System

Abstract

A system is provided.SOLUTION: A system includes a means for registering and learning voice data of each participant in advance, a means for transcribing voice in real time during a conference, a means for removing unnecessary voice from transcription data, a means for automatically summarizing conference contents, a means for automatically transmitting summary data to participants, and a means for retrieving past minutes in cooperation with a chatbot.SELECTED DRAWING: Figure 1

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] Conventional meeting recording methods have the following issues. First, transcription tools have difficulty accurately identifying speakers, making it difficult to clearly understand who said what. Second, minutes often include unnecessary sounds such as "umm" and "ah," making it time-consuming and laborious to create them. Furthermore, summarizing the meeting content is a significant burden, requiring quick and accurate summarization. Furthermore, a clear and shared record is necessary to prevent disputes over "heard" or "unsaid" that arise during or after the meeting. [Means for solving the problem]

[0005] To solve the above problems, the present invention provides a system that includes a means for registering and learning the voice data of each participant in advance, a means for transcribing the voice in real time during the meeting, a means for removing unnecessary voices from the transcript data, a means for automatically summarizing the meeting content, a means for automatically sending the summary data to the participants, and a means for searching past minutes in cooperation with a chatbot. This makes it possible to identify speakers, remove unnecessary voices, provide quick summaries, and reduce disputes over "said" and "unsaid."

[0006] "Audio data" refers to a digital audio file that records what was said during a meeting.

[0007] "Means for pre-registration and learning" refers to the function of registering the characteristic voice of each participant on the server and training it using a voice recognition model.

[0008] "Means for real-time audio transcription" refers to the function of instantly converting audio data sent from a device during a meeting into text format.

[0009] "Means for removing unnecessary audio" refers to a function that automatically removes meaningless audio such as "eh" and "um" from transcribed text data.

[0010] "Automatic summarization" refers to a function that uses generative AI to extract important content from a meeting and generate concise summary information.

[0011] "Automatic sending means" refers to a function that sends the summary data generated after the conference has ended to the registered email addresses of the participants.

[0012] "Means for linking with chatbots" refers to the ability to use a chat interface to search past minutes and provide users with the necessary information.

[0013] "Means for identifying the speaker" refers to a function that identifies who spoke from transcribed audio data.

[0014] "Minutes data" refers to a text file that summarizes and records the contents of a meeting.

[0015] The means for sending to "email addresses" is a function for sending the minutes data to the email addresses of the participants. [Brief explanation of the drawings]

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

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

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

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

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

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

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

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

[0024] [First embodiment]

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

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

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

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

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

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

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

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

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

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

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

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

[0037] This invention is a system for automatically generating and efficiently managing meeting minutes. This system pre-registers and learns from the voice data of each participant to accurately identify who spoke, transcribes the voice in real time during the meeting, removes unnecessary voice, summarizes the meeting content, and automatically sends the summary data to participants after the meeting. It is also possible to search past meeting minutes in conjunction with a chatbot.

[0038] Below, each component of this system and its processing will be specifically explained.

[0039] Server Processing

[0040] 1. Registering and learning audio data

[0041] The server receives voice data previously uploaded by users and trains a speech recognition model, building a database of each participant's distinctive voice, enabling accurate identification of who is speaking during a meeting.

[0042] 2. Transcription of audio data

[0043] When a meeting begins, the device records what is being said and sends it to the server, which receives the audio data in real time and transcribes it. With the help of a speech recognition model, the transcription is done while identifying the speaker.

[0044] 3. Remove unwanted audio

[0045] The server uses natural language processing technology to automatically remove unnecessary sounds such as "um" and "ah" from the recorded text, making the minutes more concise and easier to read.

[0046] 4. Automatic summary generation

[0047] The server inputs the transcribed data of all statements into the generation AI, which automatically extracts the key points of the meeting and creates a summary. The summarized minutes are then stored on the server.

[0048] 5. Sending meeting minutes

[0049] After the meeting, the server automatically sends the generated summary to the participants' email addresses, allowing participants to check the minutes immediately after the meeting ends.

[0050] 6. Chatbot integration

[0051] The server works in conjunction with the chatbot to search past minutes in response to user inquiries and provide the necessary information.

[0052] Terminal handling

[0053] 1. Registering audio data

[0054] Users record their own voice using their devices and upload the voice data to the server, which allows the server to learn the voice data of each participant in advance.

[0055] 2. Sending audio data during a meeting

[0056] When the meeting begins, the device records the user's speech in real time and sends it to the server, which receives it and immediately transcribes it.

[0057] 3. Checking meeting notes

[0058] After the meeting ends, the terminal receives the minutes sent from the server and displays them to the user, allowing the user to quickly grasp the main points of the meeting.

[0059] User Action

[0060] 1. Registering audio data

[0061] Users use their devices to record and register their own voice, which is then uploaded to a server and used as training data for the voice recognition model.

[0062] 2. Statements made during meetings

[0063] Users can join a conference and speak as usual. The device records what they say in real time and sends it to the server, so no special operations are required on the part of the user.

[0064] 3. Checking meeting notes

[0065] After the meeting, users can check the minutes data received from their devices. The system's automatic summary function allows users to quickly grasp the main points of the meeting.

[0066] 4. Integration with chatbots

[0067] Users can use the chatbot to easily search for and check past meeting minutes or specific comments. The chatbot extracts the relevant information from a database on the server and presents it to the user.

[0068] By introducing this system, the creation of meeting minutes will be automated, significantly improving work efficiency. It also identifies speakers and removes unnecessary audio, improving the quality of meeting minutes. This reduces disputes over "said" and "unsaid," making it possible to run meetings with greater transparency.

[0069] The processing flow will be explained below.

[0070] Server Processing

[0071] Step 1: Register and train audio data

[0072] 1. The user records audio data and saves it on the device.

[0073] The user records their own voice and saves it on the device.

[0074] 2. The device uploads the audio data to the server

[0075] The device transmits the stored voice data to the server.

[0076] 3. The server receives the voice data and learns

[0077] The server receives the audio data and trains a speech recognition model, incorporating the voice characteristics of a particular user to identify what is being said.

[0078] Step 2: Real-time transcription

[0079] 1. The device sends the audio data during the meeting to the server

[0080] After the conference starts, the terminal records the user's speech in real time and transmits it intermittently to the server.

[0081] 2. The server transcribes the audio data in real time

[0082] The server transcribes the received audio data in real time, using a speech recognition model to identify each speaker while transcribing.

[0083] Step 3: Remove unwanted audio

[0084] 1. The server uses natural language processing to remove unnecessary audio

[0085] The server uses natural language processing technology to automatically remove unnecessary sounds such as "um" and "ah" from the transcribed data.

[0086] Step 4: Automatic summarization

[0087] 1. Server generates a summary using AI

[0088] The server inputs text data containing all statements into the generation AI, which automatically extracts the key points of the meeting and creates a summary.

[0089] The created summary is saved as minutes on the server.

[0090] Step 5: Sending meeting notes

[0091] 1. The server sends the minutes to all participants

[0092] The server automatically sends the saved summary to the registered email addresses of all conference participants.

[0093] Step 6: Integrate with a chatbot

[0094] 1. The server provides database access to the chatbot

[0095] The server provides the chatbot with access to a database of past meeting minutes and configures it to return data in response to queries.

[0096] 2. Chatbots handle user inquiries

[0097] The chatbot receives a search request from the user, retrieves and displays relevant meeting minutes data.

[0098] Terminal handling

[0099] Step 1: Registering audio data

[0100] 1. The user records the audio and saves it on the device.

[0101] The user records and saves their own voice on the device.

[0102] 2. The device uploads the audio data to the server

[0103] The device sends the recorded audio data to the server.

[0104] Step 2: Sending audio data during a meeting

[0105] 1. The device records what the user says and sends it to the server

[0106] The device records the user's remarks in real time during the meeting and transmits them intermittently to the server.

[0107] Step 3: Review the meeting notes

[0108] 1. Display the minutes received by the device

[0109] After the conference ends, the terminal receives the minutes sent from the server and displays them to the user.

[0110] User Action

[0111] Step 1: Registering audio data

[0112] 1. The user records the audio and saves it on the device.

[0113] The user records and saves their own voice.

[0114] 2. The user uploads the audio data to the server using their device.

[0115] The user uploads the audio data recorded using the terminal to the server.

[0116] Step 2: Speaking during the meeting

[0117] 1. The user speaks as usual

[0118] The user speaks normally in the conference.

[0119] 2. Users use devices to record their speech.

[0120] When a user speaks, the device automatically starts recording and sends it to the server.

[0121] Step 3: Review the meeting notes

[0122] 1. User checks the minutes

[0123] After the meeting ends, the user checks the minutes that have arrived on the terminal, which displays the minutes received from the server.

[0124] Step 4: Integrate with a chatbot

[0125] 1. User searches and checks with chatbot

[0126] Users can use the chatbot to search for past minutes or specific statements and check the results.

[0127] The chatbot extracts the relevant information from the database on the server and displays it to the user.

[0128] Example 1

[0129] 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."

[0130] Conventional meeting minutes creation requires manual recording, which is time-consuming and labor-intensive, and it is difficult to identify speakers and extract important content. Furthermore, sharing minutes and searching past records is also time-consuming, making efficient information management difficult. This has led to problems such as a decline in meeting transparency and work efficiency.

[0131] 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.

[0132] In this invention, the server includes means for registering and learning the voice data of each participant in advance, means for transcribing the voice in real time during the conference, means for removing unnecessary voice from the transcribed data, means for automatically summarizing the conference content, means for automatically sending the summary data to the participants, and means for searching past minutes in cooperation with a chatbot. This automates the creation of meeting minutes, enables quick and accurate identification of speakers and extraction of important content, and improves the transparency and work efficiency of meetings.

[0133] "Audio data" refers to recorded audio information in digital or analog format.

[0134] "Learning" is the process by which a speech recognition model analyzes input speech data and makes it possible to identify the characteristics of the speaker and the content of the speech.

[0135] "Transcription" is the process of analyzing audio data and converting its content into text format data.

[0136] "Unnecessary audio" refers to meaningless, redundant audio such as "um" and "ah," which detracts from the conciseness of the minutes.

[0137] A "summary" is text data that analyzes the contents of a meeting and succinctly summarizes the main points and decisions made.

[0138] "Auto-send" is the process of automatically sending specific data to a pre-defined email address or other address.

[0139] A "chatbot" is a software application that uses natural language processing to automatically respond to user questions.

[0140] A "minutes" is a document that records what was said and what was decided at a meeting.

[0141] "Speaker" refers to each participant who speaks during a conference.

[0142] A "database" is an electromagnetic storage device for efficiently storing, managing, and retrieving specific data.

[0143] "Email" is a means of communication for sending and receiving text and files over the Internet.

[0144] MODE FOR CARRYING OUT THE INVENTION

[0145] This invention is a system for automatically generating and efficiently managing meeting minutes. This system registers and learns the voice data of each participant in advance, identifies who spoke, transcribes the meeting in real time, removes unnecessary sounds, and summarizes the meeting content. It also automatically sends the summary data to participants after the meeting ends, and can also search for past minutes in conjunction with a chatbot.

[0146] Audio data registration and training

[0147] First, the user records their own voice on their device using a standard recording application (e.g., Voice Recorder). This voice data is then uploaded to a server. The server then analyzes the uploaded voice data using the Google Cloud Speech-to-Text API or similar to learn the characteristics of each speaker. This makes it easier to identify speakers during a meeting.

[0148] Transcription of audio data

[0149] When a meeting starts, the device records the user's speech in real time. This recording data is then sent to the server. The server then uses the Google Cloud Speech-to-Text API to convert the received audio data into text in real time. At the same time, it identifies the speaker using a pre-trained model.

[0150] Removing unwanted audio

[0151] The server removes unnecessary sounds such as "um" and "ah" from the transcribed text using natural language processing libraries such as the Natural Language Toolkit (NLTK) and SpaCy. This process makes the minutes easier to read.

[0152] Automatic summary generation

[0153] Based on all the transcribed text data, a generative AI model (e.g., OpenAI GPT-4) is used to extract the main points of the meeting. The server sets prompts for summary generation, and the main points are automatically summarized based on these prompts.

[0154] Example prompt sentence:

[0155] "Please summarize the meeting and include key topics and decisions."

[0156] Sending meeting notes

[0157] After the meeting ends, the server automatically sends the generated summary to the participants' email addresses using email sending services such as SendGrid or Amazon SES, allowing participants to check the summarized minutes immediately after the meeting ends.

[0158] Collaboration with chatbots

[0159] The server works with a chatbot (e.g., Dialogflow, Microsoft Bot Framework) to respond to requests from users to search for meeting minutes. When a user searches for past meeting minutes or specific statements, the chatbot retrieves the necessary information from the database on the server and provides it to the user.

[0160] These processes automate the creation of meeting minutes, allowing for quick and accurate identification of speakers and extraction of important content, significantly improving meeting transparency and work efficiency.

[0161] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0162] Step 1: Register and train audio data

[0163] Input: User recorded voice data

[0164] How it works: The user records their own voice using a standard recording application, and the recording is uploaded from the device to the server.

[0165] Data processing: The server converts the received voice data into text data using the Google Cloud Speech-to-Text API, etc. Based on this text data, a speech recognition model is used to learn the speaker's characteristics.

[0166] Output: A database containing speaker characteristics data

[0167] Step 2: Record and send audio during the meeting

[0168] Input: The voice that the user speaks during the meeting.

[0169] Specific operation: When the conference starts, the terminal records the user's speech in real time, and the recorded data is sent to the server sequentially.

[0170] Data processing: The device transfers the recorded audio data to the server via an audio data streaming service.

[0171] Output: Real-time audio data sent to the server

[0172] Step 3: Transcribe the audio data

[0173] Input: Real-time audio data received by the server

[0174] Specific operation: The server converts the received voice data into text data using the Google Cloud Speech-to-Text API.

[0175] Data processing: Based on real-time speech recognition, the audio data is transcribed into corresponding text data, while simultaneously identifying the speaker.

[0176] Output: Real-time text data with speaker identification

[0177] Step 4: Remove unwanted audio

[0178] Input: Transcribed real-time text data

[0179] Specific operation: The server uses the Natural Language Toolkit (NLTK) and SpaCy libraries to detect and remove redundant sounds such as "eh" and "um" from text data.

[0180] Data processing: Natural language processing techniques are used to filter out redundant data and generate clean text data.

[0181] Output: Transcribed text data with redundant speech removed

[0182] Step 5: Automatically generate summaries

[0183] Input: Fairly formatted text data

[0184] How it works: The server uses a generative AI model (e.g., OpenAI GPT-4) to extract key points from the input, typed text data.

[0185] Data processing: The summarization process uses prompts, and the AI ​​model extracts key points from the text data and generates a summary.

[0186] Output: Summarized minutes data

[0187] Example prompt sentence:

[0188] "Please summarize the meeting and include key topics and decisions."

[0189] Step 6: Sending meeting notes

[0190] Input: Summarized minutes data

[0191] Specific operation: The server automatically sends summarized meeting minutes data to participants' email addresses using SendGrid or Amazon SES.

[0192] Data processing: The summarized minutes data is inserted into the body of an email and sent to everyone based on the appropriate participant list.

[0193] Output: Summary minutes sent to participants' email addresses

[0194] Step 7: Integrate with a chatbot

[0195] Input: A request to search minutes from the user

[0196] Specific behavior: A user sends a request to search for past minutes or specific statements through a chatbot (e.g., Dialogflow, Microsoft Bot Framework).

[0197] Data processing: The server receives a request from the chatbot, searches for relevant information in the database, and extracts the required data.

[0198] Output: Providing search results to the user

[0199] The above are the processing steps and their specific operations of the present invention. The steps work in conjunction with each other to create meeting minutes efficiently and accurately.

[0200] (Application example 1)

[0201] 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."

[0202] In today's factory operations, it is important to accurately record and manage the details and progress of tasks performed by workers and robots. However, manual recording and progress management takes time and effort, and is prone to human error. Searching past work records also requires a great deal of effort. Furthermore, it is difficult to remove unnecessary data from work reports and extract the key points. An efficient system to solve this problem is needed.

[0203] 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.

[0204] In this invention, the server includes means for registering and learning voice data of each worker in advance, means for transcribing voice in real time during work, means for removing unnecessary voice from the transcribed data, means for automatically summarizing work content, means for automatically sending the summary data to a manager, and means for searching past work records in cooperation with a chatbot. This makes it possible to automate work recording and management in factories, improving work efficiency and accuracy.

[0205] "Audio data" refers to verbal information uttered by a worker or speaker that has been recorded in digital form.

[0206] "Means for learning" refers to the process by which the server builds a speech recognition model based on registered speech data, enabling it to identify the voice of a specific speaker.

[0207] "Transcribing means" refers to speech recognition technology for converting audio data into text format.

[0208] The "means for removing unwanted audio" is a process that automatically removes unwanted parts such as "um" and "ah" from the transcribed text.

[0209] "Summarization" is a technique for extracting the main content from transcribed data and summarizing it briefly.

[0210] "Means for automatic sending" refers to a process in which the server automatically sends the generated summary data to a specified email address.

[0211] A "chatbot" is software that automatically converses with users and provides them with information in response to their inquiries.

[0212] The "means for searching past work records" is a process for searching past work data stored on the server based on a user request and extracting necessary information.

[0213] MODE FOR CARRYING OUT THE INVENTION

[0214] This invention is a system for improving the efficiency of work recording and management in factories. This system records the content and progress of work performed by workers and robots in real time, automatically summarizing the data and sending it to managers. It also has a function for searching past work records.

[0215] 1. Registering and learning audio data

[0216] The server receives the voice data uploaded in advance by each worker or robot and trains the speech recognition model. This builds a database of each worker's or robot's characteristic voice, making it possible to accurately identify who is speaking during work. This step uses Python and the speech_recognition library.

[0217] 2. Transcription of audio data

[0218] When a task begins, the device records what is being said and what is being done, and sends it to the server. The server receives the audio data in real time and transcribes it. With the help of a speech recognition model, the transcription is performed while identifying the speaker and the task. The speech_recognition library is also used in this step.

[0219] 3. Remove unwanted audio

[0220] The server uses natural language processing technology to automatically remove unnecessary sounds such as "um" and "ah" from the recorded text, making the work log more concise and easy to read. This step utilizes Python's text processing capabilities.

[0221] 4. Automatic summary generation

[0222] All speech and work content is transcribed and input into a generative AI model, which automatically extracts key points and creates a summary. The summarized work record is then saved on a server. This step uses Hugging Face's transformers module, and the summarize function is used to generate the summary.

[0223] 5. Sending work records

[0224] After the work is completed, the server automatically sends the generated summary to the administrator's email address. This function allows the administrator to check the work record immediately after the work is completed. This step uses the Python smtplib library.

[0225] 6. Chatbot integration

[0226] The server works with the chatbot to search past work records in response to user inquiries and provide the necessary information. This step uses the chatbot software and database search function.

[0227] Specific examples

[0228] For example, a factory robot may report work such as "Welding process 2 has been completed" or "Part A is missing." The system transcribes these statements in real time, removes unnecessary parts, generates a summary, and sends it to the manager by email after the work is completed. It also makes it easy to search past work records through the chatbot.

[0229] Prompt Sentence Examples

[0230] Collaboration with Edge AI Sensors:

[0231] Q: How is the data collected by the AI ​​sensors analyzed?

[0232] Answer: Data collected by the AI ​​sensor is transcribed in real time through a speech recognition module, unnecessary noise is automatically removed, and a summary is generated.

[0233] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0234] Step 1:

[0235] The server pre-registers the voice data of each worker and robot and trains a voice recognition model. The input for this step is the voice data of the worker and robot. The server uses this voice data to build a voice recognition model and processes the data so that it can identify the voice of each worker. The output is a voice feature database for each worker.

[0236] Step 2:

[0237] The terminal records the voices emitted by workers and robots while working and sends the voice data to the server in real time. The input for this step is the voices spoken during work. The terminal records this and sends it to the server. The server converts the received voice data into transcription data. The output is transcription data that converts the voices into text.

[0238] Step 3:

[0239] The server removes unwanted speech from the transcription data. The input for this step is the transcribed text data. The server uses natural language processing techniques to identify unwanted parts, such as "um" and "ah," and removes them from the text. The output is clean text data with the unwanted speech removed.

[0240] Step 4:

[0241] The server inputs the clean text data into a generative AI model to automatically extract the key points of the work content and create a summary. The input for this step is the clean text data. The server uses the generative AI model to extract important points from the text data and create a short summary. The output is a short, summarized text data.

[0242] Step 5:

[0243] The server automatically sends the generated summary to the administrator's email address. The inputs to this step are the summarized text data and the administrator's email address. The server combines these data using the SMTP protocol and sends an email to the administrator. The output is the summary data sent to the administrator's mailbox.

[0244] Step 6:

[0245] When a user searches for past work records through a chatbot, the server searches the database based on the user's inquiry and provides the required information. The input to this step is the user's search query. The server works with the chatbot to extract the relevant work records from the database and present them to the user. The output is the past work record data provided to the user.

[0246] 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.

[0247] This invention combines a conventional meeting recording system with an emotion engine to grasp the emotional state of each participant and create and manage advanced minutes. This system automatically generates meeting minutes while recognizing participants' emotions in real time, which can be used to summarize the meeting and for subsequent follow-up.

[0248] Below, each component of this system and its processing will be specifically explained.

[0249] Server Processing

[0250] 1. Registering and learning audio data

[0251] The server receives voice data uploaded by users in advance and trains a voice recognition model. At the same time, the emotion engine also learns the user's emotional state from the voice data. This allows the model to reflect the emotional characteristics of each participant as well as their voice, making it possible to identify emotional changes during the meeting.

[0252] 2. Real-time transcription and sentiment analysis of audio data

[0253] When a meeting begins, the device records what is being said and sends it to the server. The server receives the audio data in real time, transcribes it, and uses an emotion engine to analyze the emotional state of each utterance. The transcription and emotion analysis are performed while identifying the speaker using a speech recognition model.

[0254] 3. Remove unwanted audio

[0255] The server uses natural language processing technology to automatically remove unwanted sounds such as "um" and "ah" from the recorded text, and the emotion engine adjusts the sound so that the unwanted sounds do not affect the emotion analysis.

[0256] 4. Automatic summary generation

[0257] The server inputs the text data, including all statements, and the results of sentiment analysis into a generation AI, which automatically extracts the key points of the meeting and creates a summary. Emotional data is also taken into consideration, and statements that caused particularly large emotional changes are reflected in the summary. The created summary is saved on the server as minutes of the meeting.

[0258] 5. Sending meeting minutes

[0259] After the meeting, the server automatically sends the generated summary to the participants' email addresses, including emotional data, so participants can share the atmosphere of the meeting and any notable emotional reactions.

[0260] 6. Chatbot integration

[0261] The server works in conjunction with the chatbot to search past meeting minutes in response to user inquiries and provide the necessary information. Searches can also be made including emotional data, making it possible to investigate specific statements and changes in emotion.

[0262] Terminal handling

[0263] 1. Registering audio data

[0264] Users record their own voice and emotions using their devices and upload the voice data to the server, which allows the server to learn the voice and emotion data of each participant in advance.

[0265] 2. Sending voice and emotion data during a meeting

[0266] When a meeting begins, the device records the user's speech and emotions in real time and sends them to the server, which then instantly transcribes the received data and performs sentiment analysis.

[0267] 3. Checking meeting notes

[0268] After the meeting ends, the device receives the minutes sent from the server and displays them to the user. Emotional data is also displayed, allowing the user to get a more detailed understanding of the atmosphere of the meeting.

[0269] User Action

[0270] 1. Registering audio data

[0271] Users use their devices to record and register their own voice and emotions, and this data is uploaded to a server and used as training data for the voice recognition model and emotion engine.

[0272] 2. Record statements and emotions during meetings

[0273] Users can participate in a meeting as usual, make comments, and record their emotions. The device records comments and emotions in real time and sends them to the server, so no special operation is required from the user.

[0274] 3. Checking meeting notes

[0275] After the meeting, the user can check the minutes received from their device. The system's automatic summarization and sentiment analysis allow the user to quickly grasp the main points of the meeting.

[0276] 4. Integration with chatbots

[0277] Users can use the chatbot to search past meeting minutes and specific statements, and can also search for emotional data, allowing them to check specific statements and the emotional state of those statements.

[0278] This system not only automates the creation of meeting minutes, but also captures the emotional state of participants in real time and reflects it in the minutes, thereby increasing the transparency and effectiveness of meetings and promoting deeper understanding and cooperation.

[0279] The processing flow will be explained below.

[0280] Server Processing

[0281] Step 1: Register and train audio data

[0282] 1. The server receives the audio data

[0283] The user uploads the audio data recorded using the terminal to the server.

[0284] 2. The server trains the speech recognition model

[0285] The server trains a speech recognition model based on the received voice data, thereby identifying the characteristics of each participant's voice.

[0286] 3. The server learns the emotion engine

[0287] The server uses an emotion engine to learn the user's emotional state from the voice data, making it possible to understand changes in emotions that accompany speech.

[0288] Step 2: Real-time transcription and sentiment analysis

[0289] 1. The server receives the audio data in real time.

[0290] During the meeting, the device records the user's remarks in real time and sends them to the server.

[0291] 2. The server performs the transcription

[0292] The server transcribes the received audio data in real time and identifies the speaker.

[0293] 3. The server analyzes the emotional data

[0294] As the server transcribes, it uses an emotion engine to analyze the emotional state of each utterance, determining, for example, whether the speaker is excited or calm.

[0295] Step 3: Remove unwanted audio

[0296] 1. The server removes unnecessary audio

[0297] The server uses natural language processing technology to remove unnecessary sounds such as "eh" and "um" from the transcription data, and sentiment analysis is also adjusted at the same time.

[0298] Step 4: Automatic summarization

[0299] 1. The server extracts the main points

[0300] The server uses AI to automatically extract key points from meetings based on transcripts containing emotional data, taking into account statements that showed significant emotional changes.

[0301] 2. The server stores the summary

[0302] The server generates a summary and saves it as a transcript.

[0303] Step 5: Sending meeting notes

[0304] 1. The server sends the minutes

[0305] The server automatically sends the minutes, including emotional data, to the email addresses of all participants.

[0306] Step 6: Integrate with a chatbot

[0307] 1. The server provides database access to the chatbot

[0308] The server grants the chatbot access to the conference database.

[0309] 2. Chatbots handle inquiries

[0310] The chatbot receives a search request from the user, retrieves the necessary data from the server, and responds.

[0311] Terminal handling

[0312] Step 1: Registering audio data

[0313] 1. The device records audio

[0314] The user records their own voice on the device.

[0315] 2. The device sends the voice data to the server

[0316] The device uploads the recorded audio data to the server.

[0317] Step 2: Sending audio data during a meeting

[0318] 1. The device records what you say and sends it to the server

[0319] The device records what is said during the meeting in real time and sends it to the server.

[0320] Step 3: Sending emotion data

[0321] 1. The device sends emotional data

[0322] The device uses an emotion engine to analyze emotions based on the recorded voice data and sends the data to a server.

[0323] Step 4: Review the meeting notes

[0324] 1. The device receives the minutes

[0325] After the meeting ends, the terminal receives the minutes sent from the server.

[0326] 2. The device displays the minutes.

[0327] The terminal displays the received minutes to the user.

[0328] User Action

[0329] Step 1: Registering audio data

[0330] 1. The user records the audio and saves it on the device.

[0331] The user records their own voice on the device.

[0332] 2. The user uploads the audio data to the server.

[0333] Upload the recorded audio data to the server.

[0334] Step 2: Speaking during the meeting

[0335] 1. User speaks

[0336] A user speaks in a conference.

[0337] 2. The user checks the recording

[0338] The user checks whether recording is working properly on the device.

[0339] Step 3: Review the meeting notes

[0340] 1. User checks the minutes

[0341] After the meeting ends, the user checks the minutes received from the terminal.

[0342] Step 4: Integrate with a chatbot

[0343] 1. Users interact with chatbots

[0344] A user uses the chatbot to search for past meeting minutes or specific statements.

[0345] 2. User checks the results

[0346] The user checks the results provided by the chatbot.

[0347] Example 2

[0348] 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."

[0349] Conventional meeting recording systems require manual creation of meeting minutes, which is time-consuming and labor-intensive. Furthermore, it is difficult to record the emotional state of each participant during the meeting, making it difficult to grasp the atmosphere of the meeting in detail. Therefore, there is a need to improve meeting efficiency and ensure transparency.

[0350] 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.

[0351] In this invention, the server includes means for registering and learning the voice data of each participant in advance, means for transcribing the voice in real time during the conference, means for removing unnecessary voice from the transcribed data, means for inputting the transcribed data and emotion data into a generative AI model to automatically summarize the conference content, means for automatically sending the summary data to the participants, means for searching past minutes in cooperation with a chatbot, and means for analyzing and saving emotion data in real time. This automates the creation of meeting minutes and enables detailed meeting records that take emotion data into account.

[0352] "Audio data" means digitally recorded information of a participant's speech.

[0353] "Emotion data" is information that indicates the emotional state of each speaker, analyzed from the voice data.

[0354] A "speech recognition model" is a machine learning or artificial intelligence algorithm for converting speech data into text data.

[0355] An "emotion engine" is an algorithm or system that analyzes audio and text data to identify emotional states.

[0356] "Transcription data" refers to data that has been converted from audio data into text format.

[0357] A "generative artificial intelligence model" is a machine learning or artificial intelligence model that generates new data based on input data.

[0358] "Unnecessary voice" refers to speech that is not related to the significance or content of the meeting (e.g., "umm," "ah," etc.).

[0359] "Summary data" is text data that extracts important points from all statements made at a meeting and summarizes them concisely.

[0360] A "chatbot" is an automated program that interacts with a user to provide or process information.

[0361] "Real-time" refers to the time frame in which data is generated and processed immediately, without delay.

[0362] A "database" is a system for efficiently storing, searching, and managing large amounts of data.

[0363] This invention is a system that automates the creation of meeting minutes and grasps the emotional states of participants in real time, thereby improving the transparency and effectiveness of meetings. This system consists of three main components: a server, a terminal, and a user, and is specifically implemented as follows.

[0364] First, users record their own voice data using their devices and upload it to the server. At this time, the recorded voice data is analyzed using an emotion engine to determine each user's emotional state. The server then trains a voice recognition model based on the received voice data. For example, it uses Google Cloud Speech-to-Text to convert the voice data into text and uses Hume AI to learn emotional characteristics.

[0365] When a meeting starts, the device records participants' speech in real time and sends it to the server. The server converts the received audio data into text in real time and simultaneously analyzes the emotional state of the speech using an emotion engine. This process uses Amazon Transcribe, for example, and stores the generated text data and emotion data in a database.

[0366] The server also uses natural language processing techniques to automatically remove unwanted sounds like "um" and "ah" from the text data. For example, it uses the SpaCy library to analyze the text and remove unnecessary parts. The clean text data is then stored in a database.

[0367] All spoken text and sentiment data are fed into a generative AI model to extract key points from the meeting and automatically generate a summary. OpenAI's GPT-4 can be used to generate summaries, using prompts such as "Please consider all spoken data from the meeting and its sentiment data, extract key points, and generate a summary of the meeting."

[0368] After the meeting, the server automatically sends the generated summary to participants, for example via email using the SendGrid API, which also includes sentiment data, allowing participants to understand the atmosphere of the meeting and any notable emotional reactions.

[0369] In addition, the server has the ability to search past meeting minutes in cooperation with chatbots. Users can use chatbots to efficiently search for meeting minutes containing specific statements or emotional states. For example, a chatbot can be built using Dialogflow and data can be searched using Elasticsearch.

[0370] The system automates meeting minutes creation and provides detailed meeting records that take emotional data into account, improving the transparency and effectiveness of meeting content and fostering deeper understanding and collaboration.

[0371] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0372] Step 1:

[0373] The user records their own voice data on the device and uploads it to the server.

[0374] Input: User recorded voice data

[0375] Specific operation: The user records their own voice using the recording app on their device and uploads the file to the server, where the uploaded data is stored.

[0376] Output: Audio data stored on the server

[0377] Step 2:

[0378] The server receives the speech data and trains a speech recognition model.

[0379] Input: Audio data uploaded from the device

[0380] How it works: The server converts voice data into text using Google Cloud Speech-to-Text or similar, and then uses an emotion engine such as Hume AI to analyze the emotional state from the voice data and saves the learning results in a database.

[0381] Output: Text data and emotion data corresponding to the audio data

[0382] Step 3:

[0383] The device records what is said during the meeting and sends it to the server in real time.

[0384] Input: Audio data during the meeting

[0385] Specific operation: The device records speech in real time and sends it to the server using WebSocket or similar.

[0386] Output: Audio data sent to the server in real time

[0387] Step 4:

[0388] The server transcribes the audio data in real time and analyzes the emotional state.

[0389] Input: Real-time audio data

[0390] How it works: The server uses Amazon Transcribe to convert the voice data into text, and simultaneously uses the emotion engine to analyze the emotional state. This data is then stored in a database in real time.

[0391] Output: Real-time text and sentiment data

[0392] Step 5:

[0393] The server removes unwanted audio from the transcription data.

[0394] Input: Real-time generated text data

[0395] Specific operation: The server uses the SpaCy library to analyze the text data and remove unnecessary sounds such as "um" and "ah." The cleaned text data is then restored to the database.

[0396] Output: Text data with unnecessary audio removed

[0397] Step 6:

[0398] The server uses a generative AI model to automatically summarize the key points of the meeting.

[0399] Input: Clean text data and sentiment data

[0400] Specific operation: The server uses OpenAI GPT-4 to generate a meeting summary from text data and emotion data, using the prompt "Consider all speech data and their emotion data from the meeting, extract key points, and generate a meeting summary." The generated summary is stored in a database.

[0401] Output: Summarized meeting data

[0402] Step 7:

[0403] After the conference ends, the server automatically sends summary data to the participants.

[0404] Input: Summarized meeting data

[0405] Specific operation: The server uses the SendGrid API to send the saved summary data to the participant's email address.

[0406] Output: Summary data sent to participants

[0407] Step 8:

[0408] The server works in conjunction with the chatbot to search for past minutes.

[0409] Input: User's search request

[0410] Specific operation: The server uses Dialogflow to build a chatbot, searches the database using Elasticsearch, and returns the search results and sentiment data to the user.

[0411] Output: Past meeting minutes data provided to the user

[0412] Through these processing steps, meeting minutes are automated and detailed meeting records are generated that take emotional data into account. This system improves the transparency and effectiveness of meetings, promoting deeper understanding and cooperation.

[0413] (Application example 2)

[0414] 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."

[0415] In conventional meeting recording systems, simply converting speech to text makes it difficult to accurately grasp the nuances of what is being said and the emotional state of the participants.In addition, in regular and emergency meetings within a factory, it is important to understand not only the main points of the meeting but also the emotional state of the participants, but previous systems were unable to perform real-time emotional analysis.

[0416] The identification processing 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 registering and learning the voice data of each participant in advance, means for transcribing the voice in real time during the conference, means for removing unnecessary voice from the transcribed data, means for automatically summarizing the conference content, means for analyzing the emotional state, means for automatically sending summary data to the participants, and means for searching past minutes in cooperation with a chatbot. This makes it possible to grasp the emotional state of the participants in real time in addition to summarizing the conference content, and to create and manage highly accurate minutes.

[0417] "Voice data of each participant" refers to voice information emitted by each participant who participates in a conference registered in the system.

[0418] The "means for pre-registration and learning" refers to a function that allows characteristic voice data of each participant to be input into the system before the start of the meeting, and for the voice recognition model to learn the voice characteristics of each participant based on that data.

[0419] "Means for transcribing audio in real time" is a function that instantly converts audio spoken during a meeting into text data.

[0420] "Means for removing unnecessary sounds from transcribed data" is a function for automatically removing meaningless sounds such as "eh" and "um" from transcribed text data.

[0421] The "means for automatically summarizing meeting content" is a function that analyzes what is said during a meeting, extracts important points, and generates a concise summary.

[0422] The "means for analyzing emotional states" is a function that analyzes the emotions contained in each statement in real time based on transcribed text data.

[0423] The "means for automatically transmitting summary data to participants" is a function for automatically transmitting the generated summary data to the communication addresses of all participants after the conference has ended.

[0424] "Means for searching past minutes in cooperation with a chatbot" refers to a chatbot function that searches past minutes or specific statements within them and provides relevant information in response to inquiries from users after a meeting.

[0425] This invention is a system that creates minutes and grasps the emotional state of participants in factory meetings. This system consists of three main components: a server, a terminal, and a user. We will explain how each component works.

[0426] Server Action:

[0427] 1. Register and train the audio data:

[0428] The server receives voice data uploaded by users in advance and trains a voice recognition model. The voice data contains the characteristic voice information of each participant, which improves the accuracy of voice recognition during the meeting. At the same time, it also learns the emotional state of each participant using an emotion engine.

[0429] 2. Real-time transcription and sentiment analysis:

[0430] When a meeting begins, the device records the audio and sends it to the server, which receives it in real time, transcribes it, and then uses an emotion engine to analyze the emotional state of each comment.

[0431] 3. Remove unwanted audio:

[0432] Using natural language processing technology, unnecessary sounds such as "um" and "ah" are automatically removed from the recorded text data, improving the quality of the transcription data.

[0433] 4. Automatic summary generation:

[0434] Based on the results of speech data and sentiment analysis, a generative AI model is used to automatically extract key points from meetings and generate summaries, including statements that elicited particularly strong sentiment.

[0435] 5. Sending minutes:

[0436] After the meeting ends, the server automatically sends the generated summary to the communication addresses of the participants, allowing the contents of the meeting to be shared quickly.

[0437] 6. Chatbot integration:

[0438] The server works in conjunction with the chatbot to search past meeting minutes in response to user questions and provide the necessary information. Emotional data can also be searched, allowing users to check specific statements and their emotional state at the time.

[0439] Terminal handling:

[0440] 1. Audio data registration:

[0441] Users use their devices to record and register their own voice and emotions, and this data is uploaded to a server and used to train the voice recognition model and emotion engine.

[0442] 2. Sending voice and emotion data:

[0443] During a meeting, the device records the user's speech and emotions in real time and transmits them to the server, which then uses the received data for immediate transcription and sentiment analysis.

[0444] 3. Check the minutes:

[0445] After the meeting ends, the device receives the minutes sent from the server and displays them to the user, allowing the user to check the details of the meeting and the emotional state at the time.

[0446] User Action:

[0447] 1. Audio data registration:

[0448] Users use their devices to record and register their own voice and emotions, which allows the voice recognition model and emotion engine to work more accurately.

[0449] 2. Record statements and emotions during the meeting:

[0450] Users simply join a meeting as usual and record their comments and emotions on their device, with no special operations required.

[0451] 3. Check the minutes:

[0452] After the meeting, users can review the minutes received from their devices and quickly grasp the main points by utilizing the system's automatic summarization and sentiment analysis.

[0453] 4. Chatbot integration:

[0454] Through the chatbot, users can search past meeting minutes and specific statements. They can also search for emotion data, allowing them to check specific statements and the emotional state of those statements.

[0455] Examples:

[0456] During regular meetings held within a factory, an assistant robot summarizes the meeting and analyzes sentiment. The minutes are then sent to participants' mailboxes after the meeting. For example, if a participant asks the chatbot, "What were the most common opinions expressed in last month's meeting?", the robot can provide a summary along with past sentiment.

[0457] Example prompt sentence:

[0458] "What were the most common opinions and comments that generated a lot of emotion at last month's meeting?"

[0459] As described above, the system of the present invention can significantly improve the efficiency of meetings within a factory.

[0460] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0461] Step 1: Register and train audio data

[0462] Users use their devices to record their own voice and emotions and upload them to the server. The server then trains the received voice data based on a speech recognition model. The input is the voice data and emotion data from the users, and the server learns from this to understand the voice and emotion features of each participant. The output is the speech recognition model after training is complete.

[0463] Step 2: Real-time transcription and sentiment analysis

[0464] The device records the audio during the meeting in real time and sends this audio data to a server, which then instantly transcribes the audio data and performs sentiment analysis using a generative AI model.The input is the audio data from the meeting, and the server outputs text data and the emotional state of each utterance.

[0465] Step 3: Remove unwanted audio

[0466] The server uses natural language processing technology to automatically remove unnecessary sounds such as "eh" and "um" from the generated text data. The input is transcribed text data, and unnecessary sounds are removed as part of the data processing. The output is clear transcribed text data.

[0467] Step 4: Automatically generate a summary

[0468] The server uses the transcription data and the results of sentiment analysis to automatically summarize the meeting content based on a generative AI model. The input is clear transcription text data and sentiment data, which are then processed into a summary using the generative AI model. The output is the summarized text data.

[0469] Step 5: Sending meeting notes

[0470] After the conference ends, the server automatically sends the generated summary data to the participants' communication addresses. The input is the summarized text data and the participants' communication addresses, and the server calculates and sends this as data in the form of email. The output is the email sent to the participants.

[0471] Step 6: Integrate with a chatbot

[0472] The user uses a chatbot from their device to search for past minutes or specific statements. The server searches for past minutes and emotion data via the chatbot and provides them to the user. The input is a search query (prompt sentence) from the user, and the server calculates and outputs the appropriate minutes and emotion data from the database. The output is the search results provided to the user.

[0473] 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.

[0474] 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.

[0475] 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.

[0476] [Second embodiment]

[0477] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0478] 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.

[0479] 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).

[0480] 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.

[0481] 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.

[0482] 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).

[0483] 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.

[0484] 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.

[0485] 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.

[0486] 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.

[0487] 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.

[0488] 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."

[0489] This invention is a system for automatically generating and efficiently managing meeting minutes. This system pre-registers and learns from the voice data of each participant to accurately identify who spoke, transcribes the voice in real time during the meeting, removes unnecessary voice, summarizes the meeting content, and automatically sends the summary data to participants after the meeting. It is also possible to search past meeting minutes in conjunction with a chatbot.

[0490] Below, each component of this system and its processing will be specifically explained.

[0491] Server Processing

[0492] 1. Registering and learning audio data

[0493] The server receives voice data previously uploaded by users and trains a speech recognition model, building a database of each participant's distinctive voice, enabling accurate identification of who is speaking during a meeting.

[0494] 2. Transcription of audio data

[0495] When a meeting begins, the device records what is being said and sends it to the server, which receives the audio data in real time and transcribes it. With the help of a speech recognition model, the transcription is done while identifying the speaker.

[0496] 3. Remove unwanted audio

[0497] The server uses natural language processing technology to automatically remove unnecessary sounds such as "um" and "ah" from the recorded text, making the minutes more concise and easier to read.

[0498] 4. Automatic summary generation

[0499] The server inputs the transcribed data of all statements into the generation AI, which automatically extracts the key points of the meeting and creates a summary. The summarized minutes are then stored on the server.

[0500] 5. Sending meeting minutes

[0501] After the meeting, the server automatically sends the generated summary to the participants' email addresses, allowing participants to check the minutes immediately after the meeting ends.

[0502] 6. Chatbot integration

[0503] The server works in conjunction with the chatbot to search past minutes in response to user inquiries and provide the necessary information.

[0504] Terminal handling

[0505] 1. Registering audio data

[0506] Users record their own voice using their devices and upload the voice data to the server, which allows the server to learn the voice data of each participant in advance.

[0507] 2. Sending audio data during a meeting

[0508] When the meeting begins, the device records the user's speech in real time and sends it to the server, which receives it and immediately transcribes it.

[0509] 3. Checking meeting notes

[0510] After the meeting ends, the terminal receives the minutes sent from the server and displays them to the user, allowing the user to quickly grasp the main points of the meeting.

[0511] User Action

[0512] 1. Registering audio data

[0513] Users use their devices to record and register their own voice, which is then uploaded to a server and used as training data for the voice recognition model.

[0514] 2. Statements made during meetings

[0515] Users can join a conference and speak as usual. The device records what they say in real time and sends it to the server, so no special operations are required on the part of the user.

[0516] 3. Checking meeting notes

[0517] After the meeting, users can check the minutes data received from their devices. The system's automatic summary function allows users to quickly grasp the main points of the meeting.

[0518] 4. Integration with chatbots

[0519] Users can use the chatbot to easily search for and check past meeting minutes or specific comments. The chatbot extracts the relevant information from a database on the server and presents it to the user.

[0520] By introducing this system, the creation of meeting minutes will be automated, significantly improving work efficiency. It also identifies speakers and removes unnecessary audio, improving the quality of meeting minutes. This reduces disputes over "said" and "unsaid," making it possible to run meetings with greater transparency.

[0521] The processing flow will be explained below.

[0522] Server Processing

[0523] Step 1: Register and train audio data

[0524] 1. The user records audio data and saves it on the device.

[0525] The user records their own voice and saves it on the device.

[0526] 2. The device uploads the audio data to the server

[0527] The device transmits the stored voice data to the server.

[0528] 3. The server receives the voice data and learns

[0529] The server receives the audio data and trains a speech recognition model, incorporating the voice characteristics of a particular user to identify what is being said.

[0530] Step 2: Real-time transcription

[0531] 1. The device sends the audio data during the meeting to the server

[0532] After the conference starts, the terminal records the user's speech in real time and transmits it intermittently to the server.

[0533] 2. The server transcribes the audio data in real time

[0534] The server transcribes the received audio data in real time, using a speech recognition model to identify each speaker while transcribing.

[0535] Step 3: Remove unwanted audio

[0536] 1. The server uses natural language processing to remove unnecessary audio

[0537] The server uses natural language processing technology to automatically remove unnecessary sounds such as "um" and "ah" from the transcribed data.

[0538] Step 4: Automatic summarization

[0539] 1. Server generates a summary using AI

[0540] The server inputs text data containing all statements into the generation AI, which automatically extracts the key points of the meeting and creates a summary.

[0541] The created summary is saved as minutes on the server.

[0542] Step 5: Sending meeting notes

[0543] 1. The server sends the minutes to all participants

[0544] The server automatically sends the saved summary to the registered email addresses of all conference participants.

[0545] Step 6: Integrate with a chatbot

[0546] 1. The server provides database access to the chatbot

[0547] The server provides the chatbot with access to a database of past meeting minutes and configures it to return data in response to queries.

[0548] 2. Chatbots handle user inquiries

[0549] The chatbot receives a search request from the user, retrieves and displays relevant meeting minutes data.

[0550] Terminal handling

[0551] Step 1: Registering audio data

[0552] 1. The user records the audio and saves it on the device.

[0553] The user records and saves their own voice on the device.

[0554] 2. The device uploads the audio data to the server

[0555] The device sends the recorded audio data to the server.

[0556] Step 2: Sending audio data during a meeting

[0557] 1. The device records what the user says and sends it to the server

[0558] The device records the user's remarks in real time during the meeting and transmits them intermittently to the server.

[0559] Step 3: Review the meeting notes

[0560] 1. Display the minutes received by the device

[0561] After the conference ends, the terminal receives the minutes sent from the server and displays them to the user.

[0562] User Action

[0563] Step 1: Registering audio data

[0564] 1. The user records the audio and saves it on the device.

[0565] The user records and saves their own voice.

[0566] 2. The user uploads the audio data to the server using their device.

[0567] The user uploads the audio data recorded using the terminal to the server.

[0568] Step 2: Speaking during the meeting

[0569] 1. The user speaks as usual

[0570] The user speaks normally in the conference.

[0571] 2. Users use devices to record their speech.

[0572] When a user speaks, the device automatically starts recording and sends it to the server.

[0573] Step 3: Review the meeting notes

[0574] 1. User checks the minutes

[0575] After the meeting ends, the user checks the minutes that have arrived on the terminal, which displays the minutes received from the server.

[0576] Step 4: Integrate with a chatbot

[0577] 1. User searches and checks with chatbot

[0578] Users can use the chatbot to search for past minutes or specific statements and check the results.

[0579] The chatbot extracts the relevant information from the database on the server and displays it to the user.

[0580] Example 1

[0581] 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."

[0582] Conventional meeting minutes creation requires manual recording, which is time-consuming and labor-intensive, and it is difficult to identify speakers and extract important content. Furthermore, sharing minutes and searching past records is also time-consuming, making efficient information management difficult. This has led to problems such as a decline in meeting transparency and work efficiency.

[0583] 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.

[0584] In this invention, the server includes means for registering and learning the voice data of each participant in advance, means for transcribing the voice in real time during the conference, means for removing unnecessary voice from the transcribed data, means for automatically summarizing the conference content, means for automatically sending the summary data to the participants, and means for searching past minutes in cooperation with a chatbot. This automates the creation of meeting minutes, enables quick and accurate identification of speakers and extraction of important content, and improves the transparency and work efficiency of meetings.

[0585] "Audio data" refers to recorded audio information in digital or analog format.

[0586] "Learning" is the process by which a speech recognition model analyzes input speech data and makes it possible to identify the characteristics of the speaker and the content of the speech.

[0587] "Transcription" is the process of analyzing audio data and converting its content into text format data.

[0588] "Unnecessary audio" refers to meaningless, redundant audio such as "um" and "ah," which detracts from the conciseness of the minutes.

[0589] A "summary" is text data that analyzes the contents of a meeting and succinctly summarizes the main points and decisions made.

[0590] "Auto-send" is the process of automatically sending specific data to a pre-defined email address or other address.

[0591] A "chatbot" is a software application that uses natural language processing to automatically respond to user questions.

[0592] A "minutes" is a document that records what was said and what was decided at a meeting.

[0593] "Speaker" refers to each participant who speaks during a conference.

[0594] A "database" is an electromagnetic storage device for efficiently storing, managing, and retrieving specific data.

[0595] "Email" is a means of communication for sending and receiving text and files over the Internet.

[0596] MODE FOR CARRYING OUT THE INVENTION

[0597] This invention is a system for automatically generating and efficiently managing meeting minutes. This system registers and learns the voice data of each participant in advance, identifies who spoke, transcribes the meeting in real time, removes unnecessary sounds, and summarizes the meeting content. It also automatically sends the summary data to participants after the meeting ends, and can also search for past minutes in conjunction with a chatbot.

[0598] Audio data registration and training

[0599] First, the user records their own voice on their device using a standard recording application (e.g., Voice Recorder). This voice data is then uploaded to a server. The server then analyzes the uploaded voice data using the Google Cloud Speech-to-Text API or similar to learn the characteristics of each speaker. This makes it easier to identify speakers during a meeting.

[0600] Transcription of audio data

[0601] When a meeting starts, the device records the user's speech in real time. This recording data is then sent to the server. The server then uses the Google Cloud Speech-to-Text API to convert the received audio data into text in real time. At the same time, it identifies the speaker using a pre-trained model.

[0602] Removing unwanted audio

[0603] The server removes unnecessary sounds such as "um" and "ah" from the transcribed text using natural language processing libraries such as the Natural Language Toolkit (NLTK) and SpaCy. This process makes the minutes easier to read.

[0604] Automatic summary generation

[0605] Based on all the transcribed text data, a generative AI model (e.g., OpenAI GPT-4) is used to extract the main points of the meeting. The server sets prompts for summary generation, and the main points are automatically summarized based on these prompts.

[0606] Example prompt sentence:

[0607] "Please summarize the meeting and include key topics and decisions."

[0608] Sending meeting notes

[0609] After the meeting ends, the server automatically sends the generated summary to the participants' email addresses using email sending services such as SendGrid or Amazon SES, allowing participants to check the summarized minutes immediately after the meeting ends.

[0610] Collaboration with chatbots

[0611] The server works with a chatbot (e.g., Dialogflow, Microsoft Bot Framework) to respond to requests from users to search for meeting minutes. When a user searches for past meeting minutes or specific statements, the chatbot retrieves the necessary information from the database on the server and provides it to the user.

[0612] These processes automate the creation of meeting minutes, allowing for quick and accurate identification of speakers and extraction of important content, significantly improving meeting transparency and work efficiency.

[0613] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0614] Step 1: Register and train audio data

[0615] Input: User recorded voice data

[0616] How it works: The user records their own voice using a standard recording application, and the recording is uploaded from the device to the server.

[0617] Data processing: The server converts the received voice data into text data using the Google Cloud Speech-to-Text API, etc. Based on this text data, a speech recognition model is used to learn the speaker's characteristics.

[0618] Output: A database containing speaker characteristics data

[0619] Step 2: Record and send audio during the meeting

[0620] Input: The voice that the user speaks during the meeting.

[0621] Specific operation: When the conference starts, the terminal records the user's speech in real time, and the recorded data is sent to the server sequentially.

[0622] Data processing: The device transfers the recorded audio data to the server via an audio data streaming service.

[0623] Output: Real-time audio data sent to the server

[0624] Step 3: Transcribe the audio data

[0625] Input: Real-time audio data received by the server

[0626] Specific operation: The server converts the received voice data into text data using the Google Cloud Speech-to-Text API.

[0627] Data processing: Based on real-time speech recognition, the audio data is transcribed into corresponding text data, while simultaneously identifying the speaker.

[0628] Output: Real-time text data with speaker identification

[0629] Step 4: Remove unwanted audio

[0630] Input: Transcribed real-time text data

[0631] Specific operation: The server uses the Natural Language Toolkit (NLTK) and SpaCy libraries to detect and remove redundant sounds such as "eh" and "um" from text data.

[0632] Data processing: Natural language processing techniques are used to filter out redundant data and generate clean text data.

[0633] Output: Transcribed text data with redundant speech removed

[0634] Step 5: Automatically generate summaries

[0635] Input: Fairly formatted text data

[0636] How it works: The server uses a generative AI model (e.g., OpenAI GPT-4) to extract key points from the input, typed text data.

[0637] Data processing: The summarization process uses prompts, and the AI ​​model extracts key points from the text data and generates a summary.

[0638] Output: Summarized minutes data

[0639] Example prompt sentence:

[0640] "Please summarize the meeting and include key topics and decisions."

[0641] Step 6: Sending meeting notes

[0642] Input: Summarized minutes data

[0643] Specific operation: The server automatically sends summarized meeting minutes data to participants' email addresses using SendGrid or Amazon SES.

[0644] Data processing: The summarized minutes data is inserted into the body of an email and sent to everyone based on the appropriate participant list.

[0645] Output: Summary minutes sent to participants' email addresses

[0646] Step 7: Integrate with a chatbot

[0647] Input: A request to search minutes from the user

[0648] Specific behavior: A user sends a request to search for past minutes or specific statements through a chatbot (e.g., Dialogflow, Microsoft Bot Framework).

[0649] Data processing: The server receives a request from the chatbot, searches for relevant information in the database, and extracts the required data.

[0650] Output: Providing search results to the user

[0651] The above are the processing steps and their specific operations of the present invention. The steps work in conjunction with each other to create meeting minutes efficiently and accurately.

[0652] (Application example 1)

[0653] 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."

[0654] In today's factory operations, it is important to accurately record and manage the details and progress of tasks performed by workers and robots. However, manual recording and progress management takes time and effort, and is prone to human error. Searching past work records also requires a great deal of effort. Furthermore, it is difficult to remove unnecessary data from work reports and extract the key points. An efficient system to solve this problem is needed.

[0655] 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.

[0656] In this invention, the server includes means for registering and learning voice data of each worker in advance, means for transcribing voice in real time during work, means for removing unnecessary voice from the transcribed data, means for automatically summarizing work content, means for automatically sending the summary data to a manager, and means for searching past work records in cooperation with a chatbot. This makes it possible to automate work recording and management in factories, improving work efficiency and accuracy.

[0657] "Audio data" refers to verbal information uttered by a worker or speaker that has been recorded in digital form.

[0658] "Means for learning" refers to the process by which the server builds a speech recognition model based on registered speech data, enabling it to identify the voice of a specific speaker.

[0659] "Transcribing means" refers to speech recognition technology for converting audio data into text format.

[0660] The "means for removing unwanted audio" is a process that automatically removes unwanted parts such as "um" and "ah" from the transcribed text.

[0661] "Summarization" is a technique for extracting the main content from transcribed data and summarizing it briefly.

[0662] "Means for automatic sending" refers to a process in which the server automatically sends the generated summary data to a specified email address.

[0663] A "chatbot" is software that automatically converses with users and provides them with information in response to their inquiries.

[0664] The "means for searching past work records" is a process for searching past work data stored on the server based on a user request and extracting necessary information.

[0665] MODE FOR CARRYING OUT THE INVENTION

[0666] This invention is a system for improving the efficiency of work recording and management in factories. This system records the content and progress of work performed by workers and robots in real time, automatically summarizing the data and sending it to managers. It also has a function for searching past work records.

[0667] 1. Registering and learning audio data

[0668] The server receives the voice data uploaded in advance by each worker or robot and trains the speech recognition model. This builds a database of each worker's or robot's characteristic voice, making it possible to accurately identify who is speaking during work. This step uses Python and the speech_recognition library.

[0669] 2. Transcription of audio data

[0670] When a task begins, the device records what is being said and what is being done, and sends it to the server. The server receives the audio data in real time and transcribes it. With the help of a speech recognition model, the transcription is performed while identifying the speaker and the task. The speech_recognition library is also used in this step.

[0671] 3. Remove unwanted audio

[0672] The server uses natural language processing technology to automatically remove unnecessary sounds such as "um" and "ah" from the recorded text, making the work log more concise and easy to read. This step utilizes Python's text processing capabilities.

[0673] 4. Automatic summary generation

[0674] All speech and work content is transcribed and input into a generative AI model, which automatically extracts key points and creates a summary. The summarized work record is then saved on a server. This step uses Hugging Face's transformers module, and the summarize function is used to generate the summary.

[0675] 5. Sending work records

[0676] After the work is completed, the server automatically sends the generated summary to the administrator's email address. This function allows the administrator to check the work record immediately after the work is completed. This step uses the Python smtplib library.

[0677] 6. Chatbot integration

[0678] The server works with the chatbot to search past work records in response to user inquiries and provide the necessary information. This step uses the chatbot software and database search function.

[0679] Specific examples

[0680] For example, a factory robot may report work such as "Welding process 2 has been completed" or "Part A is missing." The system transcribes these statements in real time, removes unnecessary parts, generates a summary, and sends it to the manager by email after the work is completed. It also makes it easy to search past work records through the chatbot.

[0681] Prompt Sentence Examples

[0682] Collaboration with Edge AI Sensors:

[0683] Q: How is the data collected by the AI ​​sensors analyzed?

[0684] Answer: Data collected by the AI ​​sensor is transcribed in real time through a speech recognition module, unnecessary noise is automatically removed, and a summary is generated.

[0685] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0686] Step 1:

[0687] The server pre-registers the voice data of each worker and robot and trains a voice recognition model. The input for this step is the voice data of the worker and robot. The server uses this voice data to build a voice recognition model and processes the data so that it can identify the voice of each worker. The output is a voice feature database for each worker.

[0688] Step 2:

[0689] The terminal records the voices emitted by workers and robots while working and sends the voice data to the server in real time. The input for this step is the voices spoken during work. The terminal records this and sends it to the server. The server converts the received voice data into transcription data. The output is transcription data that converts the voices into text.

[0690] Step 3:

[0691] The server removes unwanted speech from the transcription data. The input for this step is the transcribed text data. The server uses natural language processing techniques to identify unwanted parts, such as "um" and "ah," and removes them from the text. The output is clean text data with the unwanted speech removed.

[0692] Step 4:

[0693] The server inputs the clean text data into a generative AI model to automatically extract the key points of the work content and create a summary. The input for this step is the clean text data. The server uses the generative AI model to extract important points from the text data and create a short summary. The output is a short, summarized text data.

[0694] Step 5:

[0695] The server automatically sends the generated summary to the administrator's email address. The inputs to this step are the summarized text data and the administrator's email address. The server combines these data using the SMTP protocol and sends an email to the administrator. The output is the summary data sent to the administrator's mailbox.

[0696] Step 6:

[0697] When a user searches for past work records through a chatbot, the server searches the database based on the user's inquiry and provides the required information. The input to this step is the user's search query. The server works with the chatbot to extract the relevant work records from the database and present them to the user. The output is the past work record data provided to the user.

[0698] 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.

[0699] This invention combines a conventional meeting recording system with an emotion engine to grasp the emotional state of each participant and create and manage advanced minutes. This system automatically generates meeting minutes while recognizing participants' emotions in real time, which can be used to summarize the meeting and for subsequent follow-up.

[0700] Below, each component of this system and its processing will be specifically explained.

[0701] Server Processing

[0702] 1. Registering and learning audio data

[0703] The server receives voice data uploaded by users in advance and trains a voice recognition model. At the same time, the emotion engine also learns the user's emotional state from the voice data. This allows the model to reflect the emotional characteristics of each participant as well as their voice, making it possible to identify emotional changes during the meeting.

[0704] 2. Real-time transcription and sentiment analysis of audio data

[0705] When a meeting begins, the device records what is being said and sends it to the server. The server receives the audio data in real time, transcribes it, and uses an emotion engine to analyze the emotional state of each utterance. The transcription and emotion analysis are performed while identifying the speaker using a speech recognition model.

[0706] 3. Remove unwanted audio

[0707] The server uses natural language processing technology to automatically remove unwanted sounds such as "um" and "ah" from the recorded text, and the emotion engine adjusts the sound so that the unwanted sounds do not affect the emotion analysis.

[0708] 4. Automatic summary generation

[0709] The server inputs the text data, including all statements, and the results of sentiment analysis into a generation AI, which automatically extracts the key points of the meeting and creates a summary. Emotional data is also taken into consideration, and statements that caused particularly large emotional changes are reflected in the summary. The created summary is saved on the server as minutes of the meeting.

[0710] 5. Sending meeting minutes

[0711] After the meeting, the server automatically sends the generated summary to the participants' email addresses, including emotional data, so participants can share the atmosphere of the meeting and any notable emotional reactions.

[0712] 6. Chatbot integration

[0713] The server works in conjunction with the chatbot to search past meeting minutes in response to user inquiries and provide the necessary information. Searches can also be made including emotional data, making it possible to investigate specific statements and changes in emotion.

[0714] Terminal handling

[0715] 1. Registering audio data

[0716] Users record their own voice and emotions using their devices and upload the voice data to the server, which allows the server to learn the voice and emotion data of each participant in advance.

[0717] 2. Sending voice and emotion data during a meeting

[0718] When a meeting begins, the device records the user's speech and emotions in real time and sends them to the server, which then instantly transcribes the received data and performs sentiment analysis.

[0719] 3. Checking meeting notes

[0720] After the meeting ends, the device receives the minutes sent from the server and displays them to the user. Emotional data is also displayed, allowing the user to get a more detailed understanding of the atmosphere of the meeting.

[0721] User Action

[0722] 1. Registering audio data

[0723] Users use their devices to record and register their own voice and emotions, and this data is uploaded to a server and used as training data for the voice recognition model and emotion engine.

[0724] 2. Record statements and emotions during meetings

[0725] Users can participate in a meeting as usual, make comments, and record their emotions. The device records comments and emotions in real time and sends them to the server, so no special operation is required from the user.

[0726] 3. Checking meeting notes

[0727] After the meeting, the user can check the minutes received from their device. The system's automatic summarization and sentiment analysis allow the user to quickly grasp the main points of the meeting.

[0728] 4. Integration with chatbots

[0729] Users can use the chatbot to search past meeting minutes and specific statements, and can also search for emotional data, allowing them to check specific statements and the emotional state of those statements.

[0730] This system not only automates the creation of meeting minutes, but also captures the emotional state of participants in real time and reflects it in the minutes, thereby increasing the transparency and effectiveness of meetings and promoting deeper understanding and cooperation.

[0731] The processing flow will be explained below.

[0732] Server Processing

[0733] Step 1: Register and train audio data

[0734] 1. The server receives the audio data

[0735] The user uploads the audio data recorded using the terminal to the server.

[0736] 2. The server trains the speech recognition model

[0737] The server trains a speech recognition model based on the received voice data, thereby identifying the characteristics of each participant's voice.

[0738] 3. The server learns the emotion engine

[0739] The server uses an emotion engine to learn the user's emotional state from the voice data, making it possible to understand changes in emotions that accompany speech.

[0740] Step 2: Real-time transcription and sentiment analysis

[0741] 1. The server receives the audio data in real time.

[0742] During the meeting, the device records the user's remarks in real time and sends them to the server.

[0743] 2. The server performs the transcription

[0744] The server transcribes the received audio data in real time and identifies the speaker.

[0745] 3. The server analyzes the emotional data

[0746] As the server transcribes, it uses an emotion engine to analyze the emotional state of each utterance, determining, for example, whether the speaker is excited or calm.

[0747] Step 3: Remove unwanted audio

[0748] 1. The server removes unnecessary audio

[0749] The server uses natural language processing technology to remove unnecessary sounds such as "eh" and "um" from the transcription data, and sentiment analysis is also adjusted at the same time.

[0750] Step 4: Automatic summarization

[0751] 1. The server extracts the main points

[0752] The server uses AI to automatically extract key points from meetings based on transcripts containing emotional data, taking into account statements that showed significant emotional changes.

[0753] 2. The server stores the summary

[0754] The server generates a summary and saves it as a transcript.

[0755] Step 5: Sending meeting notes

[0756] 1. The server sends the minutes

[0757] The server automatically sends the minutes, including emotional data, to the email addresses of all participants.

[0758] Step 6: Integrate with a chatbot

[0759] 1. The server provides database access to the chatbot

[0760] The server grants the chatbot access to the conference database.

[0761] 2. Chatbots handle inquiries

[0762] The chatbot receives a search request from the user, retrieves the necessary data from the server, and responds.

[0763] Terminal handling

[0764] Step 1: Registering audio data

[0765] 1. The device records audio

[0766] The user records their own voice on the device.

[0767] 2. The device sends the voice data to the server

[0768] The device uploads the recorded audio data to the server.

[0769] Step 2: Sending audio data during a meeting

[0770] 1. The device records what you say and sends it to the server

[0771] The device records what is said during the meeting in real time and sends it to the server.

[0772] Step 3: Sending emotion data

[0773] 1. The device sends emotional data

[0774] The device uses an emotion engine to analyze emotions based on the recorded voice data and sends the data to a server.

[0775] Step 4: Review the meeting notes

[0776] 1. The device receives the minutes

[0777] After the meeting ends, the terminal receives the minutes sent from the server.

[0778] 2. The device displays the minutes.

[0779] The terminal displays the received minutes to the user.

[0780] User Action

[0781] Step 1: Registering audio data

[0782] 1. The user records the audio and saves it on the device.

[0783] The user records their own voice on the device.

[0784] 2. The user uploads the audio data to the server.

[0785] Upload the recorded audio data to the server.

[0786] Step 2: Speaking during the meeting

[0787] 1. User speaks

[0788] A user speaks in a conference.

[0789] 2. The user checks the recording

[0790] The user checks whether recording is working properly on the device.

[0791] Step 3: Review the meeting notes

[0792] 1. User checks the minutes

[0793] After the meeting ends, the user checks the minutes received from the terminal.

[0794] Step 4: Integrate with a chatbot

[0795] 1. Users interact with chatbots

[0796] A user uses the chatbot to search for past meeting minutes or specific statements.

[0797] 2. User checks the results

[0798] The user checks the results provided by the chatbot.

[0799] Example 2

[0800] 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."

[0801] Conventional meeting recording systems require manual creation of meeting minutes, which is time-consuming and labor-intensive. Furthermore, it is difficult to record the emotional state of each participant during the meeting, making it difficult to grasp the atmosphere of the meeting in detail. Therefore, there is a need to improve meeting efficiency and ensure transparency.

[0802] 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.

[0803] In this invention, the server includes means for registering and learning the voice data of each participant in advance, means for transcribing the voice in real time during the conference, means for removing unnecessary voice from the transcribed data, means for inputting the transcribed data and emotion data into a generative AI model to automatically summarize the conference content, means for automatically sending the summary data to the participants, means for searching past minutes in cooperation with a chatbot, and means for analyzing and saving emotion data in real time. This automates the creation of meeting minutes and enables detailed meeting records that take emotion data into account.

[0804] "Audio data" means digitally recorded information of a participant's speech.

[0805] "Emotion data" is information that indicates the emotional state of each speaker, analyzed from the voice data.

[0806] A "speech recognition model" is a machine learning or artificial intelligence algorithm for converting speech data into text data.

[0807] An "emotion engine" is an algorithm or system that analyzes audio and text data to identify emotional states.

[0808] "Transcription data" refers to data that has been converted from audio data into text format.

[0809] A "generative artificial intelligence model" is a machine learning or artificial intelligence model that generates new data based on input data.

[0810] "Unnecessary voice" refers to speech that is not related to the significance or content of the meeting (e.g., "umm," "ah," etc.).

[0811] "Summary data" is text data that extracts important points from all statements made at a meeting and summarizes them concisely.

[0812] A "chatbot" is an automated program that interacts with a user to provide or process information.

[0813] "Real-time" refers to the time frame in which data is generated and processed immediately, without delay.

[0814] A "database" is a system for efficiently storing, searching, and managing large amounts of data.

[0815] This invention is a system that automates the creation of meeting minutes and grasps the emotional states of participants in real time, thereby improving the transparency and effectiveness of meetings. This system consists of three main components: a server, a terminal, and a user, and is specifically implemented as follows.

[0816] First, users record their own voice data using their devices and upload it to the server. At this time, the recorded voice data is analyzed using an emotion engine to determine each user's emotional state. The server then trains a voice recognition model based on the received voice data. For example, it uses Google Cloud Speech-to-Text to convert the voice data into text and uses Hume AI to learn emotional characteristics.

[0817] When a meeting starts, the device records participants' speech in real time and sends it to the server. The server converts the received audio data into text in real time and simultaneously analyzes the emotional state of the speech using an emotion engine. This process uses Amazon Transcribe, for example, and stores the generated text data and emotion data in a database.

[0818] The server also uses natural language processing techniques to automatically remove unwanted sounds like "um" and "ah" from the text data. For example, it uses the SpaCy library to analyze the text and remove unnecessary parts. The clean text data is then stored in a database.

[0819] All spoken text and sentiment data are fed into a generative AI model to extract key points from the meeting and automatically generate a summary. OpenAI's GPT-4 can be used to generate summaries, using prompts such as "Please consider all spoken data from the meeting and its sentiment data, extract key points, and generate a summary of the meeting."

[0820] After the meeting, the server automatically sends the generated summary to participants, for example via email using the SendGrid API, which also includes sentiment data, allowing participants to understand the atmosphere of the meeting and any notable emotional reactions.

[0821] In addition, the server has the ability to search past meeting minutes in cooperation with chatbots. Users can use chatbots to efficiently search for meeting minutes containing specific statements or emotional states. For example, a chatbot can be built using Dialogflow and data can be searched using Elasticsearch.

[0822] The system automates meeting minutes creation and provides detailed meeting records that take emotional data into account, improving the transparency and effectiveness of meeting content and fostering deeper understanding and collaboration.

[0823] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0824] Step 1:

[0825] The user records their own voice data on the device and uploads it to the server.

[0826] Input: User recorded voice data

[0827] Specific operation: The user records their own voice using the recording app on their device and uploads the file to the server, where the uploaded data is stored.

[0828] Output: Audio data stored on the server

[0829] Step 2:

[0830] The server receives the speech data and trains a speech recognition model.

[0831] Input: Audio data uploaded from the device

[0832] How it works: The server converts voice data into text using Google Cloud Speech-to-Text or similar, and then uses an emotion engine such as Hume AI to analyze the emotional state from the voice data and saves the learning results in a database.

[0833] Output: Text data and emotion data corresponding to the audio data

[0834] Step 3:

[0835] The device records what is said during the meeting and sends it to the server in real time.

[0836] Input: Audio data during the meeting

[0837] Specific operation: The device records speech in real time and sends it to the server using WebSocket or similar.

[0838] Output: Audio data sent to the server in real time

[0839] Step 4:

[0840] The server transcribes the audio data in real time and analyzes the emotional state.

[0841] Input: Real-time audio data

[0842] How it works: The server uses Amazon Transcribe to convert the voice data into text, and simultaneously uses the emotion engine to analyze the emotional state. This data is then stored in a database in real time.

[0843] Output: Real-time text and sentiment data

[0844] Step 5:

[0845] The server removes unwanted audio from the transcription data.

[0846] Input: Real-time generated text data

[0847] Specific operation: The server uses the SpaCy library to analyze the text data and remove unnecessary sounds such as "um" and "ah." The cleaned text data is then restored to the database.

[0848] Output: Text data with unnecessary audio removed

[0849] Step 6:

[0850] The server uses a generative AI model to automatically summarize the key points of the meeting.

[0851] Input: Clean text data and sentiment data

[0852] Specific operation: The server uses OpenAI GPT-4 to generate a meeting summary from text data and emotion data, using the prompt "Consider all speech data and their emotion data from the meeting, extract key points, and generate a meeting summary." The generated summary is stored in a database.

[0853] Output: Summarized meeting data

[0854] Step 7:

[0855] After the conference ends, the server automatically sends summary data to the participants.

[0856] Input: Summarized meeting data

[0857] Specific operation: The server uses the SendGrid API to send the saved summary data to the participant's email address.

[0858] Output: Summary data sent to participants

[0859] Step 8:

[0860] The server works in conjunction with the chatbot to search for past minutes.

[0861] Input: User's search request

[0862] Specific operation: The server uses Dialogflow to build a chatbot, searches the database using Elasticsearch, and returns the search results and sentiment data to the user.

[0863] Output: Past meeting minutes data provided to the user

[0864] Through these processing steps, meeting minutes are automated and detailed meeting records are generated that take emotional data into account. This system improves the transparency and effectiveness of meetings, promoting deeper understanding and cooperation.

[0865] (Application example 2)

[0866] 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."

[0867] In conventional meeting recording systems, simply converting speech to text makes it difficult to accurately grasp the nuances of what is being said and the emotional state of the participants.In addition, in regular and emergency meetings within a factory, it is important to understand not only the main points of the meeting but also the emotional state of the participants, but previous systems were unable to perform real-time emotional analysis.

[0868] The identification processing 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 registering and learning the voice data of each participant in advance, means for transcribing the voice in real time during the conference, means for removing unnecessary voice from the transcribed data, means for automatically summarizing the conference content, means for analyzing the emotional state, means for automatically sending summary data to the participants, and means for searching past minutes in cooperation with a chatbot. This makes it possible to grasp the emotional state of the participants in real time in addition to summarizing the conference content, and to create and manage highly accurate minutes.

[0869] "Voice data of each participant" refers to voice information emitted by each participant who participates in a conference registered in the system.

[0870] The "means for pre-registration and learning" refers to a function that allows characteristic voice data of each participant to be input into the system before the start of the meeting, and for the voice recognition model to learn the voice characteristics of each participant based on that data.

[0871] "Means for transcribing audio in real time" is a function that instantly converts audio spoken during a meeting into text data.

[0872] "Means for removing unnecessary sounds from transcribed data" is a function for automatically removing meaningless sounds such as "eh" and "um" from transcribed text data.

[0873] The "means for automatically summarizing meeting content" is a function that analyzes what is said during a meeting, extracts important points, and generates a concise summary.

[0874] The "means for analyzing emotional states" is a function that analyzes the emotions contained in each statement in real time based on transcribed text data.

[0875] The "means for automatically transmitting summary data to participants" is a function for automatically transmitting the generated summary data to the communication addresses of all participants after the conference has ended.

[0876] "Means for searching past minutes in cooperation with a chatbot" refers to a chatbot function that searches past minutes or specific statements within them and provides relevant information in response to inquiries from users after a meeting.

[0877] This invention is a system that creates minutes and grasps the emotional state of participants in factory meetings. This system consists of three main components: a server, a terminal, and a user. We will explain how each component works.

[0878] Server Action:

[0879] 1. Register and train the audio data:

[0880] The server receives voice data uploaded by users in advance and trains a voice recognition model. The voice data contains the characteristic voice information of each participant, which improves the accuracy of voice recognition during the meeting. At the same time, it also learns the emotional state of each participant using an emotion engine.

[0881] 2. Real-time transcription and sentiment analysis:

[0882] When a meeting begins, the device records the audio and sends it to the server, which receives it in real time, transcribes it, and then uses an emotion engine to analyze the emotional state of each comment.

[0883] 3. Remove unwanted audio:

[0884] Using natural language processing technology, unnecessary sounds such as "um" and "ah" are automatically removed from the recorded text data, improving the quality of the transcription data.

[0885] 4. Automatic summary generation:

[0886] Based on the results of speech data and sentiment analysis, a generative AI model is used to automatically extract key points from meetings and generate summaries, including statements that elicited particularly strong sentiment.

[0887] 5. Sending minutes:

[0888] After the meeting ends, the server automatically sends the generated summary to the communication addresses of the participants, allowing the contents of the meeting to be shared quickly.

[0889] 6. Chatbot integration:

[0890] The server works in conjunction with the chatbot to search past meeting minutes in response to user questions and provide the necessary information. Emotional data can also be searched, allowing users to check specific statements and their emotional state at the time.

[0891] Terminal handling:

[0892] 1. Audio data registration:

[0893] Users use their devices to record and register their own voice and emotions, and this data is uploaded to a server and used to train the voice recognition model and emotion engine.

[0894] 2. Sending voice and emotion data:

[0895] During a meeting, the device records the user's speech and emotions in real time and transmits them to the server, which then uses the received data for immediate transcription and sentiment analysis.

[0896] 3. Check the minutes:

[0897] After the meeting ends, the device receives the minutes sent from the server and displays them to the user, allowing the user to check the details of the meeting and the emotional state at the time.

[0898] User Action:

[0899] 1. Audio data registration:

[0900] Users use their devices to record and register their own voice and emotions, which allows the voice recognition model and emotion engine to work more accurately.

[0901] 2. Record statements and emotions during the meeting:

[0902] Users simply join a meeting as usual and record their comments and emotions on their device, with no special operations required.

[0903] 3. Check the minutes:

[0904] After the meeting, users can review the minutes received from their devices and quickly grasp the main points by utilizing the system's automatic summarization and sentiment analysis.

[0905] 4. Chatbot integration:

[0906] Through the chatbot, users can search past meeting minutes and specific statements. They can also search for emotion data, allowing them to check specific statements and the emotional state of those statements.

[0907] Examples:

[0908] During regular meetings held within a factory, an assistant robot summarizes the meeting and analyzes sentiment. The minutes are then sent to participants' mailboxes after the meeting. For example, if a participant asks the chatbot, "What were the most common opinions expressed in last month's meeting?", the robot can provide a summary along with past sentiment.

[0909] Example prompt sentence:

[0910] "What were the most common opinions and comments that generated a lot of emotion at last month's meeting?"

[0911] As described above, the system of the present invention can significantly improve the efficiency of meetings within a factory.

[0912] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0913] Step 1: Register and train audio data

[0914] Users use their devices to record their own voice and emotions and upload them to the server. The server then trains the received voice data based on a speech recognition model. The input is the voice data and emotion data from the users, and the server learns from this to understand the voice and emotion features of each participant. The output is the speech recognition model after training is complete.

[0915] Step 2: Real-time transcription and sentiment analysis

[0916] The device records the audio during the meeting in real time and sends this audio data to a server, which then instantly transcribes the audio data and performs sentiment analysis using a generative AI model.The input is the audio data from the meeting, and the server outputs text data and the emotional state of each utterance.

[0917] Step 3: Remove unwanted audio

[0918] The server uses natural language processing technology to automatically remove unnecessary sounds such as "eh" and "um" from the generated text data. The input is transcribed text data, and unnecessary sounds are removed as part of the data processing. The output is clear transcribed text data.

[0919] Step 4: Automatically generate a summary

[0920] The server uses the transcription data and the results of sentiment analysis to automatically summarize the meeting content based on a generative AI model. The input is clear transcription text data and sentiment data, which are then processed into a summary using the generative AI model. The output is the summarized text data.

[0921] Step 5: Sending meeting notes

[0922] After the conference ends, the server automatically sends the generated summary data to the participants' communication addresses. The input is the summarized text data and the participants' communication addresses, and the server calculates and sends this as data in the form of email. The output is the email sent to the participants.

[0923] Step 6: Integrate with a chatbot

[0924] The user uses a chatbot from their device to search for past minutes or specific statements. The server searches for past minutes and emotion data via the chatbot and provides them to the user. The input is a search query (prompt sentence) from the user, and the server calculates and outputs the appropriate minutes and emotion data from the database. The output is the search results provided to the user.

[0925] 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.

[0926] 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.

[0927] 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.

[0928] [Third embodiment]

[0929] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0930] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0931] 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).

[0932] 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.

[0933] 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.

[0934] 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).

[0935] 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.

[0936] 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.

[0937] 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.

[0938] 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.

[0939] 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.

[0940] 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."

[0941] This invention is a system for automatically generating and efficiently managing meeting minutes. This system pre-registers and learns from the voice data of each participant to accurately identify who spoke, transcribes the voice in real time during the meeting, removes unnecessary voice, summarizes the meeting content, and automatically sends the summary data to participants after the meeting. It is also possible to search past meeting minutes in conjunction with a chatbot.

[0942] Below, each component of this system and its processing will be specifically explained.

[0943] Server Processing

[0944] 1. Registering and learning audio data

[0945] The server receives voice data previously uploaded by users and trains a speech recognition model, building a database of each participant's distinctive voice, enabling accurate identification of who is speaking during a meeting.

[0946] 2. Transcription of audio data

[0947] When a meeting begins, the device records what is being said and sends it to the server, which receives the audio data in real time and transcribes it. With the help of a speech recognition model, the transcription is done while identifying the speaker.

[0948] 3. Remove unwanted audio

[0949] The server uses natural language processing technology to automatically remove unnecessary sounds such as "um" and "ah" from the recorded text, making the minutes more concise and easier to read.

[0950] 4. Automatic summary generation

[0951] The server inputs the transcribed data of all statements into the generation AI, which automatically extracts the key points of the meeting and creates a summary. The summarized minutes are then stored on the server.

[0952] 5. Sending meeting minutes

[0953] After the meeting, the server automatically sends the generated summary to the participants' email addresses, allowing participants to check the minutes immediately after the meeting ends.

[0954] 6. Chatbot integration

[0955] The server works in conjunction with the chatbot to search past minutes in response to user inquiries and provide the necessary information.

[0956] Terminal handling

[0957] 1. Registering audio data

[0958] Users record their own voice using their devices and upload the voice data to the server, which allows the server to learn the voice data of each participant in advance.

[0959] 2. Sending audio data during a meeting

[0960] When the meeting begins, the device records the user's speech in real time and sends it to the server, which receives it and immediately transcribes it.

[0961] 3. Checking meeting notes

[0962] After the meeting ends, the terminal receives the minutes sent from the server and displays them to the user, allowing the user to quickly grasp the main points of the meeting.

[0963] User Action

[0964] 1. Registering audio data

[0965] Users use their devices to record and register their own voice, which is then uploaded to a server and used as training data for the voice recognition model.

[0966] 2. Statements made during meetings

[0967] Users can join a conference and speak as usual. The device records what they say in real time and sends it to the server, so no special operations are required on the part of the user.

[0968] 3. Checking meeting notes

[0969] After the meeting, users can check the minutes data received from their devices. The system's automatic summary function allows users to quickly grasp the main points of the meeting.

[0970] 4. Integration with chatbots

[0971] Users can use the chatbot to easily search for and check past meeting minutes or specific comments. The chatbot extracts the relevant information from a database on the server and presents it to the user.

[0972] By introducing this system, the creation of meeting minutes will be automated, significantly improving work efficiency. It also identifies speakers and removes unnecessary audio, improving the quality of meeting minutes. This reduces disputes over "said" and "unsaid," making it possible to run meetings with greater transparency.

[0973] The processing flow will be explained below.

[0974] Server Processing

[0975] Step 1: Register and train audio data

[0976] 1. The user records audio data and saves it on the device.

[0977] The user records their own voice and saves it on the device.

[0978] 2. The device uploads the audio data to the server

[0979] The device transmits the stored voice data to the server.

[0980] 3. The server receives the voice data and learns

[0981] The server receives the audio data and trains a speech recognition model, incorporating the voice characteristics of a particular user to identify what is being said.

[0982] Step 2: Real-time transcription

[0983] 1. The device sends the audio data during the meeting to the server

[0984] After the conference starts, the terminal records the user's speech in real time and transmits it intermittently to the server.

[0985] 2. The server transcribes the audio data in real time

[0986] The server transcribes the received audio data in real time, using a speech recognition model to identify each speaker while transcribing.

[0987] Step 3: Remove unwanted audio

[0988] 1. The server uses natural language processing to remove unnecessary audio

[0989] The server uses natural language processing technology to automatically remove unnecessary sounds such as "um" and "ah" from the transcribed data.

[0990] Step 4: Automatic summarization

[0991] 1. Server generates a summary using AI

[0992] The server inputs text data containing all statements into the generation AI, which automatically extracts the key points of the meeting and creates a summary.

[0993] The created summary is saved as minutes on the server.

[0994] Step 5: Sending meeting notes

[0995] 1. The server sends the minutes to all participants

[0996] The server automatically sends the saved summary to the registered email addresses of all conference participants.

[0997] Step 6: Integrate with a chatbot

[0998] 1. The server provides database access to the chatbot

[0999] The server provides the chatbot with access to a database of past meeting minutes and configures it to return data in response to queries.

[1000] 2. Chatbots handle user inquiries

[1001] The chatbot receives a search request from the user, retrieves and displays relevant meeting minutes data.

[1002] Terminal handling

[1003] Step 1: Registering audio data

[1004] 1. The user records the audio and saves it on the device.

[1005] The user records and saves their own voice on the device.

[1006] 2. The device uploads the audio data to the server

[1007] The device sends the recorded audio data to the server.

[1008] Step 2: Sending audio data during a meeting

[1009] 1. The device records what the user says and sends it to the server

[1010] The device records the user's remarks in real time during the meeting and transmits them intermittently to the server.

[1011] Step 3: Review the meeting notes

[1012] 1. Display the minutes received by the device

[1013] After the conference ends, the terminal receives the minutes sent from the server and displays them to the user.

[1014] User Action

[1015] Step 1: Registering audio data

[1016] 1. The user records the audio and saves it on the device.

[1017] The user records and saves their own voice.

[1018] 2. The user uploads the audio data to the server using their device.

[1019] The user uploads the audio data recorded using the terminal to the server.

[1020] Step 2: Speaking during the meeting

[1021] 1. The user speaks as usual

[1022] The user speaks normally in the conference.

[1023] 2. Users use devices to record their speech.

[1024] When a user speaks, the device automatically starts recording and sends it to the server.

[1025] Step 3: Review the meeting notes

[1026] 1. User checks the minutes

[1027] After the meeting ends, the user checks the minutes that have arrived on the terminal, which displays the minutes received from the server.

[1028] Step 4: Integrate with a chatbot

[1029] 1. User searches and checks with chatbot

[1030] Users can use the chatbot to search for past minutes or specific statements and check the results.

[1031] The chatbot extracts the relevant information from the database on the server and displays it to the user.

[1032] Example 1

[1033] 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."

[1034] Conventional meeting minutes creation requires manual recording, which is time-consuming and labor-intensive, and it is difficult to identify speakers and extract important content. Furthermore, sharing minutes and searching past records is also time-consuming, making efficient information management difficult. This has led to problems such as a decline in meeting transparency and work efficiency.

[1035] 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.

[1036] In this invention, the server includes means for registering and learning the voice data of each participant in advance, means for transcribing the voice in real time during the conference, means for removing unnecessary voice from the transcribed data, means for automatically summarizing the conference content, means for automatically sending the summary data to the participants, and means for searching past minutes in cooperation with a chatbot. This automates the creation of meeting minutes, enables quick and accurate identification of speakers and extraction of important content, and improves the transparency and work efficiency of meetings.

[1037] "Audio data" refers to recorded audio information in digital or analog format.

[1038] "Learning" is the process by which a speech recognition model analyzes input speech data and makes it possible to identify the characteristics of the speaker and the content of the speech.

[1039] "Transcription" is the process of analyzing audio data and converting its content into text format data.

[1040] "Unnecessary audio" refers to meaningless, redundant audio such as "um" and "ah," which detracts from the conciseness of the minutes.

[1041] A "summary" is text data that analyzes the contents of a meeting and succinctly summarizes the main points and decisions made.

[1042] "Auto-send" is the process of automatically sending specific data to a pre-defined email address or other address.

[1043] A "chatbot" is a software application that uses natural language processing to automatically respond to user questions.

[1044] A "minutes" is a document that records what was said and what was decided at a meeting.

[1045] "Speaker" refers to each participant who speaks during a conference.

[1046] A "database" is an electromagnetic storage device for efficiently storing, managing, and retrieving specific data.

[1047] "Email" is a means of communication for sending and receiving text and files over the Internet.

[1048] MODE FOR CARRYING OUT THE INVENTION

[1049] This invention is a system for automatically generating and efficiently managing meeting minutes. This system registers and learns the voice data of each participant in advance, identifies who spoke, transcribes the meeting in real time, removes unnecessary sounds, and summarizes the meeting content. It also automatically sends the summary data to participants after the meeting ends, and can also search for past minutes in conjunction with a chatbot.

[1050] Audio data registration and training

[1051] First, the user records their own voice on their device using a standard recording application (e.g., Voice Recorder). This voice data is then uploaded to a server. The server then analyzes the uploaded voice data using the Google Cloud Speech-to-Text API or similar to learn the characteristics of each speaker. This makes it easier to identify speakers during a meeting.

[1052] Transcription of audio data

[1053] When a meeting starts, the device records the user's speech in real time. This recording data is then sent to the server. The server then uses the Google Cloud Speech-to-Text API to convert the received audio data into text in real time. At the same time, it identifies the speaker using a pre-trained model.

[1054] Removing unwanted audio

[1055] The server removes unnecessary sounds such as "um" and "ah" from the transcribed text using natural language processing libraries such as the Natural Language Toolkit (NLTK) and SpaCy. This process makes the minutes easier to read.

[1056] Automatic summary generation

[1057] Based on all the transcribed text data, a generative AI model (e.g., OpenAI GPT-4) is used to extract the main points of the meeting. The server sets prompts for summary generation, and the main points are automatically summarized based on these prompts.

[1058] Example prompt sentence:

[1059] "Please summarize the meeting and include key topics and decisions."

[1060] Sending meeting notes

[1061] After the meeting ends, the server automatically sends the generated summary to the participants' email addresses using email sending services such as SendGrid or Amazon SES, allowing participants to check the summarized minutes immediately after the meeting ends.

[1062] Collaboration with chatbots

[1063] The server works with a chatbot (e.g., Dialogflow, Microsoft Bot Framework) to respond to requests from users to search for meeting minutes. When a user searches for past meeting minutes or specific statements, the chatbot retrieves the necessary information from the database on the server and provides it to the user.

[1064] These processes automate the creation of meeting minutes, allowing for quick and accurate identification of speakers and extraction of important content, significantly improving meeting transparency and work efficiency.

[1065] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1066] Step 1: Register and train audio data

[1067] Input: User recorded voice data

[1068] How it works: The user records their own voice using a standard recording application, and the recording is uploaded from the device to the server.

[1069] Data processing: The server converts the received voice data into text data using the Google Cloud Speech-to-Text API, etc. Based on this text data, a speech recognition model is used to learn the speaker's characteristics.

[1070] Output: A database containing speaker characteristics data

[1071] Step 2: Record and send audio during the meeting

[1072] Input: The voice that the user speaks during the meeting.

[1073] Specific operation: When the conference starts, the terminal records the user's speech in real time, and the recorded data is sent to the server sequentially.

[1074] Data processing: The device transfers the recorded audio data to the server via an audio data streaming service.

[1075] Output: Real-time audio data sent to the server

[1076] Step 3: Transcribe the audio data

[1077] Input: Real-time audio data received by the server

[1078] Specific operation: The server converts the received voice data into text data using the Google Cloud Speech-to-Text API.

[1079] Data processing: Based on real-time speech recognition, the audio data is transcribed into corresponding text data, while simultaneously identifying the speaker.

[1080] Output: Real-time text data with speaker identification

[1081] Step 4: Remove unwanted audio

[1082] Input: Transcribed real-time text data

[1083] Specific operation: The server uses the Natural Language Toolkit (NLTK) and SpaCy libraries to detect and remove redundant sounds such as "eh" and "um" from text data.

[1084] Data processing: Natural language processing techniques are used to filter out redundant data and generate clean text data.

[1085] Output: Transcribed text data with redundant speech removed

[1086] Step 5: Automatically generate summaries

[1087] Input: Fairly formatted text data

[1088] How it works: The server uses a generative AI model (e.g., OpenAI GPT-4) to extract key points from the input, typed text data.

[1089] Data processing: The summarization process uses prompts, and the AI ​​model extracts key points from the text data and generates a summary.

[1090] Output: Summarized minutes data

[1091] Example prompt sentence:

[1092] "Please summarize the meeting and include key topics and decisions."

[1093] Step 6: Sending meeting notes

[1094] Input: Summarized minutes data

[1095] Specific operation: The server automatically sends summarized meeting minutes data to participants' email addresses using SendGrid or Amazon SES.

[1096] Data processing: The summarized minutes data is inserted into the body of an email and sent to everyone based on the appropriate participant list.

[1097] Output: Summary minutes sent to participants' email addresses

[1098] Step 7: Integrate with a chatbot

[1099] Input: A request to search minutes from the user

[1100] Specific behavior: A user sends a request to search for past minutes or specific statements through a chatbot (e.g., Dialogflow, Microsoft Bot Framework).

[1101] Data processing: The server receives a request from the chatbot, searches for relevant information in the database, and extracts the required data.

[1102] Output: Providing search results to the user

[1103] The above are the processing steps and their specific operations of the present invention. The steps work in conjunction with each other to create meeting minutes efficiently and accurately.

[1104] (Application example 1)

[1105] 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."

[1106] In today's factory operations, it is important to accurately record and manage the details and progress of tasks performed by workers and robots. However, manual recording and progress management takes time and effort, and is prone to human error. Searching past work records also requires a great deal of effort. Furthermore, it is difficult to remove unnecessary data from work reports and extract the key points. An efficient system to solve this problem is needed.

[1107] 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.

[1108] In this invention, the server includes means for registering and learning voice data of each worker in advance, means for transcribing voice in real time during work, means for removing unnecessary voice from the transcribed data, means for automatically summarizing work content, means for automatically sending the summary data to a manager, and means for searching past work records in cooperation with a chatbot. This makes it possible to automate work recording and management in factories, improving work efficiency and accuracy.

[1109] "Audio data" refers to verbal information uttered by a worker or speaker that has been recorded in digital form.

[1110] "Means for learning" refers to the process by which the server builds a speech recognition model based on registered speech data, enabling it to identify the voice of a specific speaker.

[1111] "Transcribing means" refers to speech recognition technology for converting audio data into text format.

[1112] The "means for removing unwanted audio" is a process that automatically removes unwanted parts such as "um" and "ah" from the transcribed text.

[1113] "Summarization" is a technique for extracting the main content from transcribed data and summarizing it briefly.

[1114] "Means for automatic sending" refers to a process in which the server automatically sends the generated summary data to a specified email address.

[1115] A "chatbot" is software that automatically converses with users and provides them with information in response to their inquiries.

[1116] The "means for searching past work records" is a process for searching past work data stored on the server based on a user request and extracting necessary information.

[1117] MODE FOR CARRYING OUT THE INVENTION

[1118] This invention is a system for improving the efficiency of work recording and management in factories. This system records the content and progress of work performed by workers and robots in real time, automatically summarizing the data and sending it to managers. It also has a function for searching past work records.

[1119] 1. Registering and learning audio data

[1120] The server receives the voice data uploaded in advance by each worker or robot and trains the speech recognition model. This builds a database of each worker's or robot's characteristic voice, making it possible to accurately identify who is speaking during work. This step uses Python and the speech_recognition library.

[1121] 2. Transcription of audio data

[1122] When a task begins, the device records what is being said and what is being done, and sends it to the server. The server receives the audio data in real time and transcribes it. With the help of a speech recognition model, the transcription is performed while identifying the speaker and the task. The speech_recognition library is also used in this step.

[1123] 3. Remove unwanted audio

[1124] The server uses natural language processing technology to automatically remove unnecessary sounds such as "um" and "ah" from the recorded text, making the work log more concise and easy to read. This step utilizes Python's text processing capabilities.

[1125] 4. Automatic summary generation

[1126] All speech and work content is transcribed and input into a generative AI model, which automatically extracts key points and creates a summary. The summarized work record is then saved on a server. This step uses Hugging Face's transformers module, and the summarize function is used to generate the summary.

[1127] 5. Sending work records

[1128] After the work is completed, the server automatically sends the generated summary to the administrator's email address. This function allows the administrator to check the work record immediately after the work is completed. This step uses the Python smtplib library.

[1129] 6. Chatbot integration

[1130] The server works with the chatbot to search past work records in response to user inquiries and provide the necessary information. This step uses the chatbot software and database search function.

[1131] Specific examples

[1132] For example, a factory robot may report work such as "Welding process 2 has been completed" or "Part A is missing." The system transcribes these statements in real time, removes unnecessary parts, generates a summary, and sends it to the manager by email after the work is completed. It also makes it easy to search past work records through the chatbot.

[1133] Prompt Sentence Examples

[1134] Collaboration with Edge AI Sensors:

[1135] Q: How is the data collected by the AI ​​sensors analyzed?

[1136] Answer: Data collected by the AI ​​sensor is transcribed in real time through a speech recognition module, unnecessary noise is automatically removed, and a summary is generated.

[1137] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1138] Step 1:

[1139] The server pre-registers the voice data of each worker and robot and trains a voice recognition model. The input for this step is the voice data of the worker and robot. The server uses this voice data to build a voice recognition model and processes the data so that it can identify the voice of each worker. The output is a voice feature database for each worker.

[1140] Step 2:

[1141] The terminal records the voices emitted by workers and robots while working and sends the voice data to the server in real time. The input for this step is the voices spoken during work. The terminal records this and sends it to the server. The server converts the received voice data into transcription data. The output is transcription data that converts the voices into text.

[1142] Step 3:

[1143] The server removes unwanted speech from the transcription data. The input for this step is the transcribed text data. The server uses natural language processing techniques to identify unwanted parts, such as "um" and "ah," and removes them from the text. The output is clean text data with the unwanted speech removed.

[1144] Step 4:

[1145] The server inputs the clean text data into a generative AI model to automatically extract the key points of the work content and create a summary. The input for this step is the clean text data. The server uses the generative AI model to extract important points from the text data and create a short summary. The output is a short, summarized text data.

[1146] Step 5:

[1147] The server automatically sends the generated summary to the administrator's email address. The inputs to this step are the summarized text data and the administrator's email address. The server combines these data using the SMTP protocol and sends an email to the administrator. The output is the summary data sent to the administrator's mailbox.

[1148] Step 6:

[1149] When a user searches for past work records through a chatbot, the server searches the database based on the user's inquiry and provides the required information. The input to this step is the user's search query. The server works with the chatbot to extract the relevant work records from the database and present them to the user. The output is the past work record data provided to the user.

[1150] 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.

[1151] This invention combines a conventional meeting recording system with an emotion engine to grasp the emotional state of each participant and create and manage advanced minutes. This system automatically generates meeting minutes while recognizing participants' emotions in real time, which can be used to summarize the meeting and for subsequent follow-up.

[1152] Below, each component of this system and its processing will be specifically explained.

[1153] Server Processing

[1154] 1. Registering and learning audio data

[1155] The server receives voice data uploaded by users in advance and trains a voice recognition model. At the same time, the emotion engine also learns the user's emotional state from the voice data. This allows the model to reflect the emotional characteristics of each participant as well as their voice, making it possible to identify emotional changes during the meeting.

[1156] 2. Real-time transcription and sentiment analysis of audio data

[1157] When a meeting begins, the device records what is being said and sends it to the server. The server receives the audio data in real time, transcribes it, and uses an emotion engine to analyze the emotional state of each utterance. The transcription and emotion analysis are performed while identifying the speaker using a speech recognition model.

[1158] 3. Remove unwanted audio

[1159] The server uses natural language processing technology to automatically remove unwanted sounds such as "um" and "ah" from the recorded text, and the emotion engine adjusts the sound so that the unwanted sounds do not affect the emotion analysis.

[1160] 4. Automatic summary generation

[1161] The server inputs the text data, including all statements, and the results of sentiment analysis into a generation AI, which automatically extracts the key points of the meeting and creates a summary. Emotional data is also taken into consideration, and statements that caused particularly large emotional changes are reflected in the summary. The created summary is saved on the server as minutes of the meeting.

[1162] 5. Sending meeting minutes

[1163] After the meeting, the server automatically sends the generated summary to the participants' email addresses, including emotional data, so participants can share the atmosphere of the meeting and any notable emotional reactions.

[1164] 6. Chatbot integration

[1165] The server works in conjunction with the chatbot to search past meeting minutes in response to user inquiries and provide the necessary information. Searches can also be made including emotional data, making it possible to investigate specific statements and changes in emotion.

[1166] Terminal handling

[1167] 1. Registering audio data

[1168] Users record their own voice and emotions using their devices and upload the voice data to the server, which allows the server to learn the voice and emotion data of each participant in advance.

[1169] 2. Sending voice and emotion data during a meeting

[1170] When a meeting begins, the device records the user's speech and emotions in real time and sends them to the server, which then instantly transcribes the received data and performs sentiment analysis.

[1171] 3. Checking meeting notes

[1172] After the meeting ends, the device receives the minutes sent from the server and displays them to the user. Emotional data is also displayed, allowing the user to get a more detailed understanding of the atmosphere of the meeting.

[1173] User Action

[1174] 1. Registering audio data

[1175] Users use their devices to record and register their own voice and emotions, and this data is uploaded to a server and used as training data for the voice recognition model and emotion engine.

[1176] 2. Record statements and emotions during meetings

[1177] Users can participate in a meeting as usual, make comments, and record their emotions. The device records comments and emotions in real time and sends them to the server, so no special operation is required from the user.

[1178] 3. Checking meeting notes

[1179] After the meeting, the user can check the minutes received from their device. The system's automatic summarization and sentiment analysis allow the user to quickly grasp the main points of the meeting.

[1180] 4. Integration with chatbots

[1181] Users can use the chatbot to search past meeting minutes and specific statements, and can also search for emotional data, allowing them to check specific statements and the emotional state of those statements.

[1182] This system not only automates the creation of meeting minutes, but also captures the emotional state of participants in real time and reflects it in the minutes, thereby increasing the transparency and effectiveness of meetings and promoting deeper understanding and cooperation.

[1183] The processing flow will be explained below.

[1184] Server Processing

[1185] Step 1: Register and train audio data

[1186] 1. The server receives the audio data

[1187] The user uploads the audio data recorded using the terminal to the server.

[1188] 2. The server trains the speech recognition model

[1189] The server trains a speech recognition model based on the received voice data, thereby identifying the characteristics of each participant's voice.

[1190] 3. The server learns the emotion engine

[1191] The server uses an emotion engine to learn the user's emotional state from the voice data, making it possible to understand changes in emotions that accompany speech.

[1192] Step 2: Real-time transcription and sentiment analysis

[1193] 1. The server receives the audio data in real time.

[1194] During the meeting, the device records the user's remarks in real time and sends them to the server.

[1195] 2. The server performs the transcription

[1196] The server transcribes the received audio data in real time and identifies the speaker.

[1197] 3. The server analyzes the emotional data

[1198] As the server transcribes, it uses an emotion engine to analyze the emotional state of each utterance, determining, for example, whether the speaker is excited or calm.

[1199] Step 3: Remove unwanted audio

[1200] 1. The server removes unnecessary audio

[1201] The server uses natural language processing technology to remove unnecessary sounds such as "eh" and "um" from the transcription data, and sentiment analysis is also adjusted at the same time.

[1202] Step 4: Automatic summarization

[1203] 1. The server extracts the main points

[1204] The server uses AI to automatically extract key points from meetings based on transcripts containing emotional data, taking into account statements that showed significant emotional changes.

[1205] 2. The server stores the summary

[1206] The server generates a summary and saves it as a transcript.

[1207] Step 5: Sending meeting notes

[1208] 1. The server sends the minutes

[1209] The server automatically sends the minutes, including emotional data, to the email addresses of all participants.

[1210] Step 6: Integrate with a chatbot

[1211] 1. The server provides database access to the chatbot

[1212] The server grants the chatbot access to the conference database.

[1213] 2. Chatbots handle inquiries

[1214] The chatbot receives a search request from the user, retrieves the necessary data from the server, and responds.

[1215] Terminal handling

[1216] Step 1: Registering audio data

[1217] 1. The device records audio

[1218] The user records their own voice on the device.

[1219] 2. The device sends the voice data to the server

[1220] The device uploads the recorded audio data to the server.

[1221] Step 2: Sending audio data during a meeting

[1222] 1. The device records what you say and sends it to the server

[1223] The device records what is said during the meeting in real time and sends it to the server.

[1224] Step 3: Sending emotion data

[1225] 1. The device sends emotional data

[1226] The device uses an emotion engine to analyze emotions based on the recorded voice data and sends the data to a server.

[1227] Step 4: Review the meeting notes

[1228] 1. The device receives the minutes

[1229] After the meeting ends, the terminal receives the minutes sent from the server.

[1230] 2. The device displays the minutes.

[1231] The terminal displays the received minutes to the user.

[1232] User Action

[1233] Step 1: Registering audio data

[1234] 1. The user records the audio and saves it on the device.

[1235] The user records their own voice on the device.

[1236] 2. The user uploads the audio data to the server.

[1237] Upload the recorded audio data to the server.

[1238] Step 2: Speaking during the meeting

[1239] 1. User speaks

[1240] A user speaks in a conference.

[1241] 2. The user checks the recording

[1242] The user checks whether recording is working properly on the device.

[1243] Step 3: Review the meeting notes

[1244] 1. User checks the minutes

[1245] After the meeting ends, the user checks the minutes received from the terminal.

[1246] Step 4: Integrate with a chatbot

[1247] 1. Users interact with chatbots

[1248] A user uses the chatbot to search for past meeting minutes or specific statements.

[1249] 2. User checks the results

[1250] The user checks the results provided by the chatbot.

[1251] Example 2

[1252] 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."

[1253] Conventional meeting recording systems require manual creation of meeting minutes, which is time-consuming and labor-intensive. Furthermore, it is difficult to record the emotional state of each participant during the meeting, making it difficult to grasp the atmosphere of the meeting in detail. Therefore, there is a need to improve meeting efficiency and ensure transparency.

[1254] 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.

[1255] In this invention, the server includes means for registering and learning the voice data of each participant in advance, means for transcribing the voice in real time during the conference, means for removing unnecessary voice from the transcribed data, means for inputting the transcribed data and emotion data into a generative AI model to automatically summarize the conference content, means for automatically sending the summary data to the participants, means for searching past minutes in cooperation with a chatbot, and means for analyzing and saving emotion data in real time. This automates the creation of meeting minutes and enables detailed meeting records that take emotion data into account.

[1256] "Audio data" means digitally recorded information of a participant's speech.

[1257] "Emotion data" is information that indicates the emotional state of each speaker, analyzed from the voice data.

[1258] A "speech recognition model" is a machine learning or artificial intelligence algorithm for converting speech data into text data.

[1259] An "emotion engine" is an algorithm or system that analyzes audio and text data to identify emotional states.

[1260] "Transcription data" refers to data that has been converted from audio data into text format.

[1261] A "generative artificial intelligence model" is a machine learning or artificial intelligence model that generates new data based on input data.

[1262] "Unnecessary voice" refers to speech that is not related to the significance or content of the meeting (e.g., "umm," "ah," etc.).

[1263] "Summary data" is text data that extracts important points from all statements made at a meeting and summarizes them concisely.

[1264] A "chatbot" is an automated program that interacts with a user to provide or process information.

[1265] "Real-time" refers to the time frame in which data is generated and processed immediately, without delay.

[1266] A "database" is a system for efficiently storing, searching, and managing large amounts of data.

[1267] This invention is a system that automates the creation of meeting minutes and grasps the emotional states of participants in real time, thereby improving the transparency and effectiveness of meetings. This system consists of three main components: a server, a terminal, and a user, and is specifically implemented as follows.

[1268] First, users record their own voice data using their devices and upload it to the server. At this time, the recorded voice data is analyzed using an emotion engine to determine each user's emotional state. The server then trains a voice recognition model based on the received voice data. For example, it uses Google Cloud Speech-to-Text to convert the voice data into text and uses Hume AI to learn emotional characteristics.

[1269] When a meeting starts, the device records participants' speech in real time and sends it to the server. The server converts the received audio data into text in real time and simultaneously analyzes the emotional state of the speech using an emotion engine. This process uses Amazon Transcribe, for example, and stores the generated text data and emotion data in a database.

[1270] The server also uses natural language processing techniques to automatically remove unwanted sounds like "um" and "ah" from the text data. For example, it uses the SpaCy library to analyze the text and remove unnecessary parts. The clean text data is then stored in a database.

[1271] All spoken text and sentiment data are fed into a generative AI model to extract key points from the meeting and automatically generate a summary. OpenAI's GPT-4 can be used to generate summaries, using prompts such as "Please consider all spoken data from the meeting and its sentiment data, extract key points, and generate a summary of the meeting."

[1272] After the meeting, the server automatically sends the generated summary to participants, for example via email using the SendGrid API, which also includes sentiment data, allowing participants to understand the atmosphere of the meeting and any notable emotional reactions.

[1273] In addition, the server has the ability to search past meeting minutes in cooperation with chatbots. Users can use chatbots to efficiently search for meeting minutes containing specific statements or emotional states. For example, a chatbot can be built using Dialogflow and data can be searched using Elasticsearch.

[1274] The system automates meeting minutes creation and provides detailed meeting records that take emotional data into account, improving the transparency and effectiveness of meeting content and fostering deeper understanding and collaboration.

[1275] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1276] Step 1:

[1277] The user records their own voice data on the device and uploads it to the server.

[1278] Input: User recorded voice data

[1279] Specific operation: The user records their own voice using the recording app on their device and uploads the file to the server, where the uploaded data is stored.

[1280] Output: Audio data stored on the server

[1281] Step 2:

[1282] The server receives the speech data and trains a speech recognition model.

[1283] Input: Audio data uploaded from the device

[1284] How it works: The server converts voice data into text using Google Cloud Speech-to-Text or similar, and then uses an emotion engine such as Hume AI to analyze the emotional state from the voice data and saves the learning results in a database.

[1285] Output: Text data and emotion data corresponding to the audio data

[1286] Step 3:

[1287] The device records what is said during the meeting and sends it to the server in real time.

[1288] Input: Audio data during the meeting

[1289] Specific operation: The device records speech in real time and sends it to the server using WebSocket or similar.

[1290] Output: Audio data sent to the server in real time

[1291] Step 4:

[1292] The server transcribes the audio data in real time and analyzes the emotional state.

[1293] Input: Real-time audio data

[1294] How it works: The server uses Amazon Transcribe to convert the voice data into text, and simultaneously uses the emotion engine to analyze the emotional state. This data is then stored in a database in real time.

[1295] Output: Real-time text and sentiment data

[1296] Step 5:

[1297] The server removes unwanted audio from the transcription data.

[1298] Input: Real-time generated text data

[1299] Specific operation: The server uses the SpaCy library to analyze the text data and remove unnecessary sounds such as "um" and "ah." The cleaned text data is then restored to the database.

[1300] Output: Text data with unnecessary audio removed

[1301] Step 6:

[1302] The server uses a generative AI model to automatically summarize the key points of the meeting.

[1303] Input: Clean text data and sentiment data

[1304] Specific operation: The server uses OpenAI GPT-4 to generate a meeting summary from text data and emotion data, using the prompt "Consider all speech data and their emotion data from the meeting, extract key points, and generate a meeting summary." The generated summary is stored in a database.

[1305] Output: Summarized meeting data

[1306] Step 7:

[1307] After the conference ends, the server automatically sends summary data to the participants.

[1308] Input: Summarized meeting data

[1309] Specific operation: The server uses the SendGrid API to send the saved summary data to the participant's email address.

[1310] Output: Summary data sent to participants

[1311] Step 8:

[1312] The server works in conjunction with the chatbot to search for past minutes.

[1313] Input: User's search request

[1314] Specific operation: The server uses Dialogflow to build a chatbot, searches the database using Elasticsearch, and returns the search results and sentiment data to the user.

[1315] Output: Past meeting minutes data provided to the user

[1316] Through these processing steps, meeting minutes are automated and detailed meeting records are generated that take emotional data into account. This system improves the transparency and effectiveness of meetings, promoting deeper understanding and cooperation.

[1317] (Application example 2)

[1318] 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."

[1319] In conventional meeting recording systems, simply converting speech to text makes it difficult to accurately grasp the nuances of what is being said and the emotional state of the participants.In addition, in regular and emergency meetings within a factory, it is important to understand not only the main points of the meeting but also the emotional state of the participants, but previous systems were unable to perform real-time emotional analysis.

[1320] The identification processing 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 registering and learning the voice data of each participant in advance, means for transcribing the voice in real time during the conference, means for removing unnecessary voice from the transcribed data, means for automatically summarizing the conference content, means for analyzing the emotional state, means for automatically sending summary data to the participants, and means for searching past minutes in cooperation with a chatbot. This makes it possible to grasp the emotional state of the participants in real time in addition to summarizing the conference content, and to create and manage highly accurate minutes.

[1321] "Voice data of each participant" refers to voice information emitted by each participant who participates in a conference registered in the system.

[1322] The "means for pre-registration and learning" refers to a function that allows characteristic voice data of each participant to be input into the system before the start of the meeting, and for the voice recognition model to learn the voice characteristics of each participant based on that data.

[1323] "Means for transcribing audio in real time" is a function that instantly converts audio spoken during a meeting into text data.

[1324] "Means for removing unnecessary sounds from transcribed data" is a function for automatically removing meaningless sounds such as "eh" and "um" from transcribed text data.

[1325] The "means for automatically summarizing meeting content" is a function that analyzes what is said during a meeting, extracts important points, and generates a concise summary.

[1326] The "means for analyzing emotional states" is a function that analyzes the emotions contained in each statement in real time based on transcribed text data.

[1327] The "means for automatically transmitting summary data to participants" is a function for automatically transmitting the generated summary data to the communication addresses of all participants after the conference has ended.

[1328] "Means for searching past minutes in cooperation with a chatbot" refers to a chatbot function that searches past minutes or specific statements within them and provides relevant information in response to inquiries from users after a meeting.

[1329] This invention is a system that creates minutes and grasps the emotional state of participants in factory meetings. This system consists of three main components: a server, a terminal, and a user. We will explain how each component works.

[1330] Server Action:

[1331] 1. Register and train the audio data:

[1332] The server receives voice data uploaded by users in advance and trains a voice recognition model. The voice data contains the characteristic voice information of each participant, which improves the accuracy of voice recognition during the meeting. At the same time, it also learns the emotional state of each participant using an emotion engine.

[1333] 2. Real-time transcription and sentiment analysis:

[1334] When a meeting begins, the device records the audio and sends it to the server, which receives it in real time, transcribes it, and then uses an emotion engine to analyze the emotional state of each comment.

[1335] 3. Remove unwanted audio:

[1336] Using natural language processing technology, unnecessary sounds such as "um" and "ah" are automatically removed from the recorded text data, improving the quality of the transcription data.

[1337] 4. Automatic summary generation:

[1338] Based on the results of speech data and sentiment analysis, a generative AI model is used to automatically extract key points from meetings and generate summaries, including statements that elicited particularly strong sentiment.

[1339] 5. Sending minutes:

[1340] After the meeting ends, the server automatically sends the generated summary to the communication addresses of the participants, allowing the contents of the meeting to be shared quickly.

[1341] 6. Chatbot integration:

[1342] The server works in conjunction with the chatbot to search past meeting minutes in response to user questions and provide the necessary information. Emotional data can also be searched, allowing users to check specific statements and their emotional state at the time.

[1343] Terminal handling:

[1344] 1. Audio data registration:

[1345] Users use their devices to record and register their own voice and emotions, and this data is uploaded to a server and used to train the voice recognition model and emotion engine.

[1346] 2. Sending voice and emotion data:

[1347] During a meeting, the device records the user's speech and emotions in real time and transmits them to the server, which then uses the received data for immediate transcription and sentiment analysis.

[1348] 3. Check the minutes:

[1349] After the meeting ends, the device receives the minutes sent from the server and displays them to the user, allowing the user to check the details of the meeting and the emotional state at the time.

[1350] User Action:

[1351] 1. Audio data registration:

[1352] Users use their devices to record and register their own voice and emotions, which allows the voice recognition model and emotion engine to work more accurately.

[1353] 2. Record statements and emotions during the meeting:

[1354] Users simply join a meeting as usual and record their comments and emotions on their device, with no special operations required.

[1355] 3. Check the minutes:

[1356] After the meeting, users can review the minutes received from their devices and quickly grasp the main points by utilizing the system's automatic summarization and sentiment analysis.

[1357] 4. Chatbot integration:

[1358] Through the chatbot, users can search past meeting minutes and specific statements. They can also search for emotion data, allowing them to check specific statements and the emotional state of those statements.

[1359] Examples:

[1360] During regular meetings held within a factory, an assistant robot summarizes the meeting and analyzes sentiment. The minutes are then sent to participants' mailboxes after the meeting. For example, if a participant asks the chatbot, "What were the most common opinions expressed in last month's meeting?", the robot can provide a summary along with past sentiment.

[1361] Example prompt sentence:

[1362] "What were the most common opinions and comments that generated a lot of emotion at last month's meeting?"

[1363] As described above, the system of the present invention can significantly improve the efficiency of meetings within a factory.

[1364] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1365] Step 1: Register and train audio data

[1366] Users use their devices to record their own voice and emotions and upload them to the server. The server then trains the received voice data based on a speech recognition model. The input is the voice data and emotion data from the users, and the server learns from this to understand the voice and emotion features of each participant. The output is the speech recognition model after training is complete.

[1367] Step 2: Real-time transcription and sentiment analysis

[1368] The device records the audio during the meeting in real time and sends this audio data to a server, which then instantly transcribes the audio data and performs sentiment analysis using a generative AI model.The input is the audio data from the meeting, and the server outputs text data and the emotional state of each utterance.

[1369] Step 3: Remove unwanted audio

[1370] The server uses natural language processing technology to automatically remove unnecessary sounds such as "eh" and "um" from the generated text data. The input is transcribed text data, and unnecessary sounds are removed as part of the data processing. The output is clear transcribed text data.

[1371] Step 4: Automatically generate a summary

[1372] The server uses the transcription data and the results of sentiment analysis to automatically summarize the meeting content based on a generative AI model. The input is clear transcription text data and sentiment data, which are then processed into a summary using the generative AI model. The output is the summarized text data.

[1373] Step 5: Sending meeting notes

[1374] After the conference ends, the server automatically sends the generated summary data to the participants' communication addresses. The input is the summarized text data and the participants' communication addresses, and the server calculates and sends this as data in the form of email. The output is the email sent to the participants.

[1375] Step 6: Integrate with a chatbot

[1376] The user uses a chatbot from their device to search for past minutes or specific statements. The server searches for past minutes and emotion data via the chatbot and provides them to the user. The input is a search query (prompt sentence) from the user, and the server calculates and outputs the appropriate minutes and emotion data from the database. The output is the search results provided to the user.

[1377] 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.

[1378] 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.

[1379] 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.

[1380] [Fourth embodiment]

[1381] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1382] 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.

[1383] 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).

[1384] 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.

[1385] 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.

[1386] 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).

[1387] 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.

[1388] 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.

[1389] 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.

[1390] 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.

[1391] 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.

[1392] 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.

[1393] 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."

[1394] This invention is a system for automatically generating and efficiently managing meeting minutes. This system pre-registers and learns from the voice data of each participant to accurately identify who spoke, transcribes the voice in real time during the meeting, removes unnecessary voice, summarizes the meeting content, and automatically sends the summary data to participants after the meeting. It is also possible to search past meeting minutes in conjunction with a chatbot.

[1395] Below, each component of this system and its processing will be specifically explained.

[1396] Server Processing

[1397] 1. Registering and learning audio data

[1398] The server receives voice data previously uploaded by users and trains a speech recognition model, building a database of each participant's distinctive voice, enabling accurate identification of who is speaking during a meeting.

[1399] 2. Transcription of audio data

[1400] When a meeting begins, the device records what is being said and sends it to the server, which receives the audio data in real time and transcribes it. With the help of a speech recognition model, the transcription is done while identifying the speaker.

[1401] 3. Remove unwanted audio

[1402] The server uses natural language processing technology to automatically remove unnecessary sounds such as "um" and "ah" from the recorded text, making the minutes more concise and easier to read.

[1403] 4. Automatic summary generation

[1404] The server inputs the transcribed data of all statements into the generation AI, which automatically extracts the key points of the meeting and creates a summary. The summarized minutes are then stored on the server.

[1405] 5. Sending meeting minutes

[1406] After the meeting, the server automatically sends the generated summary to the participants' email addresses, allowing participants to check the minutes immediately after the meeting ends.

[1407] 6. Chatbot integration

[1408] The server works in conjunction with the chatbot to search past minutes in response to user inquiries and provide the necessary information.

[1409] Terminal handling

[1410] 1. Registering audio data

[1411] Users record their own voice using their devices and upload the voice data to the server, which allows the server to learn the voice data of each participant in advance.

[1412] 2. Sending audio data during a meeting

[1413] When the meeting begins, the device records the user's speech in real time and sends it to the server, which receives it and immediately transcribes it.

[1414] 3. Checking meeting notes

[1415] After the meeting ends, the terminal receives the minutes sent from the server and displays them to the user, allowing the user to quickly grasp the main points of the meeting.

[1416] User Action

[1417] 1. Registering audio data

[1418] Users use their devices to record and register their own voice, which is then uploaded to a server and used as training data for the voice recognition model.

[1419] 2. Statements made during meetings

[1420] Users can join a conference and speak as usual. The device records what they say in real time and sends it to the server, so no special operations are required on the part of the user.

[1421] 3. Checking meeting notes

[1422] After the meeting, users can check the minutes data received from their devices. The system's automatic summary function allows users to quickly grasp the main points of the meeting.

[1423] 4. Integration with chatbots

[1424] Users can use the chatbot to easily search for and check past meeting minutes or specific comments. The chatbot extracts the relevant information from a database on the server and presents it to the user.

[1425] By introducing this system, the creation of meeting minutes will be automated, significantly improving work efficiency. It also identifies speakers and removes unnecessary audio, improving the quality of meeting minutes. This reduces disputes over "said" and "unsaid," making it possible to run meetings with greater transparency.

[1426] The processing flow will be explained below.

[1427] Server Processing

[1428] Step 1: Register and train audio data

[1429] 1. The user records audio data and saves it on the device.

[1430] The user records their own voice and saves it on the device.

[1431] 2. The device uploads the audio data to the server

[1432] The device transmits the stored voice data to the server.

[1433] 3. The server receives the voice data and learns

[1434] The server receives the audio data and trains a speech recognition model, incorporating the voice characteristics of a particular user to identify what is being said.

[1435] Step 2: Real-time transcription

[1436] 1. The device sends the audio data during the meeting to the server

[1437] After the conference starts, the terminal records the user's speech in real time and transmits it intermittently to the server.

[1438] 2. The server transcribes the audio data in real time

[1439] The server transcribes the received audio data in real time, using a speech recognition model to identify each speaker while transcribing.

[1440] Step 3: Remove unwanted audio

[1441] 1. The server uses natural language processing to remove unnecessary audio

[1442] The server uses natural language processing technology to automatically remove unnecessary sounds such as "um" and "ah" from the transcribed data.

[1443] Step 4: Automatic summarization

[1444] 1. Server generates a summary using AI

[1445] The server inputs text data containing all statements into the generation AI, which automatically extracts the key points of the meeting and creates a summary.

[1446] The created summary is saved as minutes on the server.

[1447] Step 5: Sending meeting notes

[1448] 1. The server sends the minutes to all participants

[1449] The server automatically sends the saved summary to the registered email addresses of all conference participants.

[1450] Step 6: Integrate with a chatbot

[1451] 1. The server provides database access to the chatbot

[1452] The server provides the chatbot with access to a database of past meeting minutes and configures it to return data in response to queries.

[1453] 2. Chatbots handle user inquiries

[1454] The chatbot receives a search request from the user, retrieves and displays relevant meeting minutes data.

[1455] Terminal handling

[1456] Step 1: Registering audio data

[1457] 1. The user records the audio and saves it on the device.

[1458] The user records and saves their own voice on the device.

[1459] 2. The device uploads the audio data to the server

[1460] The device sends the recorded audio data to the server.

[1461] Step 2: Sending audio data during a meeting

[1462] 1. The device records what the user says and sends it to the server

[1463] The device records the user's remarks in real time during the meeting and transmits them intermittently to the server.

[1464] Step 3: Review the meeting notes

[1465] 1. Display the minutes received by the device

[1466] After the conference ends, the terminal receives the minutes sent from the server and displays them to the user.

[1467] User Action

[1468] Step 1: Registering audio data

[1469] 1. The user records the audio and saves it on the device.

[1470] The user records and saves their own voice.

[1471] 2. The user uploads the audio data to the server using their device.

[1472] The user uploads the audio data recorded using the terminal to the server.

[1473] Step 2: Speaking during the meeting

[1474] 1. The user speaks as usual

[1475] The user speaks normally in the conference.

[1476] 2. Users use devices to record their speech.

[1477] When a user speaks, the device automatically starts recording and sends it to the server.

[1478] Step 3: Review the meeting notes

[1479] 1. User checks the minutes

[1480] After the meeting ends, the user checks the minutes that have arrived on the terminal, which displays the minutes received from the server.

[1481] Step 4: Integrate with a chatbot

[1482] 1. User searches and checks with chatbot

[1483] Users can use the chatbot to search for past minutes or specific statements and check the results.

[1484] The chatbot extracts the relevant information from the database on the server and displays it to the user.

[1485] Example 1

[1486] 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."

[1487] Conventional meeting minutes creation requires manual recording, which is time-consuming and labor-intensive, and it is difficult to identify speakers and extract important content. Furthermore, sharing minutes and searching past records is also time-consuming, making efficient information management difficult. This has led to problems such as a decline in meeting transparency and work efficiency.

[1488] 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.

[1489] In this invention, the server includes means for registering and learning the voice data of each participant in advance, means for transcribing the voice in real time during the conference, means for removing unnecessary voice from the transcribed data, means for automatically summarizing the conference content, means for automatically sending the summary data to the participants, and means for searching past minutes in cooperation with a chatbot. This automates the creation of meeting minutes, enables quick and accurate identification of speakers and extraction of important content, and improves the transparency and work efficiency of meetings.

[1490] "Audio data" refers to recorded audio information in digital or analog format.

[1491] "Learning" is the process by which a speech recognition model analyzes input speech data and makes it possible to identify the characteristics of the speaker and the content of the speech.

[1492] "Transcription" is the process of analyzing audio data and converting its content into text format data.

[1493] "Unnecessary audio" refers to meaningless, redundant audio such as "um" and "ah," which detracts from the conciseness of the minutes.

[1494] A "summary" is text data that analyzes the contents of a meeting and succinctly summarizes the main points and decisions made.

[1495] "Auto-send" is the process of automatically sending specific data to a pre-defined email address or other address.

[1496] A "chatbot" is a software application that uses natural language processing to automatically respond to user questions.

[1497] A "minutes" is a document that records what was said and what was decided at a meeting.

[1498] "Speaker" refers to each participant who speaks during a conference.

[1499] A "database" is an electromagnetic storage device for efficiently storing, managing, and retrieving specific data.

[1500] "Email" is a means of communication for sending and receiving text and files over the Internet.

[1501] MODE FOR CARRYING OUT THE INVENTION

[1502] This invention is a system for automatically generating and efficiently managing meeting minutes. This system registers and learns the voice data of each participant in advance, identifies who spoke, transcribes the meeting in real time, removes unnecessary sounds, and summarizes the meeting content. It also automatically sends the summary data to participants after the meeting ends, and can also search for past minutes in conjunction with a chatbot.

[1503] Audio data registration and training

[1504] First, the user records their own voice on their device using a standard recording application (e.g., Voice Recorder). This voice data is then uploaded to a server. The server then analyzes the uploaded voice data using the Google Cloud Speech-to-Text API or similar to learn the characteristics of each speaker. This makes it easier to identify speakers during a meeting.

[1505] Transcription of audio data

[1506] When a meeting starts, the device records the user's speech in real time. This recording data is then sent to the server. The server then uses the Google Cloud Speech-to-Text API to convert the received audio data into text in real time. At the same time, it identifies the speaker using a pre-trained model.

[1507] Removing unwanted audio

[1508] The server removes unnecessary sounds such as "um" and "ah" from the transcribed text using natural language processing libraries such as the Natural Language Toolkit (NLTK) and SpaCy. This process makes the minutes easier to read.

[1509] Automatic summary generation

[1510] Based on all the transcribed text data, a generative AI model (e.g., OpenAI GPT-4) is used to extract the main points of the meeting. The server sets prompts for summary generation, and the main points are automatically summarized based on these prompts.

[1511] Example prompt sentence:

[1512] "Please summarize the meeting and include key topics and decisions."

[1513] Sending meeting notes

[1514] After the meeting ends, the server automatically sends the generated summary to the participants' email addresses using email sending services such as SendGrid or Amazon SES, allowing participants to check the summarized minutes immediately after the meeting ends.

[1515] Collaboration with chatbots

[1516] The server works with a chatbot (e.g., Dialogflow, Microsoft Bot Framework) to respond to requests from users to search for meeting minutes. When a user searches for past meeting minutes or specific statements, the chatbot retrieves the necessary information from the database on the server and provides it to the user.

[1517] These processes automate the creation of meeting minutes, allowing for quick and accurate identification of speakers and extraction of important content, significantly improving meeting transparency and work efficiency.

[1518] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1519] Step 1: Register and train audio data

[1520] Input: User recorded voice data

[1521] How it works: The user records their own voice using a standard recording application, and the recording is uploaded from the device to the server.

[1522] Data processing: The server converts the received voice data into text data using the Google Cloud Speech-to-Text API, etc. Based on this text data, a speech recognition model is used to learn the speaker's characteristics.

[1523] Output: A database containing speaker characteristics data

[1524] Step 2: Record and send audio during the meeting

[1525] Input: The voice that the user speaks during the meeting.

[1526] Specific operation: When the conference starts, the terminal records the user's speech in real time, and the recorded data is sent to the server sequentially.

[1527] Data processing: The device transfers the recorded audio data to the server via an audio data streaming service.

[1528] Output: Real-time audio data sent to the server

[1529] Step 3: Transcribe the audio data

[1530] Input: Real-time audio data received by the server

[1531] Specific operation: The server converts the received voice data into text data using the Google Cloud Speech-to-Text API.

[1532] Data processing: Based on real-time speech recognition, the audio data is transcribed into corresponding text data, while simultaneously identifying the speaker.

[1533] Output: Real-time text data with speaker identification

[1534] Step 4: Remove unwanted audio

[1535] Input: Transcribed real-time text data

[1536] Specific operation: The server uses the Natural Language Toolkit (NLTK) and SpaCy libraries to detect and remove redundant sounds such as "eh" and "um" from text data.

[1537] Data processing: Natural language processing techniques are used to filter out redundant data and generate clean text data.

[1538] Output: Transcribed text data with redundant speech removed

[1539] Step 5: Automatically generate summaries

[1540] Input: Fairly formatted text data

[1541] How it works: The server uses a generative AI model (e.g., OpenAI GPT-4) to extract key points from the input, typed text data.

[1542] Data processing: The summarization process uses prompts, and the AI ​​model extracts key points from the text data and generates a summary.

[1543] Output: Summarized minutes data

[1544] Example prompt sentence:

[1545] "Please summarize the meeting and include key topics and decisions."

[1546] Step 6: Sending meeting notes

[1547] Input: Summarized minutes data

[1548] Specific operation: The server automatically sends summarized meeting minutes data to participants' email addresses using SendGrid or Amazon SES.

[1549] Data processing: The summarized minutes data is inserted into the body of an email and sent to everyone based on the appropriate participant list.

[1550] Output: Summary minutes sent to participants' email addresses

[1551] Step 7: Integrate with a chatbot

[1552] Input: A request to search minutes from the user

[1553] Specific behavior: A user sends a request to search for past minutes or specific statements through a chatbot (e.g., Dialogflow, Microsoft Bot Framework).

[1554] Data processing: The server receives a request from the chatbot, searches for relevant information in the database, and extracts the required data.

[1555] Output: Providing search results to the user

[1556] The above are the processing steps and their specific operations of the present invention. The steps work in conjunction with each other to create meeting minutes efficiently and accurately.

[1557] (Application example 1)

[1558] 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."

[1559] In today's factory operations, it is important to accurately record and manage the details and progress of tasks performed by workers and robots. However, manual recording and progress management takes time and effort, and is prone to human error. Searching past work records also requires a great deal of effort. Furthermore, it is difficult to remove unnecessary data from work reports and extract the key points. An efficient system to solve this problem is needed.

[1560] 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.

[1561] In this invention, the server includes means for registering and learning voice data of each worker in advance, means for transcribing voice in real time during work, means for removing unnecessary voice from the transcribed data, means for automatically summarizing work content, means for automatically sending the summary data to a manager, and means for searching past work records in cooperation with a chatbot. This makes it possible to automate work recording and management in factories, improving work efficiency and accuracy.

[1562] "Audio data" refers to verbal information uttered by a worker or speaker that has been recorded in digital form.

[1563] "Means for learning" refers to the process by which the server builds a speech recognition model based on registered speech data, enabling it to identify the voice of a specific speaker.

[1564] "Transcribing means" refers to speech recognition technology for converting audio data into text format.

[1565] The "means for removing unwanted audio" is a process that automatically removes unwanted parts such as "um" and "ah" from the transcribed text.

[1566] "Summarization" is a technique for extracting the main content from transcribed data and summarizing it briefly.

[1567] "Means for automatic sending" refers to a process in which the server automatically sends the generated summary data to a specified email address.

[1568] A "chatbot" is software that automatically converses with users and provides them with information in response to their inquiries.

[1569] The "means for searching past work records" is a process for searching past work data stored on the server based on a user request and extracting necessary information.

[1570] MODE FOR CARRYING OUT THE INVENTION

[1571] This invention is a system for improving the efficiency of work recording and management in factories. This system records the content and progress of work performed by workers and robots in real time, automatically summarizing the data and sending it to managers. It also has a function for searching past work records.

[1572] 1. Registering and learning audio data

[1573] The server receives the voice data uploaded in advance by each worker or robot and trains the speech recognition model. This builds a database of each worker's or robot's characteristic voice, making it possible to accurately identify who is speaking during work. This step uses Python and the speech_recognition library.

[1574] 2. Transcription of audio data

[1575] When a task begins, the device records what is being said and what is being done, and sends it to the server. The server receives the audio data in real time and transcribes it. With the help of a speech recognition model, the transcription is performed while identifying the speaker and the task. The speech_recognition library is also used in this step.

[1576] 3. Remove unwanted audio

[1577] The server uses natural language processing technology to automatically remove unnecessary sounds such as "um" and "ah" from the recorded text, making the work log more concise and easy to read. This step utilizes Python's text processing capabilities.

[1578] 4. Automatic summary generation

[1579] All speech and work content is transcribed and input into a generative AI model, which automatically extracts key points and creates a summary. The summarized work record is then saved on a server. This step uses Hugging Face's transformers module, and the summarize function is used to generate the summary.

[1580] 5. Sending work records

[1581] After the work is completed, the server automatically sends the generated summary to the administrator's email address. This function allows the administrator to check the work record immediately after the work is completed. This step uses the Python smtplib library.

[1582] 6. Chatbot integration

[1583] The server works with the chatbot to search past work records in response to user inquiries and provide the necessary information. This step uses the chatbot software and database search function.

[1584] Specific examples

[1585] For example, a factory robot may report work such as "Welding process 2 has been completed" or "Part A is missing." The system transcribes these statements in real time, removes unnecessary parts, generates a summary, and sends it to the manager by email after the work is completed. It also makes it easy to search past work records through the chatbot.

[1586] Prompt Sentence Examples

[1587] Collaboration with Edge AI Sensors:

[1588] Q: How is the data collected by the AI ​​sensors analyzed?

[1589] Answer: Data collected by the AI ​​sensor is transcribed in real time through a speech recognition module, unnecessary noise is automatically removed, and a summary is generated.

[1590] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1591] Step 1:

[1592] The server pre-registers the voice data of each worker and robot and trains a voice recognition model. The input for this step is the voice data of the worker and robot. The server uses this voice data to build a voice recognition model and processes the data so that it can identify the voice of each worker. The output is a voice feature database for each worker.

[1593] Step 2:

[1594] The terminal records the voices emitted by workers and robots while working and sends the voice data to the server in real time. The input for this step is the voices spoken during work. The terminal records this and sends it to the server. The server converts the received voice data into transcription data. The output is transcription data that converts the voices into text.

[1595] Step 3:

[1596] The server removes unwanted speech from the transcription data. The input for this step is the transcribed text data. The server uses natural language processing techniques to identify unwanted parts, such as "um" and "ah," and removes them from the text. The output is clean text data with the unwanted speech removed.

[1597] Step 4:

[1598] The server inputs the clean text data into a generative AI model to automatically extract the key points of the work content and create a summary. The input for this step is the clean text data. The server uses the generative AI model to extract important points from the text data and create a short summary. The output is a short, summarized text data.

[1599] Step 5:

[1600] The server automatically sends the generated summary to the administrator's email address. The inputs to this step are the summarized text data and the administrator's email address. The server combines these data using the SMTP protocol and sends an email to the administrator. The output is the summary data sent to the administrator's mailbox.

[1601] Step 6:

[1602] When a user searches for past work records through a chatbot, the server searches the database based on the user's inquiry and provides the required information. The input to this step is the user's search query. The server works with the chatbot to extract the relevant work records from the database and present them to the user. The output is the past work record data provided to the user.

[1603] 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.

[1604] This invention combines a conventional meeting recording system with an emotion engine to grasp the emotional state of each participant and create and manage advanced minutes. This system automatically generates meeting minutes while recognizing participants' emotions in real time, which can be used to summarize the meeting and for subsequent follow-up.

[1605] Below, each component of this system and its processing will be specifically explained.

[1606] Server Processing

[1607] 1. Registering and learning audio data

[1608] The server receives voice data uploaded by users in advance and trains a voice recognition model. At the same time, the emotion engine also learns the user's emotional state from the voice data. This allows the model to reflect the emotional characteristics of each participant as well as their voice, making it possible to identify emotional changes during the meeting.

[1609] 2. Real-time transcription and sentiment analysis of audio data

[1610] When a meeting begins, the device records what is being said and sends it to the server. The server receives the audio data in real time, transcribes it, and uses an emotion engine to analyze the emotional state of each utterance. The transcription and emotion analysis are performed while identifying the speaker using a speech recognition model.

[1611] 3. Remove unwanted audio

[1612] The server uses natural language processing technology to automatically remove unwanted sounds such as "um" and "ah" from the recorded text, and the emotion engine adjusts the sound so that the unwanted sounds do not affect the emotion analysis.

[1613] 4. Automatic summary generation

[1614] The server inputs the text data, including all statements, and the results of sentiment analysis into a generation AI, which automatically extracts the key points of the meeting and creates a summary. Emotional data is also taken into consideration, and statements that caused particularly large emotional changes are reflected in the summary. The created summary is saved on the server as minutes of the meeting.

[1615] 5. Sending meeting minutes

[1616] After the meeting, the server automatically sends the generated summary to the participants' email addresses, including emotional data, so participants can share the atmosphere of the meeting and any notable emotional reactions.

[1617] 6. Chatbot integration

[1618] The server works in conjunction with the chatbot to search past meeting minutes in response to user inquiries and provide the necessary information. Searches can also be made including emotional data, making it possible to investigate specific statements and changes in emotion.

[1619] Terminal handling

[1620] 1. Registering audio data

[1621] Users record their own voice and emotions using their devices and upload the voice data to the server, which allows the server to learn the voice and emotion data of each participant in advance.

[1622] 2. Sending voice and emotion data during a meeting

[1623] When a meeting begins, the device records the user's speech and emotions in real time and sends them to the server, which then instantly transcribes the received data and performs sentiment analysis.

[1624] 3. Checking meeting notes

[1625] After the meeting ends, the device receives the minutes sent from the server and displays them to the user. Emotional data is also displayed, allowing the user to get a more detailed understanding of the atmosphere of the meeting.

[1626] User Action

[1627] 1. Registering audio data

[1628] Users use their devices to record and register their own voice and emotions, and this data is uploaded to a server and used as training data for the voice recognition model and emotion engine.

[1629] 2. Record statements and emotions during meetings

[1630] Users can participate in a meeting as usual, make comments, and record their emotions. The device records comments and emotions in real time and sends them to the server, so no special operation is required from the user.

[1631] 3. Checking meeting notes

[1632] After the meeting, the user can check the minutes received from their device. The system's automatic summarization and sentiment analysis allow the user to quickly grasp the main points of the meeting.

[1633] 4. Integration with chatbots

[1634] Users can use the chatbot to search past meeting minutes and specific statements, and can also search for emotional data, allowing them to check specific statements and the emotional state of those statements.

[1635] This system not only automates the creation of meeting minutes, but also captures the emotional state of participants in real time and reflects it in the minutes, thereby increasing the transparency and effectiveness of meetings and promoting deeper understanding and cooperation.

[1636] The processing flow will be explained below.

[1637] Server Processing

[1638] Step 1: Register and train audio data

[1639] 1. The server receives the audio data

[1640] The user uploads the audio data recorded using the terminal to the server.

[1641] 2. The server trains the speech recognition model

[1642] The server trains a speech recognition model based on the received voice data, thereby identifying the characteristics of each participant's voice.

[1643] 3. The server learns the emotion engine

[1644] The server uses an emotion engine to learn the user's emotional state from the voice data, making it possible to understand changes in emotions that accompany speech.

[1645] Step 2: Real-time transcription and sentiment analysis

[1646] 1. The server receives the audio data in real time.

[1647] During the meeting, the device records the user's remarks in real time and sends them to the server.

[1648] 2. The server performs the transcription

[1649] The server transcribes the received audio data in real time and identifies the speaker.

[1650] 3. The server analyzes the emotional data

[1651] As the server transcribes, it uses an emotion engine to analyze the emotional state of each utterance, determining, for example, whether the speaker is excited or calm.

[1652] Step 3: Remove unwanted audio

[1653] 1. The server removes unnecessary audio

[1654] The server uses natural language processing technology to remove unnecessary sounds such as "eh" and "um" from the transcription data, and sentiment analysis is also adjusted at the same time.

[1655] Step 4: Automatic summarization

[1656] 1. The server extracts the main points

[1657] The server uses AI to automatically extract key points from meetings based on transcripts containing emotional data, taking into account statements that showed significant emotional changes.

[1658] 2. The server stores the summary

[1659] The server generates a summary and saves it as a transcript.

[1660] Step 5: Sending meeting notes

[1661] 1. The server sends the minutes

[1662] The server automatically sends the minutes, including emotional data, to the email addresses of all participants.

[1663] Step 6: Integrate with a chatbot

[1664] 1. The server provides database access to the chatbot

[1665] The server grants the chatbot access to the conference database.

[1666] 2. Chatbots handle inquiries

[1667] The chatbot receives a search request from the user, retrieves the necessary data from the server, and responds.

[1668] Terminal handling

[1669] Step 1: Registering audio data

[1670] 1. The device records audio

[1671] The user records their own voice on the device.

[1672] 2. The device sends the voice data to the server

[1673] The device uploads the recorded audio data to the server.

[1674] Step 2: Sending audio data during a meeting

[1675] 1. The device records what you say and sends it to the server

[1676] The device records what is said during the meeting in real time and sends it to the server.

[1677] Step 3: Sending emotion data

[1678] 1. The device sends emotional data

[1679] The device uses an emotion engine to analyze emotions based on the recorded voice data and sends the data to a server.

[1680] Step 4: Review the meeting notes

[1681] 1. The device receives the minutes

[1682] After the meeting ends, the terminal receives the minutes sent from the server.

[1683] 2. The device displays the minutes.

[1684] The terminal displays the received minutes to the user.

[1685] User Action

[1686] Step 1: Registering audio data

[1687] 1. The user records the audio and saves it on the device.

[1688] The user records their own voice on the device.

[1689] 2. The user uploads the audio data to the server.

[1690] Upload the recorded audio data to the server.

[1691] Step 2: Speaking during the meeting

[1692] 1. User speaks

[1693] A user speaks in a conference.

[1694] 2. The user checks the recording

[1695] The user checks whether recording is working properly on the device.

[1696] Step 3: Review the meeting notes

[1697] 1. User checks the minutes

[1698] After the meeting ends, the user checks the minutes received from the terminal.

[1699] Step 4: Integrate with a chatbot

[1700] 1. Users interact with chatbots

[1701] A user uses the chatbot to search for past meeting minutes or specific statements.

[1702] 2. User checks the results

[1703] The user checks the results provided by the chatbot.

[1704] Example 2

[1705] 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."

[1706] Conventional meeting recording systems require manual creation of meeting minutes, which is time-consuming and labor-intensive. Furthermore, it is difficult to record the emotional state of each participant during the meeting, making it difficult to grasp the atmosphere of the meeting in detail. Therefore, there is a need to improve meeting efficiency and ensure transparency.

[1707] 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.

[1708] In this invention, the server includes means for registering and learning the voice data of each participant in advance, means for transcribing the voice in real time during the conference, means for removing unnecessary voice from the transcribed data, means for inputting the transcribed data and emotion data into a generative AI model to automatically summarize the conference content, means for automatically sending the summary data to the participants, means for searching past minutes in cooperation with a chatbot, and means for analyzing and saving emotion data in real time. This automates the creation of meeting minutes and enables detailed meeting records that take emotion data into account.

[1709] "Audio data" means digitally recorded information of a participant's speech.

[1710] "Emotion data" is information that indicates the emotional state of each speaker, analyzed from the voice data.

[1711] A "speech recognition model" is a machine learning or artificial intelligence algorithm for converting speech data into text data.

[1712] An "emotion engine" is an algorithm or system that analyzes audio and text data to identify emotional states.

[1713] "Transcription data" refers to data that has been converted from audio data into text format.

[1714] A "generative artificial intelligence model" is a machine learning or artificial intelligence model that generates new data based on input data.

[1715] "Unnecessary voice" refers to speech that is not related to the significance or content of the meeting (e.g., "umm," "ah," etc.).

[1716] "Summary data" is text data that extracts important points from all statements made at a meeting and summarizes them concisely.

[1717] A "chatbot" is an automated program that interacts with a user to provide or process information.

[1718] "Real-time" refers to the time frame in which data is generated and processed immediately, without delay.

[1719] A "database" is a system for efficiently storing, searching, and managing large amounts of data.

[1720] This invention is a system that automates the creation of meeting minutes and grasps the emotional states of participants in real time, thereby improving the transparency and effectiveness of meetings. This system consists of three main components: a server, a terminal, and a user, and is specifically implemented as follows.

[1721] First, users record their own voice data using their devices and upload it to the server. At this time, the recorded voice data is analyzed using an emotion engine to determine each user's emotional state. The server then trains a voice recognition model based on the received voice data. For example, it uses Google Cloud Speech-to-Text to convert the voice data into text and uses Hume AI to learn emotional characteristics.

[1722] When a meeting starts, the device records participants' speech in real time and sends it to the server. The server converts the received audio data into text in real time and simultaneously analyzes the emotional state of the speech using an emotion engine. This process uses Amazon Transcribe, for example, and stores the generated text data and emotion data in a database.

[1723] The server also uses natural language processing techniques to automatically remove unwanted sounds like "um" and "ah" from the text data. For example, it uses the SpaCy library to analyze the text and remove unnecessary parts. The clean text data is then stored in a database.

[1724] All spoken text and sentiment data are fed into a generative AI model to extract key points from the meeting and automatically generate a summary. OpenAI's GPT-4 can be used to generate summaries, using prompts such as "Please consider all spoken data from the meeting and its sentiment data, extract key points, and generate a summary of the meeting."

[1725] After the meeting, the server automatically sends the generated summary to participants, for example via email using the SendGrid API, which also includes sentiment data, allowing participants to understand the atmosphere of the meeting and any notable emotional reactions.

[1726] In addition, the server has the ability to search past meeting minutes in cooperation with chatbots. Users can use chatbots to efficiently search for meeting minutes containing specific statements or emotional states. For example, a chatbot can be built using Dialogflow and data can be searched using Elasticsearch.

[1727] The system automates meeting minutes creation and provides detailed meeting records that take emotional data into account, improving the transparency and effectiveness of meeting content and fostering deeper understanding and collaboration.

[1728] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1729] Step 1:

[1730] The user records their own voice data on the device and uploads it to the server.

[1731] Input: User recorded voice data

[1732] Specific operation: The user records their own voice using the recording app on their device and uploads the file to the server, where the uploaded data is stored.

[1733] Output: Audio data stored on the server

[1734] Step 2:

[1735] The server receives the speech data and trains a speech recognition model.

[1736] Input: Audio data uploaded from the device

[1737] How it works: The server converts voice data into text using Google Cloud Speech-to-Text or similar, and then uses an emotion engine such as Hume AI to analyze the emotional state from the voice data and saves the learning results in a database.

[1738] Output: Text data and emotion data corresponding to the audio data

[1739] Step 3:

[1740] The device records what is said during the meeting and sends it to the server in real time.

[1741] Input: Audio data during the meeting

[1742] Specific operation: The device records speech in real time and sends it to the server using WebSocket or similar.

[1743] Output: Audio data sent to the server in real time

[1744] Step 4:

[1745] The server transcribes the audio data in real time and analyzes the emotional state.

[1746] Input: Real-time audio data

[1747] How it works: The server uses Amazon Transcribe to convert the voice data into text, and simultaneously uses the emotion engine to analyze the emotional state. This data is then stored in a database in real time.

[1748] Output: Real-time text and sentiment data

[1749] Step 5:

[1750] The server removes unwanted audio from the transcription data.

[1751] Input: Real-time generated text data

[1752] Specific operation: The server uses the SpaCy library to analyze the text data and remove unnecessary sounds such as "um" and "ah." The cleaned text data is then restored to the database.

[1753] Output: Text data with unnecessary audio removed

[1754] Step 6:

[1755] The server uses a generative AI model to automatically summarize the key points of the meeting.

[1756] Input: Clean text data and sentiment data

[1757] Specific operation: The server uses OpenAI GPT-4 to generate a meeting summary from text data and emotion data, using the prompt "Consider all speech data and their emotion data from the meeting, extract key points, and generate a meeting summary." The generated summary is stored in a database.

[1758] Output: Summarized meeting data

[1759] Step 7:

[1760] After the conference ends, the server automatically sends summary data to the participants.

[1761] Input: Summarized meeting data

[1762] Specific operation: The server uses the SendGrid API to send the saved summary data to the participant's email address.

[1763] Output: Summary data sent to participants

[1764] Step 8:

[1765] The server works in conjunction with the chatbot to search for past minutes.

[1766] Input: User's search request

[1767] Specific operation: The server uses Dialogflow to build a chatbot, searches the database using Elasticsearch, and returns the search results and sentiment data to the user.

[1768] Output: Past meeting minutes data provided to the user

[1769] Through these processing steps, meeting minutes are automated and detailed meeting records are generated that take emotional data into account. This system improves the transparency and effectiveness of meetings, promoting deeper understanding and cooperation.

[1770] (Application example 2)

[1771] 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."

[1772] In conventional meeting recording systems, simply converting speech to text makes it difficult to accurately grasp the nuances of what is being said and the emotional state of the participants.In addition, in regular and emergency meetings within a factory, it is important to understand not only the main points of the meeting but also the emotional state of the participants, but previous systems were unable to perform real-time emotional analysis.

[1773] The identification processing 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 registering and learning the voice data of each participant in advance, means for transcribing the voice in real time during the conference, means for removing unnecessary voice from the transcribed data, means for automatically summarizing the conference content, means for analyzing the emotional state, means for automatically sending summary data to the participants, and means for searching past minutes in cooperation with a chatbot. This makes it possible to grasp the emotional state of the participants in real time in addition to summarizing the conference content, and to create and manage highly accurate minutes.

[1774] "Voice data of each participant" refers to voice information emitted by each participant who participates in a conference registered in the system.

[1775] The "means for pre-registration and learning" refers to a function that allows characteristic voice data of each participant to be input into the system before the start of the meeting, and for the voice recognition model to learn the voice characteristics of each participant based on that data.

[1776] "Means for transcribing audio in real time" is a function that instantly converts audio spoken during a meeting into text data.

[1777] "Means for removing unnecessary sounds from transcribed data" is a function for automatically removing meaningless sounds such as "eh" and "um" from transcribed text data.

[1778] The "means for automatically summarizing meeting content" is a function that analyzes what is said during a meeting, extracts important points, and generates a concise summary.

[1779] The "means for analyzing emotional states" is a function that analyzes the emotions contained in each statement in real time based on transcribed text data.

[1780] The "means for automatically transmitting summary data to participants" is a function for automatically transmitting the generated summary data to the communication addresses of all participants after the conference has ended.

[1781] "Means for searching past minutes in cooperation with a chatbot" refers to a chatbot function that searches past minutes or specific statements within them and provides relevant information in response to inquiries from users after a meeting.

[1782] This invention is a system that creates minutes and grasps the emotional state of participants in factory meetings. This system consists of three main components: a server, a terminal, and a user. We will explain how each component works.

[1783] Server Action:

[1784] 1. Register and train the audio data:

[1785] The server receives voice data uploaded by users in advance and trains a voice recognition model. The voice data contains the characteristic voice information of each participant, which improves the accuracy of voice recognition during the meeting. At the same time, it also learns the emotional state of each participant using an emotion engine.

[1786] 2. Real-time transcription and sentiment analysis:

[1787] When a meeting begins, the device records the audio and sends it to the server, which receives it in real time, transcribes it, and then uses an emotion engine to analyze the emotional state of each comment.

[1788] 3. Remove unwanted audio:

[1789] Using natural language processing technology, unnecessary sounds such as "um" and "ah" are automatically removed from the recorded text data, improving the quality of the transcription data.

[1790] 4. Automatic summary generation:

[1791] Based on the results of speech data and sentiment analysis, a generative AI model is used to automatically extract key points from meetings and generate summaries, including statements that elicited particularly strong sentiment.

[1792] 5. Sending minutes:

[1793] After the meeting ends, the server automatically sends the generated summary to the communication addresses of the participants, allowing the contents of the meeting to be shared quickly.

[1794] 6. Chatbot integration:

[1795] The server works in conjunction with the chatbot to search past meeting minutes in response to user questions and provide the necessary information. Emotional data can also be searched, allowing users to check specific statements and their emotional state at the time.

[1796] Terminal handling:

[1797] 1. Audio data registration:

[1798] Users use their devices to record and register their own voice and emotions, and this data is uploaded to a server and used to train the voice recognition model and emotion engine.

[1799] 2. Sending voice and emotion data:

[1800] During a meeting, the device records the user's speech and emotions in real time and transmits them to the server, which then uses the received data for immediate transcription and sentiment analysis.

[1801] 3. Check the minutes:

[1802] After the meeting ends, the device receives the minutes sent from the server and displays them to the user, allowing the user to check the details of the meeting and the emotional state at the time.

[1803] User Action:

[1804] 1. Audio data registration:

[1805] Users use their devices to record and register their own voice and emotions, which allows the voice recognition model and emotion engine to work more accurately.

[1806] 2. Record statements and emotions during the meeting:

[1807] Users simply join a meeting as usual and record their comments and emotions on their device, with no special operations required.

[1808] 3. Check the minutes:

[1809] After the meeting, users can review the minutes received from their devices and quickly grasp the main points by utilizing the system's automatic summarization and sentiment analysis.

[1810] 4. Chatbot integration:

[1811] Through the chatbot, users can search past meeting minutes and specific statements. They can also search for emotion data, allowing them to check specific statements and the emotional state of those statements.

[1812] Examples:

[1813] During regular meetings held within a factory, an assistant robot summarizes the meeting and analyzes sentiment. The minutes are then sent to participants' mailboxes after the meeting. For example, if a participant asks the chatbot, "What were the most common opinions expressed in last month's meeting?", the robot can provide a summary along with past sentiment.

[1814] Example prompt sentence:

[1815] "What were the most common opinions and comments that generated a lot of emotion at last month's meeting?"

[1816] As described above, the system of the present invention can significantly improve the efficiency of meetings within a factory.

[1817] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1818] Step 1: Register and train audio data

[1819] Users use their devices to record their own voice and emotions and upload them to the server. The server then trains the received voice data based on a speech recognition model. The input is the voice data and emotion data from the users, and the server learns from this to understand the voice and emotion features of each participant. The output is the speech recognition model after training is complete.

[1820] Step 2: Real-time transcription and sentiment analysis

[1821] The device records the audio during the meeting in real time and sends this audio data to a server, which then instantly transcribes the audio data and performs sentiment analysis using a generative AI model.The input is the audio data from the meeting, and the server outputs text data and the emotional state of each utterance.

[1822] Step 3: Remove unwanted audio

[1823] The server uses natural language processing technology to automatically remove unnecessary sounds such as "eh" and "um" from the generated text data. The input is transcribed text data, and unnecessary sounds are removed as part of the data processing. The output is clear transcribed text data.

[1824] Step 4: Automatically generate a summary

[1825] The server uses the transcription data and the results of sentiment analysis to automatically summarize the meeting content based on a generative AI model. The input is clear transcription text data and sentiment data, which are then processed into a summary using the generative AI model. The output is the summarized text data.

[1826] Step 5: Sending meeting notes

[1827] After the conference ends, the server automatically sends the generated summary data to the participants' communication addresses. The input is the summarized text data and the participants' communication addresses, and the server calculates and sends this as data in the form of email. The output is the email sent to the participants.

[1828] Step 6: Integrate with a chatbot

[1829] The user uses a chatbot from their device to search for past minutes or specific statements. The server searches for past minutes and emotion data via the chatbot and provides them to the user. The input is a search query (prompt sentence) from the user, and the server calculates and outputs the appropriate minutes and emotion data from the database. The output is the search results provided to the user.

[1830] 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.

[1831] 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.

[1832] 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.

[1833] 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.

[1834] 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.

[1835] 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.

[1836] 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).

[1837] 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.

[1838] 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."

[1839] 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.

[1840] 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).

[1841] 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.

[1842] 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.

[1843] 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.

[1844] 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.

[1845] 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.

[1846] 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.

[1847] 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.

[1848] 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.

[1849] 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.

[1850] 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.

[1851] The following is further disclosed regarding the above embodiment.

[1852] (Claim 1)

[1853] A means of registering and learning the voice data of each participant in advance,

[1854] A means to transcribe audio in real time during meetings,

[1855] a means for removing unwanted audio from the transcription data;

[1856] means for automatically summarizing meeting content;

[1857] an automated means of sending summary data to participants;

[1858] A system that includes a means to search past meeting minutes in cooperation with a chatbot.

[1859] (Claim 2)

[1860] 10. The system of claim 1, wherein audio data transmitted during a conference is transcribed in real time and the speaker is identified.

[1861] (Claim 3)

[1862] 2. The system according to claim 1, wherein the automatically summarized minutes data is sent to the email addresses of all participants.

[1863] "Example 1"

[1864] (Claim 1)

[1865] A means of registering and learning the voice data of each participant in advance,

[1866] A means to transcribe audio in real time during meetings,

[1867] a means for removing unwanted audio from the transcription data;

[1868] means for automatically summarizing meeting content;

[1869] an automated means of sending summary data to participants;

[1870] A system that includes a means to search past meeting minutes in cooperation with a chatbot.

[1871] (Claim 2)

[1872] 10. The system of claim 1, wherein real-time recorded audio data is transmitted through multiple devices, transcribed through speech recognition, and specific speakers are identified.

[1873] (Claim 3)

[1874] 10. The system of claim 1, wherein the generated summary data is sent to all participants via email.

[1875] "Application Example 1"

[1876] (Claim 1)

[1877] A means for registering and learning the voice data of each worker in advance;

[1878] A means to transcribe audio in real time while you work,

[1879] a means for removing unwanted audio from the transcription data;

[1880] a means for automatically summarizing work;

[1881] a means for automatically transmitting the summary data to an administrator;

[1882] A system that includes a means for searching past work records in cooperation with a chatbot.

[1883] (Claim 2)

[1884] The system according to claim 1, which transcribes voice data transmitted during work in real time and identifies the speaker.

[1885] (Claim 3)

[1886] 2. The system of claim 1, wherein the automatically summarized work record data is sent to an administrator's email address.

[1887] "Example 2: Combining Emotion Engines"

[1888] (Claim 1)

[1889] A means of registering and learning the voice data of each participant in advance,

[1890] A means to transcribe audio in real time during meetings,

[1891] a means for removing unwanted audio from the transcription data;

[1892] A means for automatically summarizing the meeting content by inputting the transcription data and emotion data into a generative artificial intelligence model;

[1893] an automated means of sending summary data to participants;

[1894] A means to search past minutes in conjunction with a chatbot,

[1895] A system including a means for analyzing and storing emotion data in real time.

[1896] (Claim 2)

[1897] 10. The system of claim 1, wherein audio data transmitted during a conference is transcribed in real time and the speaker is identified.

[1898] (Claim 3)

[1899] 2. The system according to claim 1, wherein the automatically summarized minutes data is sent to the email addresses of all participants.

[1900] "Application example 2 when combining emotion engines"

[1901] (Claim 1)

[1902] A means of registering and learning the voice data of each participant in advance,

[1903] A means to transcribe audio in real time during meetings,

[1904] a means for removing unwanted audio from the transcription data;

[1905] means for automatically summarizing meeting content;

[1906] a means for analyzing emotional states;

[1907] an automated means of sending summary data to participants;

[1908] A system that includes a means to search past meeting minutes in cooperation with a chatbot.

[1909] (Claim 2)

[1910] 10. The system of claim 1, wherein audio data transmitted during a conference is transcribed in real time and the speaker is identified.

[1911] (Claim 3)

[1912] 2. The system of claim 1, wherein the automatically summarized minutes data is sent to the communication addresses of all participants. [Explanation of symbols]

[1913] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. A means of registering and learning the voice data of each participant in advance, A means to transcribe audio in real time during meetings, a means for removing unwanted audio from the transcription data; means for automatically summarizing meeting content; an automated means of sending summary data to participants; A system that includes a means to search past meeting minutes in cooperation with a chatbot.

2. The system of claim 1 , wherein the system transcribes voice data transmitted during a conference in real time and identifies the speaker.

3. The system according to claim 1, wherein the automatically summarized minutes data is sent to the email addresses of all participants.

Images