] Output: Audio file (WAV, MP3 format) and emotional metadata

[0917] Step 6:

[0918] Uploading data

[0919] (Device) Upload the audio file and emotional metadata to a central server using a secure protocol such as HTTPS.

[0920] Input: Audio file and emotional metadata

[0921] Output: Data transfer to the server

[0922] Step 7:

[0923] Receiving and saving audio files

[0924] (Server) The server receives the audio file and emotional metadata uploaded from the device and stores them in a database.

[0925] Input: Uploaded audio file and sentiment metadata

[0926] Output: Save to database

[0927] Step 8:

[0928] Executing voice recognition

[0929] (Server) The server passes the saved audio file to a speech recognition system and converts it into text data, for example, using the Google Cloud Speech-to-Text API.

[0930] Input: Saved audio file

[0931] Output: Text data

[0932] Step 9:

[0933] Text data analysis

[0934] (Server) The data converted to text by speech recognition is input into a natural language processing (NLP) system for analysis. The NLP system extracts important information and key points from the text data.

[0935] Input: Text data

[0936] Output: Key information and talking points

[0937] Step 10:

[0938] Generate meeting minutes

[0939] (Server) Based on the extracted information, the minutes are converted into a format that reflects emotional information. The minutes are saved in a structured document format.

[0940] Input: Key information, conversation highlights, sentiment data

[0941] Output: Minutes document

[0942] Step 11:

[0943] Submitting a search request

[0944] (User) The user searches for past minutes by entering specific keywords or dates and times in a dedicated app.

[0945] Input: Search keyword, date and time

[0946] Output: Generate a search request

[0947] Step 12:

[0948] Submitting a search request

[0949] (Terminal) The terminal that receives the user's search request sends it to the central server.

[0950] Input: Search request

[0951] Output: Forwarding the request to the server

[0952] Step 13:

[0953] Database search

[0954] (Server) The server searches the minutes based on the specified keywords in the database and extracts relevant minutes.

[0955] Input: Search request, database

[0956] Output: Search results

[0957] Step 14:

[0958] Complementation by generative AI

[0959] (Server) If the search results are incomplete, the missing parts are filled in using a generative AI (e.g., a generative AI model). The generative AI infers the missing parts and generates complementary information.

[0960] Input: Incomplete search results

[0961] Output: Completed minutes information

[0962] Step 15:

[0963] Displaying search results

[0964] (Terminal) The minutes containing the supplemented information are presented to the user and displayed visually, allowing the user to confirm the necessary information.

[0965] Input: Completed minutes information

[0966] Output: Displaying information to the user

[0967] The above is a detailed description of the specific processing steps of the system.

[0968] (Application example 2)

[0969] 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."

[0970] In recent years, with the increasing popularity of self-driving vehicles, there is a growing need to record and analyze conversations and instructions given while driving. However, there is no system that can record and search detailed minutes that reflect the content and emotions of conversations, rather than simply recording audio. There is also a need for a system that can extract important information from conversations while driving and make it available for later review, thereby supporting driver decision-making.

[0971] The identification process by the identification 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 recording user conversations in real time; means for uploading recorded audio files to a central server; means for converting audio files stored in the central server into text using a voice recognition system; means for analyzing the converted text using natural language processing technology and converting it into a minutes format; means for searching past minutes from a database in response to a search request from a user; means for complementing missing parts using generative AI if the search results are incomplete; means for displaying the minutes including the complemented information so that the user can check them; means for recording conversations while driving using an on-board computer; and means for using an emotion analysis engine to analyze emotions from recorded audio data. This allows for detailed recording of conversations while driving and reflecting emotional information, thereby supporting the driver's decision-making and making it possible to easily search for and check important information later.

[0972] "Means for recording user conversations in real time" refers to devices or software that instantly record the user's conversations as audio on the spot.

[0973] "Means for uploading recorded audio files to a central server" refers to devices or software that transfer locally recorded audio data over a network to a remote server for storage.

[0974] "Means for converting audio files stored on a central server into text using a voice recognition system" refers to a system or software for analyzing audio data stored on a remote server and converting it into text data.

[0975] "Means of analyzing text data using natural language processing technology and converting it into minutes format" refers to devices or software that use natural language processing technology to analyze text data, extract the key points of the conversation, and format it into minutes.

[0976] "Means for searching a database for past minutes in response to a search request from a user" refers to a system or software for searching a database for minutes generated in the past based on keywords or conditions specified by the user.

[0977] "Means of using generative AI to fill in gaps when search results are incomplete" refers to devices or software that use generative AI technology to fill in gaps when search results do not contain sufficient information.

[0978] "Means for displaying the minutes including the completed information so that the user can check them" refers to a device or interface for displaying the minutes with the missing information completed so that the user can visually check them.

[0979] "Means for recording conversations while driving using an on-board computer" refers to devices or software that use a computer system installed in a vehicle to record conversations while driving.

[0980] "Means using an emotion analysis engine that analyzes emotions from recorded voice data" refers to an engine or software that analyzes recorded voice data and identifies and classifies emotions in conversations.

[0981] The present invention is a system that records user conversations in real time and analyzes and processes the recorded data. In particular, it records conversations while driving an autonomous vehicle and generates minutes that incorporate emotion analysis. Specific embodiments of the present invention are described below.

[0982] System Overview

[0983] The system consists of the following main elements:

[0984] A means for users to record conversations in real time

[0985] A means to upload recorded audio files to a central server

[0986] A means of converting audio files stored on a central server into text using a voice recognition system

[0987] A method for analyzing text data using natural language processing technology and converting it into minutes format

[0988] A means of searching past minutes from a database in response to a search request from a user

[0989] A means of using generative AI to fill in gaps in incomplete search results

[0990] A means of displaying the minutes with the completed information so that the user can review them

[0991] Means for recording conversations while driving using an on-board computer

[0992] A method using an emotion analysis engine to analyze emotions from recorded voice data

[0993] Hardware and software used

[0994] The following hardware and software is used:

[0995] Hardware:

[0996] Microphones in cars

[0997] In-vehicle computer

[0998] Internet connection device

[0999] software:

[1000] Audio recording software (e.g. PyAudio)

[1001] Sentiment analysis engine (e.g., EmotionEngine)

[1002] Speech recognition systems (e.g., Google Cloud Speech-to-Text)

[1003] Natural language processing systems (e.g., OpenAI's GPT-4)

[1004] Database management systems (e.g. MySQL, MongoDB)

[1005] Program processing overview

[1006] The overall system process is as follows:

[1007] 1. Real-time recording:

[1008] The terminal (on-board computer) uses the microphone inside the car to record the user's conversation in real time. The recorded voice is temporarily stored in a buffer, and the emotion analysis engine analyzes and records the emotional data in parallel.

[1009] 2. Generate and upload the audio file:

[1010] At regular intervals, the audio data in the buffer is converted to WAV or MP3 format, and an audio file with emotional information added as metadata is generated, which is then uploaded to a central server via a secure protocol (e.g., HTTPS).

[1011] 3. Speech Recognition and Text Transcription:

[1012] The central server inputs the received voice file into a voice recognition system and converts it into text data, which is then stored in a database.

[1013] 4. Natural Language Processing and Meeting Transcription Generation:

[1014] The text data is passed to a natural language processing system, which extracts and analyzes the main points of the conversation, generating minutes with emotional information added and saving them in a database.

[1015] 5. Search and complete:

[1016] When a user searches for past minutes, the system extracts relevant information from the database and displays it to the user. If there are any missing parts, it uses generative AI to fill in the gaps and provides that information to the user.

[1017] Specific examples

[1018] For example, when a user searches for minutes using keywords such as "Project Meeting October 20, 2023," the system will search for relevant minutes from the database and fill in any missing parts using generative AI.

[1019] Prompt Sentence Examples

[1020] "Search for minutes related to 'Project Meeting October 20, 2023' in your database. If there are any gaps, fill them in using your generative AI model."

[1021] The above is a specific embodiment of the present invention. By using this system, users can record conversations while driving in detail, generate minutes that include emotion analysis information, and easily search and check them later.

[1022] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1023] Step 1:

[1024] The terminal (on-board computer) uses the vehicle's microphone to record the user's conversation in real time. The input is the user's voice, and the output is the voice data. The voice recording software is launched and the recorded voice data is temporarily stored in a buffer. In parallel, the emotion analysis engine receives the user's voice data, analyzes and records the emotional data. The output of the emotion analysis engine is emotional information data.

[1025] Step 2:

[1026] The device periodically converts the audio data in the buffer into WAV or MP3 format. The input is the audio data stored in the buffer, and the output is an audio file (WAV or MP3). After the conversion is complete, emotional information is added as metadata and saved as tag information in the audio file.

[1027] Step 3:

[1028] The device uploads the audio file to the central server using the HTTPS protocol. The input is the audio file and its metadata, and the output is the audio file stored on the central server. The device verifies the secure connection and sends the audio file. The central server stores the received audio file in a database.

[1029] Step 4:

[1030] The server inputs the saved voice file into a voice recognition system and converts it into text data. The input is a voice file and the output is text data. The voice recognition system converts the voice data into text and records the text in a database.

[1031] Step 5:

[1032] The server passes the converted data to a natural language processing system for analysis. The input is text data, and the output is summarized data in the form of minutes. The natural language processing system analyzes the text data, extracts important content from the conversation, and converts it into minutes format. Emotional information is also reflected here.

[1033] Step 6:

[1034] The user sends a request to search for past meeting minutes from the terminal. The input is the user's search keywords and conditions, and the output is a search request. The terminal sends the request to the server, which searches the database for past records.

[1035] Step 7:

[1036] The server extracts minutes from the database that match the search criteria. The input is the user's search request, and the output is the corresponding minutes data. If the search results are incomplete, generative AI is used to fill in the missing parts and generate the final minutes data.

[1037] Step 8:

[1038] The server sends the minutes including the completed information to the terminal. The input is the complete minutes data, and the output is the data for display. The terminal visually displays the received minutes to the user, allowing them to check them.

[1039] 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.

[1040] 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.

[1041] 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.

[1042] [Third embodiment]

[1043] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1044] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[1045] 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).

[1046] 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.

[1047] 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.

[1048] 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).

[1049] 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.

[1050] 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.

[1051] 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.

[1052] 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.

[1053] 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.

[1054] 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."

[1055] The present invention relates to a system for electronically recording user conversations, analyzing and supplementing the content of the recorded conversations, and automatically generating minutes of the conversations. Hereinafter, an embodiment of the present invention will be described in detail.

[1056] System Overview

[1057] 1. Transcript of conversation:

[1058] (User) The user launches the dedicated app and gives the command to "start recording."

[1059] (Device) The device uses a microphone to record the user's conversation in real time and stores the recorded data in temporary storage.

[1060] 2. Uploading audio data:

[1061] (Device) When recording is complete, an audio file is generated and metadata such as the date, time, location, and participants is added.

[1062] (Device) Uploads the recording file to a central server in accordance with security protocols.

[1063] 3. Central server for data storage and speech recognition:

[1064] (Server) Receives the uploaded audio file and stores it in the database.

[1065] (Server) The saved audio file is passed to a speech recognition system and converted into text data.

[1066] 4. Generating meeting minutes using natural language processing:

[1067] (Server) The textual data is input into a natural language processing (NLP) system and analysis is performed.

[1068] (Server) The NLP system extracts important information and conversational highlights from the text data and converts it into minutes format.

[1069] (Server) The generated minutes are saved in a database, and users can refer to them later.

[1070] 5. Search and memory completion functions:

[1071] (User) Using a dedicated app, users can search for past minutes by specifying specific keywords or dates and times.

[1072] (Terminal) The terminal that receives the user's search request sends it to the central server.

[1073] (Server) The server performs a search based on the specified keywords in the database and extracts relevant minutes.

[1074] (Server) If the search results are incomplete, generative AI is used to infer and complete the missing parts.

[1075] (Terminal) The minutes including the completed information are presented to the user and displayed visually.

[1076] Specific examples

[1077] Example 1: Meeting recording

[1078] (User) "I'll record the meeting starting at 10 o'clock today," and opens the dedicated app.

[1079] (Device) The app launches and you press the record button to record the start of the meeting.

[1080] (Device) The device records all conversations, stops recording when the meeting ends, and generates an audio file.

[1081] (Device) Add metadata such as "10:00 Meeting October 20, 2023" to the audio file and upload it to the central server.

[1082] Example 2: Search and complete past conversations

[1083] (User) "I want to check what was decided at last year's project meeting," so he enters the search keyword "Project 2022" into the app.

[1084] (Terminal) Receives the user's search request and sends it to the central server.

[1085] (Server) Search the database for relevant minutes and extract those that match the specified keywords.

[1086] (Server) If the results are incomplete, the generative AI will complete the relevant information and add details.

[1087] (Terminal) The completed minutes are displayed to the user, who then confirms the contents.

[1088] By going through these steps, users can automatically record, complete, and search the contents of important conversations and meetings, significantly improving the efficiency of their daily work. This invention will contribute to improving productivity, especially in the business world.

[1089] The processing flow will be explained below.

[1090] Step 1:

[1091] (User) The user launches the dedicated app and gives the command to "start recording."

[1092] (Device) The app will launch a voice recording program via voice recognition or tap.

[1093] (Device) The device will begin recording the conversation in real time using its microphone.

[1094] Step 2:

[1095] (Device) During recording, audio data is temporarily stored in a buffer.

[1096] (Device) Converts the recorded data into a file format (e.g. WAV, MP3) at regular intervals.

[1097] (Device) Adds metadata (date, time, location, participants) to the generated audio file.

[1098] Step 3:

[1099] (Device) Once recording is complete, prepare the audio file for uploading to the central server.

[1100] (Device) Sends the audio file to a central server using a secure protocol (e.g., HTTPS).

[1101] (Server) Receives the uploaded audio file.

[1102] Step 4:

[1103] (Server) Save the received audio file in the database.

[1104] (Server) Prepares the saved audio file for passing to the speech recognition system.

[1105] Step 5:

[1106] (Server) The speech recognition system analyzes the audio file and converts the audio data into text data.

[1107] (Server) Obtain the converted text data and proceed to the next processing step.

[1108] Step 6:

[1109] (Server) Inputs text data into a natural language processing (NLP) system.

[1110] (Server) NLP system analyzes the text data and extracts important information (e.g., decisions, action items).

[1111] Step 7:

[1112] (Server) Convert the extracted information into minutes format.

[1113] (Server) Save the generated minutes in the database and prepare to notify the user.

[1114] Step 8:

[1115] (User) When you want to look back on a past conversation, you can enter search keywords using a dedicated app.

[1116] (Terminal) Receives the user's search request and sends it to the central server.

[1117] Step 9:

[1118] (Server) Search for minutes based on specified keywords in the database.

[1119] (Server) If the search results are incomplete, prepare to fill in the missing parts using generative AI.

[1120] Step 10:

[1121] (Server) The generative AI predicts the missing parts and generates new minutes based on the supplemented information.

[1122] (Server) Sends the completed minutes to the user's terminal.

[1123] Step 11:

[1124] (Terminal) The minutes sent are displayed on the user interface.

[1125] (User) Check the completed results and obtain the required information.

[1126] Example 1

[1127] 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."

[1128] In today's business environment, many meetings and discussions are held, and it is necessary to record their contents and organize them for later reference. However, manual recording and minutes creation is time-consuming and prone to human error. Furthermore, it is difficult to efficiently search and reference past minutes, resulting in reduced work efficiency. It is also difficult to supplement information based on incomplete records. To solve these problems, a system is needed that can automatically create minutes using real-time conversation recording, speech recognition, and natural language processing, as well as highly accurate search and completion functions.

[1129] 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.

[1130] In this invention, the server includes a means for recording user conversations in real time, a means for adding metadata such as date, time, location, and participants to the recorded audio files and uploading them to a central server, and a means for sequentially converting the audio files stored on the central server into text using voice recognition technology. This allows users to efficiently record and manage conversation content and automatically generate important conversation content in the form of minutes. Users can also search past minutes based on specific keywords or dates and fill in missing parts using generative artificial intelligence, allowing them to quickly and accurately refer to the information they need.

[1131] "User" refers to a person who uses the system to record, upload, search, and check minutes of meetings.

[1132] A "terminal" is a device operated by a user, and is an electronic device that has a microphone, speaker, display, and Internet connection capabilities.

[1133] "Server" refers to a central computer system that stores audio files, processes speech recognition, natural language processing, generates and stores minutes, and performs search and completion processing.

[1134] "Conversation recording" refers to the act of collecting a user's speech as digital audio data using a microphone.

[1135] "Central Server" means a computer system located at the center of the system, which stores, analyzes, generates, searches, and completes voice data.

[1136] "Speech recognition technology" is a technology for converting voice data into text data, and is the process of analyzing voice signals and extracting their contents as text information.

[1137] "Natural language processing technology" is a technology that analyzes text data and performs processes such as semantic understanding, information extraction, and summary generation.

[1138] A "minutes format" is a document that organizes and records the contents of a conversation or meeting, and includes the date, time, location, participants, and what was said.

[1139] A "search request" is a request made by a user to locate past minutes based on specific keywords or dates.

[1140] "Data storage" refers to a memory area on a server for storing voice data, text data, minutes data, etc.

[1141] "Generative AI" is an AI technology that complements and generates new data based on existing information.

[1142] "Metadata" is auxiliary information added to an audio file, and includes attribute information such as date, time, location, and participants.

[1143] A "user interface" is an interface that allows users to operate the system, and provides functions such as voice recording, search, and checking minutes.

[1144] The present invention relates to a system for electronically recording user conversations, analyzing and supplementing the content of the recorded conversations, and automatically generating minutes of the conversations. Hereinafter, an embodiment of the present invention will be described in detail.

[1145] Conversation Record

[1146] (User) The user launches the dedicated app and instructs it to "start recording." Specifically, the user presses the "start recording button" using the smartphone app, which starts the recording process.

[1147] (Device) The device (such as a smartphone or tablet) uses a microphone to record the user's conversation in real time and saves the recorded data in temporary storage.

[1148] Uploading audio data

[1149] (Device) When the recording is complete, an audio file is generated and metadata such as the date, time, location, and participants is added. For example, metadata is added in the format "Meeting held at 10:00 on October 20, 2023."

[1150] (Device) Uploads the recording file to a central server following a security protocol. Specifically, the device sends the audio file to the server using an HTTPS connection.

[1151] Central server for data storage and voice recognition

[1152] (Server) Receives the uploaded audio files and stores them in a database. The server database has a table called "received_audio_files" that stores audio files and metadata.

[1153] (Server) The saved audio file is converted into text data using speech recognition technology (e.g., Google Cloud Speech-to-Text API). For example, an audio file of "10 o'clock meeting" is converted into text such as "The agenda for the meeting is budget review."

[1154] Generating meeting minutes using natural language processing

[1155] (Server) The text data is input into natural language processing technology (e.g., Hugging Face's Transformers library) and analyzed. The text data is analyzed into a format such as "Important matters: budget, person in charge: Tanaka."

[1156] (Server) Extracts important information and key points from the text data and converts them into minutes format. For example, minutes may be converted into a format such as "Meeting name: 10:00 Meeting, Date and time: October 20, 2023, Content: Budget review will be on the agenda."

[1157] (Server) The generated minutes are saved in the database. The minutes data is stored in a table called "meeting_minutes".

[1158] Search and memory completion functions

[1159] (User) Search for past minutes by specifying specific keywords or dates using the dedicated app. For example, enter "Project 2022" in the app and press the search button.

[1160] (Device) The device receives the user's search request and sends it to the central server. The search query is sent to the server's API.

[1161] (Server) Perform a search based on the specified keywords in the database to extract relevant minutes, for example, filtering minutes that match the keywords "project" and "2022".

[1162] (Server) If the search results are incomplete, generative artificial intelligence (e.g., OpenAI GPT-4) is used to infer and complete the missing parts. For example, GPT-4 completes the missing part about "project progress" and adds more detailed information.

[1163] (Device) The minutes containing the supplemented information are presented to the user, who confirms the contents. The supplemented minutes are displayed on the app screen.

[1164] Specific examples

[1165] Example 1: Meeting recording

[1166] (User) Opens the dedicated app and instructs, "I want to record the meeting starting at 10:00 today."

[1167] (Device) The app launches and you press the record button to record the start of the meeting.

[1168] (Device) The device records all conversations, stops recording when the meeting ends, and generates an audio file.

[1169] (Device) Add metadata such as "10:00 Meeting October 20, 2023" to the audio file and upload it to the central server.

[1170] Example 2: Search and complete past conversations

[1171] (User) In the app, he / she thinks, "I want to check what was decided at last year's project meeting," and enters the search keyword "Project 2022."

[1172] (Terminal) Receives the user's search request and sends it to the central server.

[1173] (Server) Search the database for relevant minutes and extract those that match the specified keywords.

[1174] (Server) If the results are incomplete, generative artificial intelligence (e.g., OpenAI GPT-4) will complete the relevant information and add details.

[1175] (Terminal) The completed minutes are displayed to the user, who then confirms the contents.

[1176] As described above, the present invention is a system that efficiently records user conversations and automatically converts important content into minutes format, thereby improving business efficiency.

[1177] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1178] Step 1:

[1179] Recording conversations

[1180] (User) Launches the dedicated app and instructs it to "start recording." This causes the device to receive as input the time the user started recording and the corresponding action.

[1181] (Device) Records user conversation in real time using a microphone. Audio data is collected sequentially and saved in temporary storage. An audio data file is generated as output.

[1182] Step 2:

[1183] Uploading audio data

[1184] (Device) When recording is complete, an audio file is generated and metadata such as the date, time, location, and participants are added. The command to stop recording and the metadata are given as input. This saves the audio file and metadata together.

[1185] (Device) Uploads the audio file to the central server using the HTTPS security protocol. The output is the audio file and metadata sent to the central server.

[1186] Step 3:

[1187] Central server for data storage and voice recognition

[1188] (Server) Receives uploaded audio files and stores them in a database. Receives audio files and metadata sent from the device as input. Stores audio files and metadata in a database as output.

[1189] (Server) The saved audio file is passed to speech recognition technology and converted into text data. Specifically, the Google Cloud Speech-to-Text API is used. The audio file is given as input, and the audio content is converted into text data as output.

[1190] Step 4:

[1191] Generating meeting minutes using natural language processing

[1192] (Server) The text data is input into natural language processing technology (Hugging Face's Transformers library) and analyzed. Text data is given as input, and text data with key points extracted is obtained as output.

[1193] (Server) The NLP system converts important information and conversational points from text data into a transcript format. The parsed text data is given as input, and formatted transcripts are generated as output.

[1194] (Server) The generated minutes are saved in a database. The minutes data is given as input, and saved in a database as output.

[1195] Step 5:

[1196] Search and memory completion functions

[1197] (User) uses a dedicated app to input specific keywords and dates to search for past minutes. The search keywords and dates are given as input, and the device sends a search request to the central server.

[1198] (Terminal) Receives the user's search request and sends it to the central server. The user's search keywords and date and time are given as input, and the search request is sent to the server as output.

[1199] (Server) Performs a search based on the specified keywords in the database and extracts relevant minutes. The search query is given as input, and the relevant minutes are extracted as output.

[1200] (Server) If the search results are incomplete, generative artificial intelligence (OpenAI GPT-4) is used to infer and complete the missing parts. The extracted incomplete minutes data is given as input, and the completed minutes are obtained as output.

[1201] (Terminal) Present the minutes containing the completed information to the user. The completed minutes data is given as input, and is displayed on the user's terminal as output.

[1202] As described above, the specific actions, data processing, and data calculations performed at each step form a series of flows, which realize efficient recording of conversations, generation of minutes, and search and completion functions.

[1203] (Application example 1)

[1204] 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."

[1205] Conversations held inside autonomous vehicles are likely to contain important information, but there is no appropriate means to efficiently record the content and use it as minutes. There is also a need for a simple and effective process for uploading and searching audio files, as well as for completing incomplete minutes. This invention solves these problems and provides a system for efficiently recording, searching, and completing conversations inside autonomous vehicles and using them as minutes.

[1206] 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.

[1207] In this invention, the server includes means for recording user conversations in real time, means for uploading recorded audio files to a central server, means for converting the audio files stored in the central server into text using a voice recognition system, means for analyzing the converted text using natural language processing technology and converting it into minutes format, means for searching past minutes from a database in response to a search request from a user, means for completing missing parts using generative AI when the search results are incomplete, means for displaying the minutes including the completed information so that the user can check them, means for recording conversations inside an autonomous vehicle, adding metadata to the recorded conversations, means for creating minutes including specific dates, times, and locations based on the metadata, and means for processing search requests based on the vehicle's location and ID information. This makes it possible to efficiently record conversations inside an autonomous vehicle and increase the availability of minutes.

[1208] "Recording a user's conversation in real time" means recording the user's voice immediately on the spot using an electronic device.

[1209] "Uploading the recorded audio file to the central server" means transferring the audio data stored in the device to a central storage device via a network.

[1210] "Converting voice files stored on a central server into text using a voice recognition system" means converting voice data stored in a central storage device into text data using voice recognition technology.

[1211] "Analyzing the text data using natural language processing technology and converting it into minutes format" means analyzing the data converted from audio to text using natural language processing technology, extracting summaries and important points, and formatting it into an official record format.

[1212] "Searching for past minutes from a database in response to a search request from a user" means receiving a search request from a user based on specific conditions and searching for past records in a database in accordance with that request.

[1213] "Using generative AI to fill in missing parts when search results are incomplete" means that when the results obtained through a search are incomplete, generative AI is used to supplement and add missing information.

[1214] "Displaying the minutes including the supplemented information so that the user can check them" means visually presenting the final minutes to the user, including the information supplemented by the generative AI.

[1215] "Recording conversations inside an autonomous vehicle and uploading them with metadata" means recording voice conversations inside an autonomous vehicle and sending the audio files to a central server along with additional information such as date, time, and location.

[1216] "Creating minutes with specific dates, times, and locations based on metadata" means using additional information (metadata) to generate an official record based on the content of conversations that took place at specific times and locations.

[1217] "Processing search requests based on vehicle location and ID information" means accepting search requests from users based on the vehicle's current location and identification information, and locating and providing appropriate records.

[1218] The present invention relates to a system for efficiently recording conversations in an autonomous vehicle and using the recorded conversations as minutes. Detailed embodiments of the present invention will be described below.

[1219] System configuration

[1220] User device: A smartphone or in-car smart device used to start and stop recording conversations and upload audio data to a central server.

[1221] Central server: Responsible for storing recording data, speech recognition, natural language processing, and generating meeting transcripts.

[1222] Database: A storage system for storing audio files, text data, and generated transcripts.

[1223] Generative AI: Used to complete incomplete search results and generate detailed meeting transcripts.

[1224] Program processing

[1225] Record and upload conversations

[1226] 1. The user device launches the app inside the autonomous vehicle and selects "start recording."

[1227] 2. The device uses the vehicle's microphone to record the user's conversation in real time and temporarily stores the audio data.

[1228] 3. When the recording is complete, an audio file is generated and metadata such as the date and time, vehicle ID, and location information is added to the file. This metadata specifically includes GPS information and the vehicle's identification ID.

[1229] 4. The device uploads the audio file to a central server following security protocols.

[1230] Audio data processing

[1231] 5. The central server receives the uploaded audio files and stores them in a database.

[1232] 6. The server converts the voice data into text using a speech recognition system (e.g., Google Cloud Speech-To-Text API).

[1233] 7. The text data is fed into a natural language processing system based on Hugging Face's transformers library, which extracts and summarizes important information and key points of the conversation.

[1234] 8. The generated minutes are stored in a database for users to refer to later.

[1235] Search and complete meeting minutes

[1236] 9. The user uses the app to search for minutes by specifying specific keywords, date and time, and vehicle location information.

[1237] 10. The central server uses Elasticsearch to search for the relevant minutes in the database and provide the results to the user.

[1238] 11. If the search results are incomplete, generative AI (e.g., GPT-3) will fill in the missing parts and generate a completed transcript.

[1239] 12. The completed minutes will be displayed on the terminal for visual confirmation.

[1240] Specific examples

[1241] Meeting minutes

[1242] User: When starting a sales strategy meeting in an autonomous vehicle, the user enters into the app, "I will record today's sales strategy meeting. It will start at 10:00 and will be held at the Tokyo branch."

[1243] Server: Records conversations during meetings, adds metadata, and uploads them to a central server.

[1244] Server: The uploaded audio data is converted into text and minutes format, and stored in a database.

[1245] Search past minutes

[1246] User: Type into the app, "Find minutes from last December's marketing meeting."

[1247] Server: Searches the database using Elasticsearch and displays the relevant minutes on the user's terminal.

[1248] Server: If the search results are incomplete, the generative AI fills in the gaps and provides information.

[1249] This makes it possible to efficiently record conversations and discussions inside an autonomous vehicle and easily search for and supplement the necessary information.

[1250] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1251] Step 1:

[1252] The user launches the dedicated app in the autonomous vehicle and selects "start recording." This starts the recording function.

[1253] Step 2:

[1254] The device uses the vehicle's microphone to record the user's conversation in real time. The recorded data is temporarily stored in the device's storage. The input is the user's voice data, and the output is a temporary audio file.

[1255] Step 3:

[1256] When the user selects "stop recording," the device stops recording and generates an audio file. This audio file is assigned metadata such as the date and time, vehicle ID, and location information. The input is a temporary audio file and metadata, and the output is an audio file with metadata.

[1257] Step 4:

[1258] The device uploads the generated audio file to a central server according to a security protocol. The input is the audio file with metadata, and the output is the saved audio file on the central server.

[1259] Step 5:

[1260] The server receives the uploaded audio files and stores them in a database. The input is the audio file, and the output is the audio data stored in the database.

[1261] Step 6:

[1262] The server passes the audio file to a speech recognition system and converts it into text data, for example, using the Google Cloud Speech-To-Text API. The input is audio data, and the output is text data.

[1263] Step 7:

[1264] Text data is input into a natural language processing system to extract and analyze important information and key points of the conversation. This analysis uses the Hugging Face transformers library. The input is text data, and the output is summary data in the form of meeting minutes.

[1265] Step 8:

[1266] The server saves the generated minutes in a database so that users can refer to them later. The input is summary data in minutes format, and the output is minutes saved in the database.

[1267] Step 9:

[1268] The user searches for meeting minutes using a dedicated app by specifying specific keywords, date and time, and vehicle location information. This operation sends a search request to the server. The input is a search query, and the output is a search request.

[1269] Step 10:

[1270] The server uses Elasticsearch to search for the relevant minutes in the database and provide the results to the user. The input is the search request and the output is the search results.

[1271] Step 11:

[1272] If the search results are incomplete, we use a generative AI (e.g., GPT-3) to fill in the gaps and generate a complete transcript. The input is the incomplete search results, and the output is the completed transcript.

[1273] Step 12:

[1274] The completed minutes are displayed on the user's terminal, allowing the user to visually check them. The input is the completed minutes, and the output is the minutes displayed on the user's terminal.

[1275] 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.

[1276] The present invention relates to a system that electronically records user conversations, analyzes and supplements the content, and automatically formats it into minutes. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, emotional information is added to the conversation content, generating more detailed minutes.

[1277] System Overview

[1278] 1. Transcript of conversation:

[1279] (User) The user launches the dedicated app and gives the command to "start recording."

[1280] (Terminal) The terminal launches a voice recording program and begins recording the conversation in real time using the microphone.

[1281] (Device) At the same time, the emotion engine is also activated, analyzing emotions from the user's conversation in real time.

[1282] 2. Integration and upload of voice and emotion data:

[1283] (Device) During recording, audio data is temporarily stored in a buffer.

[1284] (Device) Along with the recorded data, emotion data recognized by the emotion engine is also recorded.

[1285] (Device) Convert the recorded data into a file format (e.g. WAV, MP3) at regular intervals and add emotional information as metadata.

[1286] (Device) Upload the audio file to a central server using a secure protocol (e.g., HTTPS).

[1287] 3. Central server for data storage and speech recognition:

[1288] (Server) Receives the uploaded audio file and emotion metadata and stores them in a database.

[1289] (Server) The saved audio file is passed to a speech recognition system and converted into text data.

[1290] 4. Generating meeting minutes using natural language processing:

[1291] (Server) The textual data is input into a natural language processing (NLP) system and analysis is performed.

[1292] (Server) The NLP system extracts important information and key points of the conversation from the text data and converts them into minutes format that reflects emotional information.

[1293] (Server) The generated minutes are saved in a database, and users can refer to them later.

[1294] 5. Search and memory completion functions:

[1295] (User) Using a dedicated app, users can search for past minutes by specifying specific keywords or dates and times.

[1296] (Terminal) The terminal that receives the user's search request sends it to the central server.

[1297] (Server) The server searches the minutes based on the specified keywords in the database and extracts relevant minutes.

[1298] (Server) If the search results are incomplete, prepare to fill in the missing parts using generative AI.

[1299] (Server) The generative AI predicts the missing parts and generates new minutes based on the supplemented information.

[1300] (Terminal) The minutes including the completed information are presented to the user and displayed visually.

[1301] Specific examples

[1302] Example 1: Meeting recording and sentiment analysis

[1303] (User) "I'll record the meeting starting at 10 o'clock today," and opens the dedicated app.

[1304] (Device) The app launches and you press the record button to record the start of the meeting.

[1305] (Device) The device records all conversations, and the emotion engine analyzes and records emotions from the conversation.

[1306] (Device) Stop recording when the meeting ends and generate an audio file.

[1307] (Device) The audio file is given metadata such as "10:00 Meeting, October 20, 2023," as well as emotional data, and then uploaded to a central server.

[1308] Example 2: Search and complete past conversations

[1309] (User) "I want to check what was decided at last year's project meeting," so he enters the search keyword "Project 2022" into the app.

[1310] (Terminal) Receives the user's search request and sends it to the central server.

[1311] (Server) Search the database for relevant minutes and extract those that match the specified keywords.

[1312] (Server) If the results are incomplete, the generative AI will complete the relevant information and add details.

[1313] (Terminal) The completed minutes are displayed to the user, who then confirms the contents.

[1314] The system allows users to obtain not just minutes, but detailed records that reflect the emotions expressed during the conversation, leading to a deeper understanding of the conversation and more reliable decision-making.

[1315] The processing flow will be explained below.

[1316] Step 1:

[1317] (User) The user launches the dedicated app and gives the command to "start recording."

[1318] (Device) The app will launch the voice recording program and emotion engine and begin recording your conversation in real time using the microphone.

[1319] (Device) The emotion engine simultaneously analyzes the user's tone of voice and vocabulary to generate emotion data.

[1320] Step 2:

[1321] (Device) During recording, the voice data and emotion data are temporarily stored in a buffer.

[1322] (Device) Convert the recorded data into a file format (e.g. WAV, MP3) at regular intervals and add emotional information as metadata.

[1323] Step 3:

[1324] (Device) Once the recording is complete, prepare the audio file and emotion data for uploading to the central server.

[1325] (Device) Sends the audio file to a central server using a secure protocol (e.g., HTTPS).

[1326] (Server) Receives the uploaded audio file and emotional metadata.

[1327] Step 4:

[1328] (Server) Store the received audio files and emotion data in a database.

[1329] (Server) The saved audio file is passed to a speech recognition system and converted into text data.

[1330] Step 5:

[1331] (Server) The speech recognition system converts the speech data into text data.

[1332] (Server) Emotion data is extracted along with the converted text data, and the process proceeds to the next step.

[1333] Step 6:

[1334] (Server) Input text data and emotion data into a natural language processing (NLP) system.

[1335] (Server) The NLP system analyzes the text data and extracts key information (e.g., decisions, action items) and sentiment changes.

[1336] Step 7:

[1337] (Server) Based on the extracted information, convert it into a minutes format that reflects changes in emotions.

[1338] (Server) Save the generated minutes in the database and prepare to notify the user.

[1339] Step 8:

[1340] (User) When you want to look back on a past conversation, you can use a dedicated app to enter search keywords and date and time.

[1341] (Terminal) The terminal that receives the user's search request sends it to the central server.

[1342] Step 9:

[1343] (Server) Search for minutes based on specified keywords in the database.

[1344] (Server) Search results also include related emotional information.

[1345] Step 10:

[1346] (Server) If the search results are incomplete, prepare to fill in the missing parts using generative AI.

[1347] Step 11:

[1348] (Server) The generative AI predicts the missing parts and generates new minutes based on the supplemented information.

[1349] (Server) Sends the completed minutes and emotion information to the user terminal.

[1350] Step 12:

[1351] (Terminal) The transmitted minutes and emotion information are displayed on the user interface.

[1352] (User) Check the completed results and sentiment information to obtain the required information.

[1353] Example 2

[1354] 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."

[1355] While conventional meeting minutes-taking systems have the ability to record and analyze the content of conversations, they are unable to reflect emotional information during the conversation, and therefore are unable to fully convey the nuances and atmosphere of the discussion.In addition, when search results are incomplete, they lack the functionality to supplement appropriate information, which can make it inconvenient for users to use past meeting minutes.

[1356] 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.

[1357] In this invention, the server includes means for recording user conversations in real time, means for analyzing and integrating emotional data into the recorded audio data in real time, means for uploading the audio files and emotional metadata to a central server, means for converting the audio files stored in the central server into text using a voice recognition system, means for analyzing the converted text using natural language processing technology and converting it into a minutes format including emotional information, means for searching a database for past minutes in response to a search request from a user, means for completing missing parts using generative AI if the search results are incomplete, and means for displaying the minutes including the completed information so that the user can review them. This makes it possible to add emotional information to the content of the conversation, creating more detailed and easy-to-understand minutes, and automatically completing missing information during searches, allowing users to efficiently utilize past minutes.

[1358] "User" refers to any person or entity that operates this system.

[1359] "Conversation" refers to the content of communication uttered by a user as voice.

[1360] "Recording in real time" refers to recording the user's speech as audio data in real time.

[1361] "Emotion data" refers to emotional information analyzed from the user's voice.

[1362] "Audio data" refers to audio files that record the user's conversation.

[1363] "Emotional metadata" refers to meta-information that includes emotional information added to audio data.

[1364] "Central Server" refers to a server for storing and processing audio data and emotional metadata.

[1365] "Speech recognition system" refers to technology or software for converting voice data into text data.

[1366] "Text conversion" refers to the analysis of audio data and conversion into text information.

[1367] "Natural language processing technology (NLP)" refers to technology for analyzing text data, understanding meaning and context, and extracting information.

[1368] "Minutes format" refers to a document format that organizes the contents of the conversation and summarizes the main points.

[1369] "Generative AI" refers to artificial intelligence that has reasoning techniques to complement incomplete data and missing information.

[1370] "Database" refers to a system for storing and managing audio data, text data, and associated metadata in a searchable manner.

[1371] A "search request" refers to a query made by a user to a system seeking specific information.

[1372] "Search Results" means the information provided in response to a search request by the system.

[1373] "Complementary information" refers to data that generative AI adds to missing or incomplete search results.

[1374] "Visually displaying" refers to displaying information on a screen in a form that can be viewed by a user.

[1375] This invention relates to a system that electronically records user conversations, analyzes and supplements the content, and automatically creates minutes. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, emotional information is added to the conversation content, generating more detailed minutes.

[1376] The user first launches the dedicated app on a device such as a smartphone and gives the command "start recording." This causes the device to launch a voice recording program and begin recording the conversation in real time using the microphone. At the same time, an emotion engine is also launched, which analyzes the emotions from the user's conversation in real time. This emotion engine may use, for example, an emotion analysis API.

[1377] During recording, the audio data is temporarily stored in a buffer memory. At the same time, emotional data analyzed by the emotion engine is also recorded. The recorded data and emotional data are converted into audio files (e.g., WAV, MP3) at regular intervals, and emotional information is added as metadata. These audio files are then uploaded to a central server using a security protocol (e.g., HTTPS).

[1378] The server receives the uploaded audio file and emotion metadata and stores them in a database. The saved audio file is converted into text data by a speech recognition system. For example, the Google Cloud Speech-to-Text API can be used as the speech recognition system.

[1379] The text data is then input into a natural language processing (NLP) system on the server for analysis. The NLP system extracts important information and key points from the text data and converts them into minutes that reflect emotional information. The generated minutes are saved in a database for users to refer to later.

[1380] Users can use a dedicated app to search for past minutes by specifying specific keywords or dates. The device that receives the user's search request sends it to a central server. The server searches the database for minutes based on the specified keywords and extracts relevant minutes.

[1381] If the search results are incomplete, the server uses a generative AI (e.g., a generative AI model) to fill in the missing parts. The generative AI infers the missing parts and generates new minutes based on the completed information. The minutes, including the completed information, are sent to the device and visually displayed to the user.

[1382] Specific examples

[1383] Example 1: Meeting recording and sentiment analysis

[1384] User: "I want to record the meeting starting at 10 o'clock today," and opens the dedicated app.

[1385] Device: The app launches and you press the record button to record the start of the meeting.

[1386] Device: The device records all conversations, and the emotion engine analyzes and records emotions from the conversation.

[1387] Terminal: Stop recording when the meeting ends and generate an audio file.

[1388] Terminal: The audio file is given metadata such as "10:00 Meeting, October 20, 2023," as well as emotional data, and uploaded to a central server.

[1389] Example 2: Search and complete past conversations

[1390] User: "I want to check what was decided at last year's project meeting," so enters the search keyword "Project 2022" into the app.

[1391] Terminal: Receives user search requests and sends them to the central server.

[1392] Server: Searches the database for relevant minutes and extracts those that match the specified keywords.

[1393] Server: If the results are incomplete, the generative AI completes the relevant information and adds details.

[1394] Terminal: The completed minutes are displayed to the user, who then confirms the contents.

[1395] This system allows users to easily obtain highly detailed and emotionally sensitive meeting minutes, leading to a deeper understanding of the meeting content and more efficient decision-making.

[1396] Prompt Sentence Examples

[1397] Please explain in detail the process by which a user records a meeting and then analyzes the audio and emotion data to generate a transcript.

[1398] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1399] Step 1:

[1400] Launch the app and start recording

[1401] (User) The user launches the dedicated app on their smartphone and commands it to "start recording." This launches the app, and the user can start recording by touch operation or voice command.

[1402] Input: "Start recording" command from user

[1403] Output: Launch an audio recording program

[1404] Step 2:

[1405] Real-time audio recording

[1406] (Device) The app will launch a voice recording program in the background and begin recording your conversation in real time using your device's microphone.

[1407] Input: Audio input from the device microphone

[1408] Output: Saving real-time audio data to a buffer

[1409] Step 3:

[1410] Emotion engine activation and analysis

[1411] (Device) The emotion engine is started at the same time as recording, and the emotion is analyzed in real time from the user's conversation. At this point, the emotion analysis API is used to generate emotion data. For example, the emotion is identified from the tone, speed, and wording of the conversation.

[1412] Input: Real-time input of audio data

[1413] Output: Real-time emotion data generation

[1414] Step 4:

[1415] Data storage and integration

[1416] (Device) During recording, the voice data is temporarily stored in buffer memory, and the analyzed emotional data is also recorded.

[1417] Input: Real-time voice data and emotion data

[1418] Output: Temporarily saved voice data and emotion data

[1419] Step 5:

[1420] File Creation and Conversion

[1421] (Device) At regular intervals (for example, every 5 minutes), the recorded data is converted into an audio file format such as WAV or MP3, and emotional information is added as metadata. During this process, the audio data is encoded and emotional information is added as metadata.

[1422] Input: Temporarily saved voice data and emotion data

[1423] Output: Audio file (WAV, MP3 format) and emotional metadata

[1424] Step 6:

[1425] Uploading data

[1426] (Device) Upload the audio file and emotional metadata to a central server using a secure protocol such as HTTPS.

[1427] Input: Audio file and emotional metadata

[1428] Output: Data transfer to the server

[1429] Step 7:

[1430] Receiving and saving audio files

[1431] (Server) The server receives the audio file and emotional metadata uploaded from the device and stores them in a database.

[1432] Input: Uploaded audio file and sentiment metadata

[1433] Output: Save to database

[1434] Step 8:

[1435] Executing voice recognition

[1436] (Server) The server passes the saved audio file to a speech recognition system and converts it into text data, for example, using the Google Cloud Speech-to-Text API.

[1437] Input: Saved audio file

[1438] Output: Text data

[1439] Step 9:

[1440] Text data analysis

[1441] (Server) The data converted to text by speech recognition is input into a natural language processing (NLP) system for analysis. The NLP system extracts important information and key points from the text data.

[1442] Input: Text data

[1443] Output: Key information and talking points

[1444] Step 10:

[1445] Generate meeting minutes

[1446] (Server) Based on the extracted information, the minutes are converted into a format that reflects emotional information. The minutes are saved in a structured document format.

[1447] Input: Key information, conversation highlights, sentiment data

[1448] Output: Minutes document

[1449] Step 11:

[1450] Submitting a search request

[1451] (User) The user searches for past minutes by entering specific keywords or dates and times in a dedicated app.

[1452] Input: Search keyword, date and time

[1453] Output: Generate a search request

[1454] Step 12:

[1455] Submitting a search request

[1456] (Terminal) The terminal that receives the user's search request sends it to the central server.

[1457] Input: Search request

[1458] Output: Forwarding the request to the server

[1459] Step 13:

[1460] Database search

[1461] (Server) The server searches the minutes based on the specified keywords in the database and extracts relevant minutes.

[1462] Input: Search request, database

[1463] Output: Search results

[1464] Step 14:

[1465] Complementation by generative AI

[1466] (Server) If the search results are incomplete, the missing parts are filled in using a generative AI (e.g., a generative AI model). The generative AI infers the missing parts and generates complementary information.

[1467] Input: Incomplete search results

[1468] Output: Completed minutes information

[1469] Step 15:

[1470] Displaying search results

[1471] (Terminal) The minutes containing the supplemented information are presented to the user and displayed visually, allowing the user to confirm the necessary information.

[1472] Input: Completed minutes information

[1473] Output: Displaying information to the user

[1474] The above is a detailed description of the specific processing steps of the system.

[1475] (Application example 2)

[1476] 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."

[1477] In recent years, with the increasing popularity of self-driving vehicles, there is a growing need to record and analyze conversations and instructions given while driving. However, there is no system that can record and search detailed minutes that reflect the content and emotions of conversations, rather than simply recording audio. There is also a need for a system that can extract important information from conversations while driving and make it available for later review, thereby supporting driver decision-making.

[1478] The identification process by the identification 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 recording user conversations in real time; means for uploading recorded audio files to a central server; means for converting audio files stored in the central server into text using a voice recognition system; means for analyzing the converted text using natural language processing technology and converting it into a minutes format; means for searching past minutes from a database in response to a search request from a user; means for complementing missing parts using generative AI if the search results are incomplete; means for displaying the minutes including the complemented information so that the user can check them; means for recording conversations while driving using an on-board computer; and means for using an emotion analysis engine to analyze emotions from recorded audio data. This allows for detailed recording of conversations while driving and reflecting emotional information, thereby supporting the driver's decision-making and making it possible to easily search for and check important information later.

[1479] "Means for recording user conversations in real time" refers to devices or software that instantly record the user's conversations as audio on the spot.

[1480] "Means for uploading recorded audio files to a central server" refers to devices or software that transfer locally recorded audio data over a network to a remote server for storage.

[1481] "Means for converting audio files stored on a central server into text using a voice recognition system" refers to a system or software for analyzing audio data stored on a remote server and converting it into text data.

[1482] "Means of analyzing text data using natural language processing technology and converting it into minutes format" refers to devices or software that use natural language processing technology to analyze text data, extract the key points of the conversation, and format it into minutes.

[1483] "Means for searching a database for past minutes in response to a search request from a user" refers to a system or software for searching a database for minutes generated in the past based on keywords or conditions specified by the user.

[1484] "Means of using generative AI to fill in gaps when search results are incomplete" refers to devices or software that use generative AI technology to fill in gaps when search results do not contain sufficient information.

[1485] "Means for displaying the minutes including the completed information so that the user can check them" refers to a device or interface for displaying the minutes with the missing information completed so that the user can visually check them.

[1486] "Means for recording conversations while driving using an on-board computer" refers to devices or software that use a computer system installed in a vehicle to record conversations while driving.

[1487] "Means using an emotion analysis engine that analyzes emotions from recorded voice data" refers to an engine or software that analyzes recorded voice data and identifies and classifies emotions in conversations.

[1488] The present invention is a system that records user conversations in real time and analyzes and processes the recorded data. In particular, it records conversations while driving an autonomous vehicle and generates minutes that incorporate emotion analysis. Specific embodiments of the present invention are described below.

[1489] System Overview

[1490] The system consists of the following main elements:

[1491] A means for users to record conversations in real time

[1492] A means to upload recorded audio files to a central server

[1493] A means of converting audio files stored on a central server into text using a voice recognition system

[1494] A method for analyzing text data using natural language processing technology and converting it into minutes format

[1495] A means of searching past minutes from a database in response to a search request from a user

[1496] A means of using generative AI to fill in gaps in incomplete search results

[1497] A means of displaying the minutes with the completed information so that the user can review them

[1498] Means for recording conversations while driving using an on-board computer

[1499] A method using an emotion analysis engine to analyze emotions from recorded voice data

[1500] Hardware and software used

[1501] The following hardware and software is used:

[1502] Hardware:

[1503] Microphones in cars

[1504] In-vehicle computer

[1505] Internet connection device

[1506] software:

[1507] Audio recording software (e.g. PyAudio)

[1508] Sentiment analysis engine (e.g., EmotionEngine)

[1509] Speech recognition systems (e.g., Google Cloud Speech-to-Text)

[1510] Natural language processing systems (e.g., OpenAI's GPT-4)

[1511] Database management systems (e.g. MySQL, MongoDB)

[1512] Program processing overview

[1513] The overall system process is as follows:

[1514] 1. Real-time recording:

[1515] The terminal (on-board computer) uses the microphone inside the car to record the user's conversation in real time. The recorded voice is temporarily stored in a buffer, and the emotion analysis engine analyzes and records the emotional data in parallel.

[1516] 2. Generate and upload the audio file:

[1517] At regular intervals, the audio data in the buffer is converted to WAV or MP3 format, and an audio file with emotional information added as metadata is generated, which is then uploaded to a central server via a secure protocol (e.g., HTTPS).

[1518] 3. Speech Recognition and Text Transcription:

[1519] The central server inputs the received voice file into a voice recognition system and converts it into text data, which is then stored in a database.

[1520] 4. Natural Language Processing and Meeting Transcription Generation:

[1521] The text data is passed to a natural language processing system, which extracts and analyzes the main points of the conversation, generating minutes with emotional information added and saving them in a database.

[1522] 5. Search and complete:

[1523] When a user searches for past minutes, the system extracts relevant information from the database and displays it to the user. If there are any missing parts, it uses generative AI to fill in the gaps and provides that information to the user.

[1524] Specific examples

[1525] For example, when a user searches for minutes using keywords such as "Project Meeting October 20, 2023," the system will search for relevant minutes from the database and fill in any missing parts using generative AI.

[1526] Prompt Sentence Examples

[1527] "Search for minutes related to 'Project Meeting October 20, 2023' in your database. If there are any gaps, fill them in using your generative AI model."

[1528] The above is a specific embodiment of the present invention. By using this system, users can record conversations while driving in detail, generate minutes that include emotion analysis information, and easily search and check them later.

[1529] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1530] Step 1:

[1531] The terminal (on-board computer) uses the vehicle's microphone to record the user's conversation in real time. The input is the user's voice, and the output is the voice data. The voice recording software is launched and the recorded voice data is temporarily stored in a buffer. In parallel, the emotion analysis engine receives the user's voice data, analyzes and records the emotional data. The output of the emotion analysis engine is emotional information data.

[1532] Step 2:

[1533] The device periodically converts the audio data in the buffer into WAV or MP3 format. The input is the audio data stored in the buffer, and the output is an audio file (WAV or MP3). After the conversion is complete, emotional information is added as metadata and saved as tag information in the audio file.

[1534] Step 3:

[1535] The device uploads the audio file to the central server using the HTTPS protocol. The input is the audio file and its metadata, and the output is the audio file stored on the central server. The device verifies the secure connection and sends the audio file. The central server stores the received audio file in a database.

[1536] Step 4:

[1537] The server inputs the saved voice file into a voice recognition system and converts it into text data. The input is a voice file and the output is text data. The voice recognition system converts the voice data into text and records the text in a database.

[1538] Step 5:

[1539] The server passes the converted data to a natural language processing system for analysis. The input is text data, and the output is summarized data in the form of minutes. The natural language processing system analyzes the text data, extracts important content from the conversation, and converts it into minutes format. Emotional information is also reflected here.

[1540] Step 6:

[1541] The user sends a request to search for past meeting minutes from the terminal. The input is the user's search keywords and conditions, and the output is a search request. The terminal sends the request to the server, which searches the database for past records.

[1542] Step 7:

[1543] The server extracts minutes from the database that match the search criteria. The input is the user's search request, and the output is the corresponding minutes data. If the search results are incomplete, generative AI is used to fill in the missing parts and generate the final minutes data.

[1544] Step 8:

[1545] The server sends the minutes including the completed information to the terminal. The input is the complete minutes data, and the output is the data for display. The terminal visually displays the received minutes to the user, allowing them to check them.

[1546] 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.

[1547] 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.

[1548] 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.

[1549] [Fourth embodiment]

[1550] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1551] 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.

[1552] 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).

[1553] 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.

[1554] 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.

[1555] 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).

[1556] 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.

[1557] 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.

[1558] 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.

[1559] 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.

[1560] 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.

[1561] 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.

[1562] 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."

[1563] The present invention relates to a system for electronically recording user conversations, analyzing and supplementing the content of the recorded conversations, and automatically generating minutes of the conversations. Hereinafter, an embodiment of the present invention will be described in detail.

[1564] System Overview

[1565] 1. Transcript of conversation:

[1566] (User) The user launches the dedicated app and gives the command to "start recording."

[1567] (Device) The device uses a microphone to record the user's conversation in real time and stores the recorded data in temporary storage.

[1568] 2. Uploading audio data:

[1569] (Device) When recording is complete, an audio file is generated and metadata such as the date, time, location, and participants is added.

[1570] (Device) Uploads the recording file to a central server in accordance with security protocols.

[1571] 3. Central server for data storage and speech recognition:

[1572] (Server) Receives the uploaded audio file and stores it in the database.

[1573] (Server) The saved audio file is passed to a speech recognition system and converted into text data.

[1574] 4. Generating meeting minutes using natural language processing:

[1575] (Server) The textual data is input into a natural language processing (NLP) system and analysis is performed.

[1576] (Server) The NLP system extracts important information and conversational highlights from the text data and converts it into minutes format.

[1577] (Server) The generated minutes are saved in a database, and users can refer to them later.

[1578] 5. Search and memory completion functions:

[1579] (User) Using a dedicated app, users can search for past minutes by specifying specific keywords or dates and times.

[1580] (Terminal) The terminal that receives the user's search request sends it to the central server.

[1581] (Server) The server performs a search based on the specified keywords in the database and extracts relevant minutes.

[1582] (Server) If the search results are incomplete, generative AI is used to infer and complete the missing parts.

[1583] (Terminal) The minutes including the completed information are presented to the user and displayed visually.

[1584] Specific examples

[1585] Example 1: Meeting recording

[1586] (User) "I'll record the meeting starting at 10 o'clock today," and opens the dedicated app.

[1587] (Device) The app launches and you press the record button to record the start of the meeting.

[1588] (Device) The device records all conversations, stops recording when the meeting ends, and generates an audio file.

[1589] (Device) Add metadata such as "10:00 Meeting October 20, 2023" to the audio file and upload it to the central server.

[1590] Example 2: Search and complete past conversations

[1591] (User) "I want to check what was decided at last year's project meeting," so he enters the search keyword "Project 2022" into the app.

[1592] (Terminal) Receives the user's search request and sends it to the central server.

[1593] (Server) Search the database for relevant minutes and extract those that match the specified keywords.

[1594] (Server) If the results are incomplete, the generative AI will complete the relevant information and add details.

[1595] (Terminal) The completed minutes are displayed to the user, who then confirms the contents.

[1596] By going through these steps, users can automatically record, complete, and search the contents of important conversations and meetings, significantly improving the efficiency of their daily work. This invention will contribute to improving productivity, especially in the business world.

[1597] The processing flow will be explained below.

[1598] Step 1:

[1599] (User) The user launches the dedicated app and gives the command to "start recording."

[1600] (Device) The app will launch a voice recording program via voice recognition or tap.

[1601] (Device) The device will begin recording the conversation in real time using its microphone.

[1602] Step 2:

[1603] (Device) During recording, audio data is temporarily stored in a buffer.

[1604] (Device) Converts the recorded data into a file format (e.g. WAV, MP3) at regular intervals.

[1605] (Device) Adds metadata (date, time, location, participants) to the generated audio file.

[1606] Step 3:

[1607] (Device) Once recording is complete, prepare the audio file for uploading to the central server.

[1608] (Device) Sends the audio file to a central server using a secure protocol (e.g., HTTPS).

[1609] (Server) Receives the uploaded audio file.

[1610] Step 4:

[1611] (Server) Save the received audio file in the database.

[1612] (Server) Prepares the saved audio file for passing to the speech recognition system.

[1613] Step 5:

[1614] (Server) The speech recognition system analyzes the audio file and converts the audio data into text data.

[1615] (Server) Obtain the converted text data and proceed to the next processing step.

[1616] Step 6:

[1617] (Server) Inputs text data into a natural language processing (NLP) system.

[1618] (Server) NLP system analyzes the text data and extracts important information (e.g., decisions, action items).

[1619] Step 7:

[1620] (Server) Convert the extracted information into minutes format.

[1621] (Server) Save the generated minutes in the database and prepare to notify the user.

[1622] Step 8:

[1623] (User) When you want to look back on a past conversation, you can enter search keywords using a dedicated app.

[1624] (Terminal) Receives the user's search request and sends it to the central server.

[1625] Step 9:

[1626] (Server) Search for minutes based on specified keywords in the database.

[1627] (Server) If the search results are incomplete, prepare to fill in the missing parts using generative AI.

[1628] Step 10:

[1629] (Server) The generative AI predicts the missing parts and generates new minutes based on the supplemented information.

[1630] (Server) Sends the completed minutes to the user's terminal.

[1631] Step 11:

[1632] (Terminal) The minutes sent are displayed on the user interface.

[1633] (User) Check the completed results and obtain the required information.

[1634] Example 1

[1635] 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."

[1636] In today's business environment, many meetings and discussions are held, and it is necessary to record their contents and organize them for later reference. However, manual recording and minutes creation is time-consuming and prone to human error. Furthermore, it is difficult to efficiently search and reference past minutes, resulting in reduced work efficiency. It is also difficult to supplement information based on incomplete records. To solve these problems, a system is needed that can automatically create minutes using real-time conversation recording, speech recognition, and natural language processing, as well as highly accurate search and completion functions.

[1637] 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.

[1638] In this invention, the server includes a means for recording user conversations in real time, a means for adding metadata such as date, time, location, and participants to the recorded audio files and uploading them to a central server, and a means for sequentially converting the audio files stored on the central server into text using voice recognition technology. This allows users to efficiently record and manage conversation content and automatically generate important conversation content in the form of minutes. Users can also search past minutes based on specific keywords or dates and fill in missing parts using generative artificial intelligence, allowing them to quickly and accurately refer to the information they need.

[1639] "User" refers to a person who uses the system to record, upload, search, and check minutes of meetings.

[1640] A "terminal" is a device operated by a user, and is an electronic device that has a microphone, speaker, display, and Internet connection capabilities.

[1641] "Server" refers to a central computer system that stores audio files, processes speech recognition, natural language processing, generates and stores minutes, and performs search and completion processing.

[1642] "Conversation recording" refers to the act of collecting a user's speech as digital audio data using a microphone.

[1643] "Central Server" means a computer system located at the center of the system, which stores, analyzes, generates, searches, and completes voice data.

[1644] "Speech recognition technology" is a technology for converting voice data into text data, and is the process of analyzing voice signals and extracting their contents as text information.

[1645] "Natural language processing technology" is a technology that analyzes text data and performs processes such as semantic understanding, information extraction, and summary generation.

[1646] A "minutes format" is a document that organizes and records the contents of a conversation or meeting, and includes the date, time, location, participants, and what was said.

[1647] A "search request" is a request made by a user to locate past minutes based on specific keywords or dates.

[1648] "Data storage" refers to a memory area on a server for storing voice data, text data, minutes data, etc.

[1649] "Generative AI" is an AI technology that complements and generates new data based on existing information.

[1650] "Metadata" is auxiliary information added to an audio file, and includes attribute information such as date, time, location, and participants.

[1651] A "user interface" is an interface that allows users to operate the system, and provides functions such as voice recording, search, and checking minutes.

[1652] The present invention relates to a system for electronically recording user conversations, analyzing and supplementing the content of the recorded conversations, and automatically generating minutes of the conversations. Hereinafter, an embodiment of the present invention will be described in detail.

[1653] Conversation Record

[1654] (User) The user launches the dedicated app and instructs it to "start recording." Specifically, the user presses the "start recording button" using the smartphone app, which starts the recording process.

[1655] (Device) The device (such as a smartphone or tablet) uses a microphone to record the user's conversation in real time and saves the recorded data in temporary storage.

[1656] Uploading audio data

[1657] (Device) When the recording is complete, an audio file is generated and metadata such as the date, time, location, and participants is added. For example, metadata is added in the format "Meeting held at 10:00 on October 20, 2023."

[1658] (Device) Uploads the recording file to a central server following a security protocol. Specifically, the device sends the audio file to the server using an HTTPS connection.

[1659] Central server for data storage and voice recognition

[1660] (Server) Receives the uploaded audio files and stores them in a database. The server database has a table called "received_audio_files" that stores audio files and metadata.

[1661] (Server) The saved audio file is converted into text data using speech recognition technology (e.g., Google Cloud Speech-to-Text API). For example, an audio file of "10 o'clock meeting" is converted into text such as "The agenda for the meeting is budget review."

[1662] Generating meeting minutes using natural language processing

[1663] (Server) The text data is input into natural language processing technology (e.g., Hugging Face's Transformers library) and analyzed. The text data is analyzed into a format such as "Important matters: budget, person in charge: Tanaka."

[1664] (Server) Extracts important information and key points from the text data and converts them into minutes format. For example, minutes may be converted into a format such as "Meeting name: 10:00 Meeting, Date and time: October 20, 2023, Content: Budget review will be on the agenda."

[1665] (Server) The generated minutes are saved in the database. The minutes data is stored in a table called "meeting_minutes".

[1666] Search and memory completion functions

[1667] (User) Search for past minutes by specifying specific keywords or dates using the dedicated app. For example, enter "Project 2022" in the app and press the search button.

[1668] (Device) The device receives the user's search request and sends it to the central server. The search query is sent to the server's API.

[1669] (Server) Perform a search based on the specified keywords in the database to extract relevant minutes, for example, filtering minutes that match the keywords "project" and "2022".

[1670] (Server) If the search results are incomplete, generative artificial intelligence (e.g., OpenAI GPT-4) is used to infer and complete the missing parts. For example, GPT-4 completes the missing part about "project progress" and adds more detailed information.

[1671] (Device) The minutes containing the supplemented information are presented to the user, who confirms the contents. The supplemented minutes are displayed on the app screen.

[1672] Specific examples

[1673] Example 1: Meeting recording

[1674] (User) Opens the dedicated app and instructs, "I want to record the meeting starting at 10:00 today."

[1675] (Device) The app launches and you press the record button to record the start of the meeting.

[1676] (Device) The device records all conversations, stops recording when the meeting ends, and generates an audio file.

[1677] (Device) Add metadata such as "10:00 Meeting October 20, 2023" to the audio file and upload it to the central server.

[1678] Example 2: Search and complete past conversations

[1679] (User) In the app, he / she thinks, "I want to check what was decided at last year's project meeting," and enters the search keyword "Project 2022."

[1680] (Terminal) Receives the user's search request and sends it to the central server.

[1681] (Server) Search the database for relevant minutes and extract those that match the specified keywords.

[1682] (Server) If the results are incomplete, generative artificial intelligence (e.g., OpenAI GPT-4) will complete the relevant information and add details.

[1683] (Terminal) The completed minutes are displayed to the user, who then confirms the contents.

[1684] As described above, the present invention is a system that efficiently records user conversations and automatically converts important content into minutes format, thereby improving business efficiency.

[1685] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1686] Step 1:

[1687] Recording conversations

[1688] (User) Launches the dedicated app and instructs it to "start recording." This causes the device to receive as input the time the user started recording and the corresponding action.

[1689] (Device) Records user conversation in real time using a microphone. Audio data is collected sequentially and saved in temporary storage. An audio data file is generated as output.

[1690] Step 2:

[1691] Uploading audio data

[1692] (Device) When recording is complete, an audio file is generated and metadata such as the date, time, location, and participants are added. The command to stop recording and the metadata are given as input. This saves the audio file and metadata together.

[1693] (Device) Uploads the audio file to the central server using the HTTPS security protocol. The output is the audio file and metadata sent to the central server.

[1694] Step 3:

[1695] Central server for data storage and voice recognition

[1696] (Server) Receives uploaded audio files and stores them in a database. Receives audio files and metadata sent from the device as input. Stores audio files and metadata in a database as output.

[1697] (Server) The saved audio file is passed to speech recognition technology and converted into text data. Specifically, the Google Cloud Speech-to-Text API is used. The audio file is given as input, and the audio content is converted into text data as output.

[1698] Step 4:

[1699] Generating meeting minutes using natural language processing

[1700] (Server) The text data is input into natural language processing technology (Hugging Face's Transformers library) and analyzed. Text data is given as input, and text data with key points extracted is obtained as output.

[1701] (Server) The NLP system converts important information and conversational points from text data into a transcript format. The parsed text data is given as input, and formatted transcripts are generated as output.

[1702] (Server) The generated minutes are saved in a database. The minutes data is given as input, and saved in a database as output.

[1703] Step 5:

[1704] Search and memory completion functions

[1705] (User) uses a dedicated app to input specific keywords and dates to search for past minutes. The search keywords and dates are given as input, and the device sends a search request to the central server.

[1706] (Terminal) Receives the user's search request and sends it to the central server. The user's search keywords and date and time are given as input, and the search request is sent to the server as output.

[1707] (Server) Performs a search based on the specified keywords in the database and extracts relevant minutes. The search query is given as input, and the relevant minutes are extracted as output.

[1708] (Server) If the search results are incomplete, generative artificial intelligence (OpenAI GPT-4) is used to infer and complete the missing parts. The extracted incomplete minutes data is given as input, and the completed minutes are obtained as output.

[1709] (Terminal) Present the minutes containing the completed information to the user. The completed minutes data is given as input, and is displayed on the user's terminal as output.

[1710] As described above, the specific actions, data processing, and data calculations performed at each step form a series of flows, which realize efficient recording of conversations, generation of minutes, and search and completion functions.

[1711] (Application example 1)

[1712] 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."

[1713] Conversations held inside autonomous vehicles are likely to contain important information, but there is no appropriate means to efficiently record the content and use it as minutes. There is also a need for a simple and effective process for uploading and searching audio files, as well as for completing incomplete minutes. This invention solves these problems and provides a system for efficiently recording, searching, and completing conversations inside autonomous vehicles and using them as minutes.

[1714] 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.

[1715] In this invention, the server includes means for recording user conversations in real time, means for uploading recorded audio files to a central server, means for converting the audio files stored in the central server into text using a voice recognition system, means for analyzing the converted text using natural language processing technology and converting it into minutes format, means for searching past minutes from a database in response to a search request from a user, means for completing missing parts using generative AI when the search results are incomplete, means for displaying the minutes including the completed information so that the user can check them, means for recording conversations inside an autonomous vehicle, adding metadata to the recorded conversations, means for creating minutes including specific dates, times, and locations based on the metadata, and means for processing search requests based on the vehicle's location and ID information. This makes it possible to efficiently record conversations inside an autonomous vehicle and increase the availability of minutes.

[1716] "Recording a user's conversation in real time" means recording the user's voice immediately on the spot using an electronic device.

[1717] "Uploading the recorded audio file to the central server" means transferring the audio data stored in the device to a central storage device via a network.

[1718] "Converting voice files stored on a central server into text using a voice recognition system" means converting voice data stored in a central storage device into text data using voice recognition technology.

[1719] "Analyzing the text data using natural language processing technology and converting it into minutes format" means analyzing the data converted from audio to text using natural language processing technology, extracting summaries and important points, and formatting it into an official record format.

[1720] "Searching for past minutes from a database in response to a search request from a user" means receiving a search request from a user based on specific conditions and searching for past records in a database in accordance with that request.

[1721] "Using generative AI to fill in missing parts when search results are incomplete" means that when the results obtained through a search are incomplete, generative AI is used to supplement and add missing information.

[1722] "Displaying the minutes including the supplemented information so that the user can check them" means visually presenting the final minutes to the user, including the information supplemented by the generative AI.

[1723] "Recording conversations inside an autonomous vehicle and uploading them with metadata" means recording voice conversations inside an autonomous vehicle and sending the audio files to a central server along with additional information such as date, time, and location.

[1724] "Creating minutes with specific dates, times, and locations based on metadata" means using additional information (metadata) to generate an official record based on the content of conversations that took place at specific times and locations.

[1725] "Processing search requests based on vehicle location and ID information" means accepting search requests from users based on the vehicle's current location and identification information, and locating and providing appropriate records.

[1726] The present invention relates to a system for efficiently recording conversations in an autonomous vehicle and using the recorded conversations as minutes. Detailed embodiments of the present invention will be described below.

[1727] System configuration

[1728] User device: A smartphone or in-car smart device used to start and stop recording conversations and upload audio data to a central server.

[1729] Central server: Responsible for storing recording data, speech recognition, natural language processing, and generating meeting transcripts.

[1730] Database: A storage system for storing audio files, text data, and generated transcripts.

[1731] Generative AI: Used to complete incomplete search results and generate detailed meeting transcripts.

[1732] Program processing

[1733] Record and upload conversations

[1734] 1. The user device launches the app inside the autonomous vehicle and selects "start recording."

[1735] 2. The device uses the vehicle's microphone to record the user's conversation in real time and temporarily stores the audio data.

[1736] 3. When the recording is complete, an audio file is generated and metadata such as the date and time, vehicle ID, and location information is added to the file. This metadata specifically includes GPS information and the vehicle's identification ID.

[1737] 4. The device uploads the audio file to a central server following security protocols.

[1738] Audio data processing

[1739] 5. The central server receives the uploaded audio files and stores them in a database.

[1740] 6. The server converts the voice data into text using a speech recognition system (e.g., Google Cloud Speech-To-Text API).

[1741] 7. The text data is fed into a natural language processing system based on Hugging Face's transformers library, which extracts and summarizes important information and key points of the conversation.

[1742] 8. The generated minutes are stored in a database for users to refer to later.

[1743] Search and complete meeting minutes

[1744] 9. The user uses the app to search for minutes by specifying specific keywords, date and time, and vehicle location information.

[1745] 10. The central server uses Elasticsearch to search for the relevant minutes in the database and provide the results to the user.

[1746] 11. If the search results are incomplete, generative AI (e.g., GPT-3) will fill in the missing parts and generate a completed transcript.

[1747] 12. The completed minutes will be displayed on the terminal for visual confirmation.

[1748] Specific examples

[1749] Meeting minutes

[1750] User: When starting a sales strategy meeting in an autonomous vehicle, the user enters into the app, "I will record today's sales strategy meeting. It will start at 10:00 and will be held at the Tokyo branch."

[1751] Server: Records conversations during meetings, adds metadata, and uploads them to a central server.

[1752] Server: The uploaded audio data is converted into text and minutes format, and stored in a database.

[1753] Search past minutes

[1754] User: Type into the app, "Find minutes from last December's marketing meeting."

[1755] Server: Searches the database using Elasticsearch and displays the relevant minutes on the user's terminal.

[1756] Server: If the search results are incomplete, the generative AI fills in the gaps and provides information.

[1757] This makes it possible to efficiently record conversations and discussions inside an autonomous vehicle and easily search for and supplement the necessary information.

[1758] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1759] Step 1:

[1760] The user launches the dedicated app in the autonomous vehicle and selects "start recording." This starts the recording function.

[1761] Step 2:

[1762] The device uses the vehicle's microphone to record the user's conversation in real time. The recorded data is temporarily stored in the device's storage. The input is the user's voice data, and the output is a temporary audio file.

[1763] Step 3:

[1764] When the user selects "stop recording," the device stops recording and generates an audio file. This audio file is assigned metadata such as the date and time, vehicle ID, and location information. The input is a temporary audio file and metadata, and the output is an audio file with metadata.

[1765] Step 4:

[1766] The device uploads the generated audio file to a central server according to a security protocol. The input is the audio file with metadata, and the output is the saved audio file on the central server.

[1767] Step 5:

[1768] The server receives the uploaded audio files and stores them in a database. The input is the audio file, and the output is the audio data stored in the database.

[1769] Step 6:

[1770] The server passes the audio file to a speech recognition system and converts it into text data, for example, using the Google Cloud Speech-To-Text API. The input is audio data, and the output is text data.

[1771] Step 7:

[1772] Text data is input into a natural language processing system to extract and analyze important information and key points of the conversation. This analysis uses the Hugging Face transformers library. The input is text data, and the output is summary data in the form of meeting minutes.

[1773] Step 8:

[1774] The server saves the generated minutes in a database so that users can refer to them later. The input is summary data in minutes format, and the output is minutes saved in the database.

[1775] Step 9:

[1776] The user searches for meeting minutes using a dedicated app by specifying specific keywords, date and time, and vehicle location information. This operation sends a search request to the server. The input is a search query, and the output is a search request.

[1777] Step 10:

[1778] The server uses Elasticsearch to search for the relevant minutes in the database and provide the results to the user. The input is the search request and the output is the search results.

[1779] Step 11:

[1780] If the search results are incomplete, we use a generative AI (e.g., GPT-3) to fill in the gaps and generate a complete transcript. The input is the incomplete search results, and the output is the completed transcript.

[1781] Step 12:

[1782] The completed minutes are displayed on the user's terminal, allowing the user to visually check them. The input is the completed minutes, and the output is the minutes displayed on the user's terminal.

[1783] 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.

[1784] The present invention relates to a system that electronically records user conversations, analyzes and supplements the content, and automatically formats it into minutes. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, emotional information is added to the conversation content, generating more detailed minutes.

[1785] System Overview

[1786] 1. Transcript of conversation:

[1787] (User) The user launches the dedicated app and gives the command to "start recording."

[1788] (Terminal) The terminal launches a voice recording program and begins recording the conversation in real time using the microphone.

[1789] (Device) At the same time, the emotion engine is also activated, analyzing emotions from the user's conversation in real time.

[1790] 2. Integration and upload of voice and emotion data:

[1791] (Device) During recording, audio data is temporarily stored in a buffer.

[1792] (Device) Along with the recorded data, emotion data recognized by the emotion engine is also recorded.

[1793] (Device) Convert the recorded data into a file format (e.g. WAV, MP3) at regular intervals and add emotional information as metadata.

[1794] (Device) Upload the audio file to a central server using a secure protocol (e.g., HTTPS).

[1795] 3. Central server for data storage and speech recognition:

[1796] (Server) Receives the uploaded audio file and emotion metadata and stores them in a database.

[1797] (Server) The saved audio file is passed to a speech recognition system and converted into text data.

[1798] 4. Generating meeting minutes using natural language processing:

[1799] (Server) The textual data is input into a natural language processing (NLP) system and analysis is performed.

[1800] (Server) The NLP system extracts important information and key points of the conversation from the text data and converts them into minutes format that reflects emotional information.

[1801] (Server) The generated minutes are saved in a database, and users can refer to them later.

[1802] 5. Search and memory completion functions:

[1803] (User) Using a dedicated app, users can search for past minutes by specifying specific keywords or dates and times.

[1804] (Terminal) The terminal that receives the user's search request sends it to the central server.

[1805] (Server) The server searches the minutes based on the specified keywords in the database and extracts relevant minutes.

[1806] (Server) If the search results are incomplete, prepare to fill in the missing parts using generative AI.

[1807] (Server) The generative AI predicts the missing parts and generates new minutes based on the supplemented information.

[1808] (Terminal) The minutes including the completed information are presented to the user and displayed visually.

[1809] Specific examples

[1810] Example 1: Meeting recording and sentiment analysis

[1811] (User) "I'll record the meeting starting at 10 o'clock today," and opens the dedicated app.

[1812] (Device) The app launches and you press the record button to record the start of the meeting.

[1813] (Device) The device records all conversations, and the emotion engine analyzes and records emotions from the conversation.

[1814] (Device) Stop recording when the meeting ends and generate an audio file.

[1815] (Device) The audio file is given metadata such as "10:00 Meeting, October 20, 2023," as well as emotional data, and then uploaded to a central server.

[1816] Example 2: Search and complete past conversations

[1817] (User) "I want to check what was decided at last year's project meeting," so he enters the search keyword "Project 2022" into the app.

[1818] (Terminal) Receives the user's search request and sends it to the central server.

[1819] (Server) Search the database for relevant minutes and extract those that match the specified keywords.

[1820] (Server) If the results are incomplete, the generative AI will complete the relevant information and add details.

[1821] (Terminal) The completed minutes are displayed to the user, who then confirms the contents.

[1822] The system allows users to obtain not just minutes, but detailed records that reflect the emotions expressed during the conversation, leading to a deeper understanding of the conversation and more reliable decision-making.

[1823] The processing flow will be explained below.

[1824] Step 1:

[1825] (User) The user launches the dedicated app and gives the command to "start recording."

[1826] (Device) The app will launch the voice recording program and emotion engine and begin recording your conversation in real time using the microphone.

[1827] (Device) The emotion engine simultaneously analyzes the user's tone of voice and vocabulary to generate emotion data.

[1828] Step 2:

[1829] (Device) During recording, the voice data and emotion data are temporarily stored in a buffer.

[1830] (Device) Convert the recorded data into a file format (e.g. WAV, MP3) at regular intervals and add emotional information as metadata.

[1831] Step 3:

[1832] (Device) Once the recording is complete, prepare the audio file and emotion data for uploading to the central server.

[1833] (Device) Sends the audio file to a central server using a secure protocol (e.g., HTTPS).

[1834] (Server) Receives the uploaded audio file and emotional metadata.

[1835] Step 4:

[1836] (Server) Store the received audio files and emotion data in a database.

[1837] (Server) The saved audio file is passed to a speech recognition system and converted into text data.

[1838] Step 5:

[1839] (Server) The speech recognition system converts the speech data into text data.

[1840] (Server) Emotion data is extracted along with the converted text data, and the process proceeds to the next step.

[1841] Step 6:

[1842] (Server) Input text data and emotion data into a natural language processing (NLP) system.

[1843] (Server) The NLP system analyzes the text data and extracts key information (e.g., decisions, action items) and sentiment changes.

[1844] Step 7:

[1845] (Server) Based on the extracted information, convert it into a minutes format that reflects changes in emotions.

[1846] (Server) Save the generated minutes in the database and prepare to notify the user.

[1847] Step 8:

[1848] (User) When you want to look back on a past conversation, you can use a dedicated app to enter search keywords and date and time.

[1849] (Terminal) The terminal that receives the user's search request sends it to the central server.

[1850] Step 9:

[1851] (Server) Search for minutes based on specified keywords in the database.

[1852] (Server) Search results also include related emotional information.

[1853] Step 10:

[1854] (Server) If the search results are incomplete, prepare to fill in the missing parts using generative AI.

[1855] Step 11:

[1856] (Server) The generative AI predicts the missing parts and generates new minutes based on the supplemented information.

[1857] (Server) Sends the completed minutes and emotion information to the user terminal.

[1858] Step 12:

[1859] (Terminal) The transmitted minutes and emotion information are displayed on the user interface.

[1860] (User) Check the completed results and sentiment information to obtain the required information.

[1861] Example 2

[1862] 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."

[1863] While conventional meeting minutes-taking systems have the ability to record and analyze the content of conversations, they are unable to reflect emotional information during the conversation, and therefore are unable to fully convey the nuances and atmosphere of the discussion.In addition, when search results are incomplete, they lack the functionality to supplement appropriate information, which can make it inconvenient for users to use past meeting minutes.

[1864] 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.

[1865] In this invention, the server includes means for recording user conversations in real time, means for analyzing and integrating emotional data into the recorded audio data in real time, means for uploading the audio files and emotional metadata to a central server, means for converting the audio files stored in the central server into text using a voice recognition system, means for analyzing the converted text using natural language processing technology and converting it into a minutes format including emotional information, means for searching a database for past minutes in response to a search request from a user, means for completing missing parts using generative AI if the search results are incomplete, and means for displaying the minutes including the completed information so that the user can review them. This makes it possible to add emotional information to the content of the conversation, creating more detailed and easy-to-understand minutes, and automatically completing missing information during searches, allowing users to efficiently utilize past minutes.

[1866] "User" refers to any person or entity that operates this system.

[1867] "Conversation" refers to the content of communication uttered by a user as voice.

[1868] "Recording in real time" refers to recording the user's speech as audio data in real time.

[1869] "Emotion data" refers to emotional information analyzed from the user's voice.

[1870] "Audio data" refers to audio files that record the user's conversation.

[1871] "Emotional metadata" refers to meta-information that includes emotional information added to audio data.

[1872] "Central Server" refers to a server for storing and processing audio data and emotional metadata.

[1873] "Speech recognition system" refers to technology or software for converting voice data into text data.

[1874] "Text conversion" refers to the analysis of audio data and conversion into text information.

[1875] "Natural language processing technology (NLP)" refers to technology for analyzing text data, understanding meaning and context, and extracting information.

[1876] "Minutes format" refers to a document format that organizes the contents of the conversation and summarizes the main points.

[1877] "Generative AI" refers to artificial intelligence that has reasoning techniques to complement incomplete data and missing information.

[1878] "Database" refers to a system for storing and managing audio data, text data, and associated metadata in a searchable manner.

[1879] A "search request" refers to a query made by a user to a system seeking specific information.

[1880] "Search Results" means the information provided in response to a search request by the system.

[1881] "Complementary information" refers to data that generative AI adds to missing or incomplete search results.

[1882] "Visually displaying" refers to displaying information on a screen in a form that can be viewed by a user.

[1883] This invention relates to a system that electronically records user conversations, analyzes and supplements the content, and automatically creates minutes. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, emotional information is added to the conversation content, generating more detailed minutes.

[1884] The user first launches the dedicated app on a device such as a smartphone and gives the command "start recording." This causes the device to launch a voice recording program and begin recording the conversation in real time using the microphone. At the same time, an emotion engine is also launched, which analyzes the emotions from the user's conversation in real time. This emotion engine may use, for example, an emotion analysis API.

[1885] During recording, the audio data is temporarily stored in a buffer memory. At the same time, emotional data analyzed by the emotion engine is also recorded. The recorded data and emotional data are converted into audio files (e.g., WAV, MP3) at regular intervals, and emotional information is added as metadata. These audio files are then uploaded to a central server using a security protocol (e.g., HTTPS).

[1886] The server receives the uploaded audio file and emotion metadata and stores them in a database. The saved audio file is converted into text data by a speech recognition system. For example, the Google Cloud Speech-to-Text API can be used as the speech recognition system.

[1887] The text data is then input into a natural language processing (NLP) system on the server for analysis. The NLP system extracts important information and key points from the text data and converts them into minutes that reflect emotional information. The generated minutes are saved in a database for users to refer to later.

[1888] Users can use a dedicated app to search for past minutes by specifying specific keywords or dates. The device that receives the user's search request sends it to a central server. The server searches the database for minutes based on the specified keywords and extracts relevant minutes.

[1889] If the search results are incomplete, the server uses a generative AI (e.g., a generative AI model) to fill in the missing parts. The generative AI infers the missing parts and generates new minutes based on the completed information. The minutes, including the completed information, are sent to the device and visually displayed to the user.

[1890] Specific examples

[1891] Example 1: Meeting recording and sentiment analysis

[1892] User: "I want to record the meeting starting at 10 o'clock today," and opens the dedicated app.

[1893] Device: The app launches and you press the record button to record the start of the meeting.

[1894] Device: The device records all conversations, and the emotion engine analyzes and records emotions from the conversation.

[1895] Terminal: Stop recording when the meeting ends and generate an audio file.

[1896] Terminal: The audio file is given metadata such as "10:00 Meeting, October 20, 2023," as well as emotional data, and uploaded to a central server.

[1897] Example 2: Search and complete past conversations

[1898] User: "I want to check what was decided at last year's project meeting," so enters the search keyword "Project 2022" into the app.

[1899] Terminal: Receives user search requests and sends them to the central server.

[1900] Server: Searches the database for relevant minutes and extracts those that match the specified keywords.

[1901] Server: If the results are incomplete, the generative AI completes the relevant information and adds details.

[1902] Terminal: The completed minutes are displayed to the user, who then confirms the contents.

[1903] This system allows users to easily obtain highly detailed and emotionally sensitive meeting minutes, leading to a deeper understanding of the meeting content and more efficient decision-making.

[1904] Prompt Sentence Examples

[1905] Please explain in detail the process by which a user records a meeting and then analyzes the audio and emotion data to generate a transcript.

[1906] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1907] Step 1:

[1908] Launch the app and start recording

[1909] (User) The user launches the dedicated app on their smartphone and commands it to "start recording." This launches the app, and the user can start recording by touch operation or voice command.

[1910] Input: "Start recording" command from user

[1911] Output: Launch an audio recording program

[1912] Step 2:

[1913] Real-time audio recording

[1914] (Device) The app will launch a voice recording program in the background and begin recording your conversation in real time using your device's microphone.

[1915] Input: Audio input from the device microphone

[1916] Output: Saving real-time audio data to a buffer

[1917] Step 3:

[1918] Emotion engine activation and analysis

[1919] (Device) The emotion engine is started at the same time as recording, and the emotion is analyzed in real time from the user's conversation. At this point, the emotion analysis API is used to generate emotion data. For example, the emotion is identified from the tone, speed, and wording of the conversation.

[1920] Input: Real-time input of audio data

[1921] Output: Real-time emotion data generation

[1922] Step 4:

[1923] Data storage and integration

[1924] (Device) During recording, the voice data is temporarily stored in buffer memory, and the analyzed emotional data is also recorded.

[1925] Input: Real-time voice data and emotion data

[1926] Output: Temporarily saved voice data and emotion data

[1927] Step 5:

[1928] File Creation and Conversion

[1929] (Device) At regular intervals (for example, every 5 minutes), the recorded data is converted into an audio file format such as WAV or MP3, and emotional information is added as metadata. During this process, the audio data is encoded and emotional information is added as metadata.

[1930] Input: Temporarily saved voice data and emotion data

[1931] Output: Audio file (WAV, MP3 format) and emotional metadata

[1932] Step 6:

[1933] Uploading data

[1934] (Device) Upload the audio file and emotional metadata to a central server using a secure protocol such as HTTPS.

[1935] Input: Audio file and emotional metadata

[1936] Output: Data transfer to the server

[1937] Step 7:

[1938] Receiving and saving audio files

[1939] (Server) The server receives the audio file and emotional metadata uploaded from the device and stores them in a database.

[1940] Input: Uploaded audio file and sentiment metadata

[1941] Output: Save to database

[1942] Step 8:

[1943] Executing voice recognition

[1944] (Server) The server passes the saved audio file to a speech recognition system and converts it into text data, for example, using the Google Cloud Speech-to-Text API.

[1945] Input: Saved audio file

[1946] Output: Text data

[1947] Step 9:

[1948] Text data analysis

[1949] (Server) The data converted to text by speech recognition is input into a natural language processing (NLP) system for analysis. The NLP system extracts important information and key points from the text data.

[1950] Input: Text data

[1951] Output: Key information and talking points

[1952] Step 10:

[1953] Generate meeting minutes

[1954] (Server) Based on the extracted information, the minutes are converted into a format that reflects emotional information. The minutes are saved in a structured document format.

[1955] Input: Key information, conversation highlights, sentiment data

[1956] Output: Minutes document

[1957] Step 11:

[1958] Submitting a search request

[1959] (User) The user searches for past minutes by entering specific keywords or dates and times in a dedicated app.

[1960] Input: Search keyword, date and time

[1961] Output: Generate a search request

[1962] Step 12:

[1963] Submitting a search request

[1964] (Terminal) The terminal that receives the user's search request sends it to the central server.

[1965] Input: Search request

[1966] Output: Forwarding the request to the server

[1967] Step 13:

[1968] Database search

[1969] (Server) The server searches the minutes based on the specified keywords in the database and extracts relevant minutes.

[1970] Input: Search request, database

[1971] Output: Search results

[1972] Step 14:

[1973] Complementation by generative AI

[1974] (Server) If the search results are incomplete, the missing parts are filled in using a generative AI (e.g., a generative AI model). The generative AI infers the missing parts and generates complementary information.

[1975] Input: Incomplete search results

[1976] Output: Completed minutes information

[1977] Step 15:

[1978] Displaying search results

[1979] (Terminal) The minutes containing the supplemented information are presented to the user and displayed visually, allowing the user to confirm the necessary information.

[1980] Input: Completed minutes information

[1981] Output: Displaying information to the user

[1982] The above is a detailed description of the specific processing steps of the system.

[1983] (Application example 2)

[1984] 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."

[1985] In recent years, with the increasing popularity of self-driving vehicles, there is a growing need to record and analyze conversations and instructions given while driving. However, there is no system that can record and search detailed minutes that reflect the content and emotions of conversations, rather than simply recording audio. There is also a need for a system that can extract important information from conversations while driving and make it available for later review, thereby supporting driver decision-making.

[1986] The identification process by the identification 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 recording user conversations in real time; means for uploading recorded audio files to a central server; means for converting audio files stored in the central server into text using a voice recognition system; means for analyzing the converted text using natural language processing technology and converting it into a minutes format; means for searching past minutes from a database in response to a search request from a user; means for complementing missing parts using generative AI if the search results are incomplete; means for displaying the minutes including the complemented information so that the user can check them; means for recording conversations while driving using an on-board computer; and means for using an emotion analysis engine to analyze emotions from recorded audio data. This allows for detailed recording of conversations while driving and reflecting emotional information, thereby supporting the driver's decision-making and making it possible to easily search for and check important information later.

[1987] "Means for recording user conversations in real time" refers to devices or software that instantly record the user's conversations as audio on the spot.

[1988] "Means for uploading recorded audio files to a central server" refers to devices or software that transfer locally recorded audio data over a network to a remote server for storage.

[1989] "Means for converting audio files stored on a central server into text using a voice recognition system" refers to a system or software for analyzing audio data stored on a remote server and converting it into text data.

[1990] "Means of analyzing text data using natural language processing technology and converting it into minutes format" refers to devices or software that use natural language processing technology to analyze text data, extract the key points of the conversation, and format it into minutes.

[1991] "Means for searching a database for past minutes in response to a search request from a user" refers to a system or software for searching a database for minutes generated in the past based on keywords or conditions specified by the user.

[1992] "Means of using generative AI to fill in gaps when search results are incomplete" refers to devices or software that use generative AI technology to fill in gaps when search results do not contain sufficient information.

[1993] "Means for displaying the minutes including the completed information so that the user can check them" refers to a device or interface for displaying the minutes with the missing information completed so that the user can visually check them.

[1994] "Means for recording conversations while driving using an on-board computer" refers to devices or software that use a computer system installed in a vehicle to record conversations while driving.

[1995] "Means using an emotion analysis engine that analyzes emotions from recorded voice data" refers to an engine or software that analyzes recorded voice data and identifies and classifies emotions in conversations.

[1996] The present invention is a system that records user conversations in real time and analyzes and processes the recorded data. In particular, it records conversations while driving an autonomous vehicle and generates minutes that incorporate emotion analysis. Specific embodiments of the present invention are described below.

[1997] System Overview

[1998] The system consists of the following main elements:

[1999] A means for users to record conversations in real time

[2000] A means to upload recorded audio files to a central server

[2001] A means of converting audio files stored on a central server into text using a voice recognition system

[2002] A method for analyzing text data using natural language processing technology and converting it into minutes format

[2003] A means of searching past minutes from a database in response to a search request from a user

[2004] A means of using generative AI to fill in gaps in incomplete search results

[2005] A means of displaying the minutes with the completed information so that the user can review them

[2006] Means for recording conversations while driving using an on-board computer

[2007] A method using an emotion analysis engine to analyze emotions from recorded voice data

[2008] Hardware and software used

[2009] The following hardware and software is used:

[2010] Hardware:

[2011] Microphones in cars

[2012] In-vehicle computer

[2013] Internet connection device

[2014] software:

[2015] Audio recording software (e.g. PyAudio)

[2016] Sentiment analysis engine (e.g., EmotionEngine)

[2017] Speech recognition systems (e.g., Google Cloud Speech-to-Text)

[2018] Natural language processing systems (e.g., OpenAI's GPT-4)

[2019] Database management systems (e.g. MySQL, MongoDB)

[2020] Program processing overview

[2021] The overall system process is as follows:

[2022] 1. Real-time recording:

[2023] The terminal (on-board computer) uses the microphone inside the car to record the user's conversation in real time. The recorded voice is temporarily stored in a buffer, and the emotion analysis engine analyzes and records the emotional data in parallel.

[2024] 2. Generate and upload the audio file:

[2025] At regular intervals, the audio data in the buffer is converted to WAV or MP3 format, and an audio file with emotional information added as metadata is generated, which is then uploaded to a central server via a secure protocol (e.g., HTTPS).

[2026] 3. Speech Recognition and Text Transcription:

[2027] The central server inputs the received voice file into a voice recognition system and converts it into text data, which is then stored in a database.

[2028] 4. Natural Language Processing and Meeting Transcription Generation:

[2029] The text data is passed to a natural language processing system, which extracts and analyzes the main points of the conversation, generating minutes with emotional information added and saving them in a database.

[2030] 5. Search and complete:

[2031] When a user searches for past minutes, the system extracts relevant information from the database and displays it to the user. If there are any missing parts, it uses generative AI to fill in the gaps and provides that information to the user.

[2032] Specific examples

[2033] For example, when a user searches for minutes using keywords such as "Project Meeting October 20, 2023," the system will search for relevant minutes from the database and fill in any missing parts using generative AI.

[2034] Prompt Sentence Examples

[2035] "Search for minutes related to 'Project Meeting October 20, 2023' in your database. If there are any gaps, fill them in using your generative AI model."

[2036] The above is a specific embodiment of the present invention. By using this system, users can record conversations while driving in detail, generate minutes that include emotion analysis information, and easily search and check them later.

[2037] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[2038] Step 1:

[2039] The terminal (on-board computer) uses the vehicle's microphone to record the user's conversation in real time. The input is the user's voice, and the output is the voice data. The voice recording software is launched and the recorded voice data is temporarily stored in a buffer. In parallel, the emotion analysis engine receives the user's voice data, analyzes and records the emotional data. The output of the emotion analysis engine is emotional information data.

[2040] Step 2:

[2041] The device periodically converts the audio data in the buffer into WAV or MP3 format. The input is the audio data stored in the buffer, and the output is an audio file (WAV or MP3). After the conversion is complete, emotional information is added as metadata and saved as tag information in the audio file.

[2042] Step 3:

[2043] The device uploads the audio file to the central server using the HTTPS protocol. The input is the audio file and its metadata, and the output is the audio file stored on the central server. The device verifies the secure connection and sends the audio file. The central server stores the received audio file in a database.

[2044] Step 4:

[2045] The server inputs the saved voice file into a voice recognition system and converts it into text data. The input is a voice file and the output is text data. The voice recognition system converts the voice data into text and records the text in a database.

[2046] Step 5:

[2047] The server passes the converted data to a natural language processing system for analysis. The input is text data, and the output is summarized data in the form of minutes. The natural language processing system analyzes the text data, extracts important content from the conversation, and converts it into minutes format. Emotional information is also reflected here.

[2048] Step 6:

[2049] The user sends a request to search for past meeting minutes from the terminal. The input is the user's search keywords and conditions, and the output is a search request. The terminal sends the request to the server, which searches the database for past records.

[2050] Step 7:

[2051] The server extracts minutes from the database that match the search criteria. The input is the user's search request, and the output is the corresponding minutes data. If the search results are incomplete, generative AI is used to fill in the missing parts and generate the final minutes data.

[2052] Step 8:

[2053] The server sends the minutes including the completed information to the terminal. The input is the complete minutes data, and the output is the data for display. The terminal visually displays the received minutes to the user, allowing them to check them.

[2054] 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.

[2055] 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.

[2056] 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.

[2057] 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.

[2058] 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.

[2059] 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.

[2060] 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).

[2061] 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.

[2062] 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."

[2063] 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.

[2064] 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).

[2065] 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.

[2066] 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.

[2067] 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.

[2068] 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.

[2069] 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.

[2070] 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.

[2071] 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.

[2072] 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.

[2073] 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.

[2074] 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.

[2075] The following is further disclosed regarding the above embodiment.

[2076] (Claim 1)

[2077] a means for recording user conversations in real time;

[2078] A means for uploading the recorded audio files to a central server;

[2079] A means for converting the voice files stored in the central server into text using a voice recognition system;

[2080] A means of analyzing the text data using natural language processing technology and converting it into minutes format;

[2081] A means for searching past minutes from a database in response to a search request from a user;

[2082] When search results are incomplete, generative AI can be used to fill in the gaps.

[2083] The system includes a me...

Claims

1. a means for recording user conversations in real time; A means for uploading the recorded audio files to a central server; A means for converting the voice files stored in the central server into text using a voice recognition system; A means of analyzing the text data using natural language processing technology and converting it into minutes format; A means for searching past minutes from a database in response to a search request from a user; When search results are incomplete, generative AI can be used to fill in the gaps. The system includes a means for displaying the minutes including the completed information so that the user can review them.

2. 10. The system of claim 1, wherein the system provides metadata to audio files.

3. The system according to claim 1, wherein the system uses generative AI to clearly indicate the information that has been supplemented.

Images