System

Abstract

A system is provided.SOLUTION: A system including means for recording voice data of a user, means for uploading the recorded voice data to a server, means for preprocessing the uploaded voice data and converting the voice data into text data, means for performing machine learning based on the voice data and the text data and learning a speech characteristic of the user, means for generating a dialogue in real time by the server and transmitting the dialogue to a terminal after the user dies, and means for reproducing the generated dialogue as voice by the terminal.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] In recent years, dialogue systems using AI technology have been attracting attention, but few systems can continue to dialogue with a deceased user. Furthermore, previous dialogue systems have had difficulty generating natural dialogue that reflects the user's speech characteristics and personality. Therefore, there is a need for a system that can realize dialogue that reflects the user's personality and provides a higher-quality dialogue experience. [Means for solving the problem]

[0005] The present invention provides a system that records a user's voice data, uploads the recorded voice data to a server, and preprocesses the uploaded voice data and converts it into text data. It also provides a system that performs machine learning based on the voice data and text data to learn the user's speech characteristics, and a system that, after the user's death, generates a dialogue in real time from the server and transmits it to a terminal. Furthermore, the system includes a device that plays back the generated dialogue as speech, enabling dialogue that reflects the user's individuality. Furthermore, the system includes a means for noise reduction and volume normalization in the preprocessing of the voice data, and a means for regularly updating the machine learning model, enabling consistently high-quality, natural dialogue to be provided.

[0006] "User" refers to the entity that uses this system and provides voice data.

[0007] "Voice data" refers to data that is a digital recording of what a user has said.

[0008] "Recording means" refers to a device or application that records a user's voice and stores it as digital data.

[0009] "Server" refers to a computer system that receives, processes, stores, and provides audio data over a network.

[0010] "Means of uploading" refers to the process or technology used to transfer a user's voice data from a device to a server.

[0011] "Preprocessing" refers to the process of removing noise from audio data, normalizing volume, and other optimizations.

[0012] "Text data" refers to data obtained by converting voice data into character information.

[0013] "Machine learning" refers to a technique in which computers use large amounts of data to learn specific patterns and trends.

[0014] "Speech characteristics" refers to characteristics such as a user's unique speaking style, vocabulary, pronunciation, and intonation.

[0015] "Means for generating dialogue in real time" refers to technologies and processes that instantly generate dialogue responses based on the user's past speech data.

[0016] "Terminal" refers to a device used by a user or their family that records and plays back audio data.

[0017] "Means for playing as audio" refers to technology that converts text data into audio and outputs it from a device. [Brief explanation of the drawings]

[0018] [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

[0019] 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.

[0020] First, the terms used in the following description will be explained.

[0021] 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).

[0022] 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.

[0023] 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.

[0024] 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.

[0025] 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."

[0026] [First embodiment]

[0027] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0028] 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.

[0029] 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).

[0030] 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.

[0031] 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.

[0032] 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.

[0033] 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.

[0034] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0035] 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.

[0036] 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.

[0037] 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.

[0038] 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."

[0039] This invention is a system that records the conversations and voices of a user while they are alive, and uses AI to learn from them, allowing for conversations with the user even after they have passed away. This system includes elements such as the user, a terminal, and a server, each of which plays a specific role.

[0040] Audio data recording

[0041] The user launches a dedicated recording app and leaves a voice message. For example, they can record a message like, "Today, I went on a trip with my family. It was a lot of fun." This recording reflects the user's personality and speech characteristics.

[0042] Uploading and preprocessing audio data

[0043] The device records this voice data, converts it into an appropriate format (e.g., WAV or MP3), and uploads it to a server. The uploaded voice data undergoes preprocessing, such as noise reduction and volume normalization, on the server. After preprocessing, the voice data is converted into text data using speech recognition technology.

[0044] Training a machine learning model

[0045] The server uses the voice data and corresponding text data to train a machine learning model. The model learns the user's speech characteristics and conversation style and is able to generate natural, authentic responses. The server also periodically incorporates newly uploaded data to improve the accuracy of the model.

[0046] Generating conversations after a user dies

[0047] After a user's death, a family member or friend can use the device to launch a conversation app. The device records the family member or friend's speech and sends the audio data to a server. The server analyzes the audio data and generates an appropriate response in real time. The generated response is sent to the device as text.

[0048] Dialogue playback

[0049] The device converts the text data received from the server into speech and plays back a response on behalf of the user. This allows family and friends to feel as if they are being spoken by the user. For example, if a family member asks, "How was your day?", the device will respond in a voice that sounds like the user, such as, "I went shopping today."

[0050] Specific examples

[0051] 1. The user launches the app and records a voice message saying, "I went to the park with my dog ​​today. It was fun!"

[0052] 2. The device records this audio and uploads it to the server.

[0053] 3. The server removes noise and normalizes the volume of the audio data, generating text data such as "I went to the park with my dog ​​today. It was fun!"

[0054] 4. The server uses a machine learning model to learn the user's speech characteristics.

[0055] 5. One day, a family member uses a conversation app to ask, "How's your day going?"

[0056] 6. The device sends the audio to the server, which generates a response saying, "I went for a walk today. The weather was nice."

[0057] 7. The device will play back the response as audio, providing a natural conversational experience for family members.

[0058] By combining the user's voice data from before their death with machine learning technology, the present invention makes it possible to realize natural conversations that sound like the user, even after the user has passed away.

[0059] The processing flow will be explained below.

[0060] Step 1:

[0061] The user launches a recording app and starts speaking. For example, the user might record a voice message such as, "Today, my family and I went to the park."

[0062] Step 2:

[0063] The device detects that the record button has been pressed and starts recording audio data. The recorded data is temporarily saved in local storage.

[0064] Step 3:

[0065] The device will recognize that the end recording button has been pressed and will stop recording. The recorded data will be converted to a digital format (e.g., WAV or MP3 format).

[0066] Step 4:

[0067] The device uploads the recording data to a server using a secure communication protocol (e.g., HTTPS).

[0068] Step 5:

[0069] The server receives the uploaded audio data, stores it in a database, and starts preprocessing the audio data.

[0070] Step 6:

[0071] The server performs noise reduction and volume normalization on the audio data, which improves the audio quality.

[0072] Step 7:

[0073] The server uses speech recognition technology to convert the preprocessed speech data into text data, for example, "Today, my family and I went to the park."

[0074] Step 8:

[0075] The server inputs the generated text data and corresponding audio data into a machine learning model, which begins the process of learning the user's speech characteristics.

[0076] Step 9:

[0077] The server trains the machine learning model, learning the user's speech patterns and characteristics to generate highly accurate responses.

[0078] Step 10:

[0079] The server evaluates the machine learning model and retrains it as needed, thereby maintaining the model's accuracy.

[0080] Step 11:

[0081] After the death of the user, a family member or friend can use the device to launch the conversation app and press the "Start conversation" button to begin the conversation.

[0082] Step 12:

[0083] The device records voice input from family and friends and sends the voice data to a server, for example, recording a question like "How was your day?"

[0084] Step 13:

[0085] The server analyzes the received voice data and converts it into text. Based on the converted data, it uses a machine learning model to generate an appropriate response.

[0086] Step 14:

[0087] The server sends the generated response text to the terminal, for example, generating a response such as "I went shopping today."

[0088] Step 15:

[0089] The device converts the response text received from the server into voice and plays it back to the family, enabling natural conversation.

[0090] By taking specific actions at each step, the system can provide natural dialogue that reflects the user's personality even after the user has passed away.

[0091] Example 1

[0092] 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."

[0093] There is a lack of technology to respond to the requests of family and friends who want to continue interacting with a deceased person. Furthermore, insufficient preprocessing of recorded audio data and insufficient training of machine learning models can lead to poor dialogue quality. Furthermore, it is difficult to accurately model the user's speech characteristics and generate natural dialogue.

[0094] 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.

[0095] In this invention, the server includes means for preprocessing the voice data, removing noise and normalizing the volume, and converting the voice data into text data, means for performing machine learning based on the voice data and text data to learn the user's speech characteristics, and means for the server to generate a dialogue in real time after the user's death and transmit it to the terminal, thereby enabling family and friends to experience natural dialogue even after the user's death.

[0096] "Voice data" is digital data that records a user's speech or voice message.

[0097] A "terminal" is an electronic device on which a recording application or a dialogue application used by a user is installed.

[0098] "Convert to format" refers to the process of converting audio data into an appropriate audio file format (e.g., WAV, MP3).

[0099] "Upload" refers to sending data recorded on a terminal to a server via a network.

[0100] A "server" is a central management computer that stores received data and performs various data processing and machine learning.

[0101] "Preprocessing" refers to performing data cleansing operations on the audio data, such as noise removal and volume normalization.

[0102] "Noise reduction" is a process for removing unnecessary background sounds and noise from audio data.

[0103] "Volume normalization" is a process for making the volume level of audio data uniform.

[0104] "Text data" is voice data converted into character data using voice recognition technology.

[0105] A "machine learning model" is an algorithm designed to learn from large amounts of data and perform a specific task.

[0106] "Speech characteristics" refers to the voice characteristics and speaking style of an individual user.

[0107] "Generating dialogue in real time" refers to generating and providing responses instantly as users take turns speaking.

[0108] "Converting to speech" means regenerating text data as speech using speech synthesis technology.

[0109] "Playback" means outputting the generated audio from a playback device such as a speaker.

[0110] "Family and friends" refers to people with whom the user will communicate after their death.

[0111] This invention is a system that records the conversations and voices of a user while they are alive, and uses AI to learn from them, allowing for conversations with the user even after they have passed away. This system includes elements such as the user, a terminal, and a server, each of which plays a specific role.

[0112] First, the user launches a dedicated recording app and leaves a voice message. This voice message reflects the user's personality and speech characteristics and includes specific content such as, "Today, I went on a trip with my family. It was a lot of fun."

[0113] The device then records this audio data, converts it into an appropriate format (e.g., WAV or MP3), and uploads it to a server via the Internet. The audio data format conversion is performed using software installed on the device (e.g., FFMPEG).

[0114] The server performs preprocessing on the uploaded audio data, such as noise removal and volume normalization. This preprocessing uses a noise removal algorithm (e.g., Audacity's noise removal function) and a volume normalization algorithm. After preprocessing is complete, the server converts the audio data into text data using speech recognition technology (e.g., Google Speech-to-Text API). For example, the generated text data is, "Today, I went on a trip with my family. It was a lot of fun."

[0115] The server then uses the voice data and corresponding text data to train a machine learning model (e.g., OpenAI GPT-4). The model learns the user's speech characteristics and conversational style and is able to generate natural, authentic responses. The server also periodically incorporates newly uploaded data to improve the model's accuracy.

[0116] After a user passes away, family and friends can use the device to launch a conversation app. When they speak into the microphone, the device records their voice and sends it to the server in real time. For example, if a family member asks, "How was your day?", the server receives the voice data and converts it into text using speech recognition technology. Then, a machine learning model generates a response such as, "I went for a walk today. The weather was nice."

[0117] The generated response is sent as text to the device, which then converts this text data into speech (e.g., using Amazon Polly to generate synthetic speech), and finally plays the generated speech through the device's speaker, giving family and friends the feeling that it is the user speaking.

[0118] Examples of prompt statements

[0119] Prompt: Convert the following audio data into text and use that text to generate a natural-sounding response based on the user's speaking characteristics.

[0120] Audio data: I went to the park with my dog ​​today. It was fun!

[0121] By combining a user's voice data with machine learning technology, this invention enables a user to have natural, lifelike conversations even after they have passed away. This system provides a way for family and friends to continue to relive precious memories.

[0122] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0123] Step 1:

[0124] The user launches a recording app and records a voice message. The user speaks into the microphone, saying, "Today, I went on a trip with my family. It was a lot of fun." The recording app captures this voice data in WAV format and saves it on the device.

[0125] Input: User's voice

[0126] Output: WAV format audio file

[0127] Step 2:

[0128] The device converts the stored audio data into MP3 format using software installed on the device (e.g., FFMPEG), and then uploads the converted MP3 file to a server via the Internet.

[0129] Input: WAV format audio file

[0130] Output: MP3 audio file

[0131] Step 3:

[0132] The server receives the uploaded audio data and stores it in storage. Then, preprocessing begins. Specifically, the server performs noise reduction and volume normalization on the audio data. This preprocessing uses a noise reduction algorithm (e.g., Audacity's noise reduction function) and a volume normalization algorithm.

[0133] Input: MP3 audio file

[0134] Output: Preprocessed audio file

[0135] Step 4:

[0136] The server uses the preprocessed voice data to convert it into text data using speech recognition technology (e.g., Google Speech-to-Text API). For example, the generated text data is, "Today, I went on a trip with my family. It was a lot of fun."

[0137] Input: Preprocessed audio file

[0138] Output: Text data

[0139] Step 5:

[0140] The server trains a machine learning model (e.g., OpenAI GPT-4) using the voice data and corresponding text data. Through this, the server learns the user's speech characteristics and interaction style. The server also periodically incorporates newly uploaded data and continuously retrains the model to improve its accuracy.

[0141] Input: Text and audio data

[0142] Output: A trained machine learning model

[0143] Step 6:

[0144] After the death of the user, family and friends can launch a conversation app and speak into the microphone, for example, saying, "How was your day?" The device will record the voice and send it to the server in real time.

[0145] Input: Voice of family or friends

[0146] Output: Recorded audio data

[0147] Step 7:

[0148] The server converts the received voice data into text data using speech recognition technology. For example, the text "How was your day?" is generated. Then, using a machine learning model trained on the server, it generates a response such as "I went for a walk today. The weather was nice."

[0149] Input: Recorded audio data

[0150] Output: Response text data

[0151] Step 8:

[0152] The device converts the response text data received from the server into speech, using speech synthesis technology (e.g., Amazon Polly) to generate a speech that says, "I went for a walk today. The weather was nice."

[0153] Input: Response text data

[0154] Output: Audio data

[0155] Step 9:

[0156] The device then plays the generated audio through the speaker, allowing family and friends to hear it as if you were actually speaking.

[0157] Input: Audio data

[0158] Output: Played audio

[0159] (Application example 1)

[0160] 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."

[0161] As the aging society progresses, many people want to cherish memories and conversations with their deceased loved ones. However, existing technology makes it difficult to realize natural conversations with the deceased. It is necessary to solve this problem and provide a new customer experience that enriches memories with the deceased.

[0162] 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.

[0163] In this invention, the server includes means for recording user voice data, means for uploading the recorded voice data to the server, means for preprocessing the uploaded voice data and converting it into text data, means for performing machine learning based on the voice data and text data to learn the user's speech characteristics, means for the server to generate a dialogue in real time after the user's death and transmit it to the terminal, means for the terminal to play back the generated dialogue as audio and provide a customer dialogue experience, and means for providing the customer with a dialogue experience with the deceased via a head-mounted display or smartphone, thereby allowing customers to experience a natural dialogue with the deceased and enrich their memories.

[0164] "User voice data" refers to voice information uttered by a user, and is digital data that can be recorded and analyzed.

[0165] A "server" is a computer system that stores, processes, and manages data on a network, and is responsible for managing user voice data and training machine learning models.

[0166] "Recording means" refers to the functionality of the device or software for recording and storing a user's voice data.

[0167] "Means for uploading to a server" refers to the functionality of a device or software for transmitting recorded audio data to a server via a network.

[0168] "Preprocessing" refers to the process of removing noise and normalizing the volume of uploaded audio data to make it easier to analyze.

[0169] "Means for converting into text data" refers to the technology and functions for analyzing voice data and converting it into text data.

[0170] "Means for conducting machine learning to learn a user's speech characteristics" refers to technologies and functions that use voice data and text data to train a model on a user's speech patterns and characteristics.

[0171] "Means for generating dialogue in real time" refers to technologies and functions that analyze dialogue with family and friends on the spot and instantly generate appropriate responses.

[0172] "Terminal" refers to devices used by users, such as computers, smartphones, and head-mounted displays.

[0173] "Means for playing as audio" refers to the technology or functionality for converting text data into audio and playing it back on a user device.

[0174] "Means for providing a customer interaction experience" refers to technologies and functions that provide natural and individual interactions with users via terminals and realize interactions with users.

[0175] A "head-mounted display" is a display device worn by a user that provides information visually and audibly.

[0176] A "smartphone" is a portable information terminal that combines the functions of a mobile phone with those of a personal computer, and is a device that can record, upload, and play back audio data.

[0177] The present invention provides a system for providing a conversational experience even after a user has passed away by recording the user's voice data and training a machine learning model based on the recorded voice data. This system is particularly applicable to customer conversational experiences using head-mounted displays and smartphones. The components and processing steps of this system are described in detail below.

[0178] Recording and uploading audio data

[0179] Users use a dedicated recording app to record everyday conversations and messages. This recording is done via a device such as a smartphone. The recorded voice data is saved on the device and then uploaded to a server.

[0180] Audio data preprocessing and text conversion

[0181] The server performs preprocessing on the uploaded audio data, removing noise and normalizing the volume, making the data easier to analyze. After preprocessing, the audio data is converted into text using speech recognition technology.

[0182] Training a machine learning model

[0183] The server uses the preprocessed audio data and corresponding text data to train a machine learning model. The model learns the user's speech characteristics and conversational style and generates natural-sounding responses. This allows the model to reproduce a user's conversations in the future, even after the user has passed away.

[0184] Dialogue generation and playback

[0185] After the death of a user, family and friends launch a dedicated conversation app on a head-mounted display or smartphone. The device records the family and friends' speech and sends the audio data to a server. The server analyzes the audio data in real time and generates an appropriate response. The generated response is sent as text data to the device, which converts it into audio and plays it back.

[0186] Specific examples

[0187] 1. A user uses a dedicated app to record a voice message saying, "I went to the park with my dog ​​today. It was fun!"

[0188] 2. The device uploads this audio to the server.

[0189] 3. The server performs preprocessing and generates text data such as "I went to the park with my dog ​​today. It was fun!"

[0190] 4. The server uses a machine learning model to learn the user's speech characteristics.

[0191] 5. One day, a family member uses a conversation app to ask, "How's your day going?"

[0192] 6. The device sends the audio to the server, which generates a response saying, "I went for a walk today. The weather was nice."

[0193] 7. The device will play back the response as audio, providing a natural conversational experience for family members.

[0194] Prompt Sentence Examples

[0195] "User-recorded question: 'Mom, what would you recommend today?'"

[0196] In this way, this invention combines the user's voice data from before death with machine learning technology to enable natural conversations that sound like the user even after the user has passed away. Furthermore, this system is expected to have a wide range of applications when used with head-mounted displays and smartphones.

[0197] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0198] Step 1:

[0199] The user launches a dedicated recording app and records a voice message. The input is the user's speech, and the output is the recorded voice data. Specifically, the user speaks into the smartphone's microphone and records a message such as, "Today, I went on a trip with my family. It was a lot of fun."

[0200] Step 2:

[0201] The device converts the recorded audio data into an appropriate format (e.g., WAV or MP3) and uploads it to the server. The input is the recorded audio data, and the output is the uploaded audio data. Specifically, the smartphone converts the recorded audio data and sends it to the server via the Internet.

[0202] Step 3:

[0203] The server performs preprocessing on the uploaded audio data, such as noise reduction and volume normalization. The input is the uploaded audio data, and the output is the preprocessed audio data. Specifically, the server runs the audio processing algorithm and processes the data to make it easier to analyze.

[0204] Step 4:

[0205] The server converts the preprocessed speech data into text data using speech recognition technology. The input is the preprocessed speech data, and the output is text data. The specific operation is to extract strings of characters from the speech signal using a speech recognition model (e.g., Wav2Vec2).

[0206] Step 5:

[0207] The server trains a machine learning model using voice data and the corresponding text data. The input is voice data and text data, and the output is a trained machine learning model. The specific operation is to input data into the model, and iteratively learns the user's speech characteristics.

[0208] Step 6:

[0209] After the death of a user, family and friends can launch a dialogue app and ask questions by voice. The input is the speech of the family or friend, and the output is recorded voice data. Specifically, family and friends use a smartphone or head-mounted display to record questions such as "How was your day?"

[0210] Step 7:

[0211] The device sends the audio data to the server. The input is the recorded audio data, and the output is the audio data uploaded to the server. The specific operation is that the smartphone sends the recorded audio data to the server.

[0212] Step 8:

[0213] The server analyzes the voice data in real time and generates an appropriate response. The input is the uploaded voice data, and the output is the generated text response. The specific operation is to use the speech recognition and generation AI model on the server to generate a response corresponding to the input voice data.

[0214] Step 9:

[0215] The device converts the text response received from the server into speech, recreating a natural conversational experience even after the user has passed away. The input is the generated text response, and the output is a speech response. The specific operation is that the smartphone or head-mounted display plays the text as speech using speech synthesis technology (e.g., Text-to-Speech).

[0216] 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.

[0217] This invention is a system that records a user's conversations and voices while they are alive and uses AI to learn from them, enabling conversations with the deceased person even after they have passed away. This system includes the user, terminal, and server elements, each of which plays a specific role. Furthermore, by combining it with an emotion engine, the system aims to provide natural conversations that correspond to the user's emotions.

[0218] Audio data recording

[0219] A user launches a dedicated recording app and leaves a voice message. For example, the user might record a message like, "Today, I went on a trip with my family. It was a lot of fun." This recording reflects the user's personality and speech characteristics.

[0220] Uploading and preprocessing audio data

[0221] The device records this voice data, converts it into an appropriate format (e.g., WAV or MP3), and uploads it to a server. The uploaded voice data undergoes preprocessing, such as noise reduction and volume normalization, on the server. After preprocessing, the voice data is converted into text data using speech recognition technology.

[0222] Training a machine learning model

[0223] The server uses the voice data and corresponding text data to train a machine learning model. The model learns the user's speech characteristics and conversation style and is able to generate natural, authentic responses. The server also periodically incorporates newly uploaded data to improve the accuracy of the model.

[0224] Emotion recognition by emotion engine

[0225] The server uses an emotion engine to extract emotional information from the user's voice data. This emotional information is used as additional data when the machine learning model learns the user's speech characteristics. By taking emotional information into account, more natural and emotionally relevant responses are generated.

[0226] Generating conversations after a user dies

[0227] After a user passes away, a family member or friend launches a conversation app on the device. The device records the family member or friend's speech and sends the audio data to a server. The server analyzes the audio data and converts it into text. After conversion, an emotion engine is used to analyze emotions and generate an appropriate response in real time. The generated response is sent to the device as text.

[0228] Dialogue playback

[0229] The device converts the text data received from the server into voice and plays back the response on behalf of the user. This allows family and friends to feel as if they are being spoken by the user. For example, if a family member asks, "How was your day?", the device will play back a voice response that sounds like the user, such as, "I went shopping today. The weather was nice, so it felt good."

[0230] Specific examples

[0231] 1. The user launches the app and records a voice message saying, "I went to the park with my dog ​​today. It was fun!"

[0232] 2. The device records this audio and uploads it to the server.

[0233] 3. The server removes noise and normalizes the volume of the audio data, generating text data such as "I went to the park with my dog ​​today. It was fun!"

[0234] 4. The server uses an emotion engine to extract emotion information from the voice data.

[0235] 5. The server uses a machine learning model to learn the user's emotional information along with their speech characteristics.

[0236] 6. One day, a family member uses a conversation app to ask, "How's your day going?"

[0237] 7. The device sends the audio to the server, which generates a response saying, "I went for a walk today. I was happy because the weather was nice."

[0238] 8. The device will play back the response as audio, providing a natural conversational experience for family members.

[0239] By combining the user's voice data from their lifetime with machine learning technology and an emotion engine, this invention makes it possible to realize natural conversations that reflect the user's personality and are sensitive to their emotions even after they have passed away.

[0240] The processing flow will be explained below.

[0241] Step 1:

[0242] The user starts a dedicated recording application and records a voice message. For example, the user might record a message such as, "Today, I went on a trip with my family. It was a lot of fun."

[0243] Step 2:

[0244] The device detects that the record button has been pressed and starts recording audio data. The recorded data is temporarily saved in local storage.

[0245] Step 3:

[0246] The device will recognize that the end recording button has been pressed and will stop recording. The recorded data will be converted to a digital format (e.g., WAV or MP3 format).

[0247] Step 4:

[0248] The device uploads the recording data to a server using a secure communication protocol (e.g., HTTPS).

[0249] Step 5:

[0250] The server receives the uploaded voice data, stores it in a database, confirms receipt of the data, and moves on to the next processing step.

[0251] Step 6:

[0252] The server starts pre-processing the audio data, performing noise reduction and volume normalization to improve the audio quality.

[0253] Step 7:

[0254] The server uses speech recognition technology to convert the preprocessed speech data into text data, for example, "Today, I went on a trip with my family."

[0255] Step 8:

[0256] The server uses an emotion engine to extract the user's emotional information from the voice data. For example, it recognizes positive emotions from voice characteristics such as "fun."

[0257] Step 9:

[0258] The server inputs voice data, text data, and emotional information into the machine learning model, which then learns responses based on the user's speech characteristics and emotions.

[0259] Step 10:

[0260] The server trains the machine learning model to generate natural-looking dialogue that reflects the user's emotions, and periodically ingests new data to update the model.

[0261] Step 11:

[0262] After the death of the user, a family member or friend can use the device to launch the conversation app and press the "Start conversation" button to begin the conversation.

[0263] Step 12:

[0264] The device records voice input from family and friends and sends the voice data to a server, for example, recording questions like "How was your day?"

[0265] Step 13:

[0266] The server analyzes the received voice data and converts it into text. After conversion, an emotion engine is used to extract emotional information and generate an appropriate response in real time.

[0267] Step 14:

[0268] The server sends the generated response text to the terminal, for example, "I went shopping today. The weather was nice, so it felt good."

[0269] Step 15:

[0270] The device converts the response text received from the server into voice and plays it back to the family, enabling natural conversation.

[0271] By taking specific actions at each step, the system is able to provide natural, emotionally sensitive conversations that reflect the user's personality, even after the user has passed away.

[0272] Example 2

[0273] 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."

[0274] Currently, there is no system that records a user's conversations and voices while they are alive and allows them to have a conversation with the user even after they have passed away. In particular, it is difficult to provide natural conversations that reflect the user's speech characteristics and emotions. The purpose of this invention is to provide natural conversations that reflect the user's personality and are in tune with their emotions even after the user has passed away.

[0275] 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.

[0276] In this invention, the server includes means for recording user voice data, means for converting the format of the recorded voice data by the terminal and uploading it to the server, means for preprocessing the uploaded voice data and converting it into text data, means for performing machine learning based on the voice data and text data to learn the user's speech characteristics and emotional information, means for extracting emotional information from the voice data using an emotion engine and using it as training data for a machine learning model, means for the server to generate a dialogue in real time after the user's death and send it to the terminal, and means for the terminal to play back the generated dialogue as audio. This makes it possible to provide natural dialogue that reflects the user's individuality and is in tune with their emotions.

[0277] A "user" is an individual who provides voice data to the system.

[0278] "Audio data" refers to data of a voice message generated by a user using a recording application.

[0279] A "terminal" is an electronic device that a user uses to record audio data and upload it to a server.

[0280] The "server" is a central processing unit responsible for preprocessing voice data, training machine learning models, generating dialogue, and more.

[0281] "Format conversion" is the process of changing audio data into an appropriate format, such as WAV or MP3 format.

[0282] "Noise reduction" is a process for removing unnecessary background noise contained in audio data.

[0283] "Volume normalization" is a process for making the volume level of audio data uniform.

[0284] "Text data" is voice data that has been analyzed and converted into text information.

[0285] "Machine learning" is the process of using data to train a model and learn patterns and rules from the data.

[0286] "Speech characteristics" refer to the characteristics of a user's voice and speaking style.

[0287] "Emotion information" is data that expresses the user's emotions and feelings extracted from the voice data.

[0288] An "emotion engine" is a technology or system for extracting emotional information from voice data.

[0289] "Dialogue generation" is the process of creating natural responses and dialogue based on data learned by a machine learning model.

[0290] "Real-time" refers to time characteristics that result in near-instant processing and response.

[0291] "Audio playback" means converting text data into audio and outputting it as if the user were speaking.

[0292] This invention is a system that records a user's voice data while they are alive and uses a machine learning model to learn from it, enabling conversations that sound like the user even after they have passed away. This system mainly uses the user, a terminal, and a server. The specific operation of each element is described below.

[0293] First, the user launches a dedicated recording app and records a voice message. For example, a user might record a message like, "Today, I went on a family trip. It was a lot of fun." This voice data reflects the user's personality and speech characteristics in detail. The recording app runs on a device such as a smartphone or tablet, and the user taps the "Start Recording" button to start recording and the "Stop Recording" button to end recording.

[0294] The device then converts the recorded audio data into an appropriate format (e.g., WAV or MP3). After conversion, the device uploads the audio data to the server. The server then applies a noise reduction filter to the uploaded audio data and normalizes the volume. Specifically, open-source tools such as FFmpeg are used. This preprocessing improves the quality of the audio data and optimizes it for subsequent processing.

[0295] The server converts the preprocessed voice data into text data using speech recognition technology. This process uses speech recognition services such as IBM Watson or Google Cloud Speech-to-Text. A machine learning model is then trained using the converted text data and the original voice data. Examples of models used include OpenAI's GPT-3. The model is trained using machine learning frameworks such as TensorFlow and PyTorch. During the training process, the model learns the user's speech characteristics and dialogue style.

[0296] The server then uses an emotion engine, such as IBM Watson's natural language understanding (NLU) service, to extract emotional information from the voice data. The extracted emotional information is used as additional data for machine learning models to help generate more natural and emotionally relevant responses.

[0297] After the death of a user, family and friends can use the device to launch a conversation app and ask questions such as, "How was your day?" The device records the voice and sends it to a server. The server analyzes the voice data and converts it into text. The server then uses an emotion engine to analyze emotions and generate an appropriate response in real time. This response is then sent to the device in text format.

[0298] Finally, the device converts the text data sent from the server into speech using speech synthesis technology. For example, it uses speech synthesis technology such as Google TTS (Text-to-Speech). This allows family and friends to feel as if the user is speaking. For example, if a family member asks, "How was your day?", the device will respond in a voice that sounds like the user, such as, "I went shopping today. The weather was nice, so it felt good."

[0299] Prompt Sentence Examples

[0300] "Imagine a scenario where you're speaking for a user whose family member has passed away. For example, create an example response to the question, 'How was your day?'"

[0301] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0302] Step 1:

[0303] The user starts a dedicated recording application and records a voice message.

[0304] Input: User's spoken utterance

[0305] Output: Recorded audio data (WAV or MP3 format)

[0306] Specific operation: The user launches the app, taps the "Start Recording" button to record audio, and taps the "Stop Recording" button to end the recording. For example, the user can record a voice message such as "Today, I went on a trip with my family. It was a lot of fun."

[0307] Step 2:

[0308] The device converts the recorded audio data into an appropriate format and uploads it to the server.

[0309] Input: Recorded audio data (WAV or MP3 format)

[0310] Output: Audio data uploaded to the server

[0311] Specific operation: The device converts the audio data using a format conversion tool (e.g., FFmpeg) and uploads it to a server via the Internet.

[0312] Step 3:

[0313] The server preprocesses the uploaded audio data and converts it into text data.

[0314] Input: Uploaded audio data

[0315] Output: Denoised and volume-normalised audio data, converted text data

[0316] Specific operation: The server applies a noise reduction filter, normalizes the volume, and converts the speech to text using speech recognition technology (e.g., Google Cloud Speech-to-Text). For example, speech data such as "Today, I went on a trip with my family. It was a lot of fun" is converted into text data such as "Today, I went on a trip with my family. It was a lot of fun."

[0317] Step 4:

[0318] The server uses the audio and text data to train a machine learning model.

[0319] Input: Preprocessed audio data, converted text data

[0320] Output: A trained machine learning model

[0321] How it works: The server uses the voice and text data to train a machine learning model (e.g., GPT-3) using TensorFlow or PyTorch. The model learns the user's speech characteristics and interaction style.

[0322] Step 5:

[0323] The server extracts emotion information from the voice data using an emotion engine.

[0324] Input: Audio data

[0325] Output: Extracted emotion information

[0326] What it does: The server uses sentiment analysis tools (e.g., IBM Watson NLU) to extract emotional information from the voice data and add it to the training data for the machine learning model, enabling natural, emotionally relevant responses.

[0327] Step 6:

[0328] After a user passes away, if family or friends launch a conversation app and speak to them, the device will record the audio and send it to the server.

[0329] Input: Voice utterances from family and friends

[0330] Output: Sending audio data to the server

[0331] Specific operation: When a family member or friend asks, "How was your day?", the device records the audio, preprocesses it, and sends it to the server.

[0332] Step 7:

[0333] The server analyzes the voice data, converts it into text, and then uses an emotion engine to analyze emotions and generate an appropriate response.

[0334] Input: Transmitted audio data

[0335] Output: The generated response text

[0336] How it works: The server uses speech recognition technology to convert the voice data into text, performs sentiment analysis, and uses machine learning models to generate an appropriate response, such as "I went for a walk today. I was happy because the weather was nice."

[0337] Step 8:

[0338] The device converts the generated response text into speech and plays it back to family and friends.

[0339] Input: Generated response text

[0340] Output: A spoken response

[0341] What it does: Your device uses text-to-speech technology (e.g., Google TTS) to convert the text into speech and play it back to your family or friends, providing a voice response that sounds like you.

[0342] (Application example 2)

[0343] 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."

[0344] In conventional virtual store systems, customer service staff respond mechanically and without emotion, lacking friendliness or individualized consideration for visitors. Furthermore, there was a need for a system that could retain a user's speech characteristics and enable natural conversations even after their death. Therefore, a new technology was needed that could utilize the user's voice data while they were alive to understand their emotions and provide friendly customer service in real time.

[0345] 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 recording user voice data, means for uploading the recorded voice data to the server, means for preprocessing the uploaded voice data and converting it into text data, means for performing machine learning on the voice data and text data to learn the user's speech characteristics, means for the server to generate a dialogue in real time after the user's death and transmit it to the terminal, means for the terminal to play back the generated dialogue as voice, and means for providing emotionally considerate customer service to visitors in the virtual store. This makes it possible to provide emotionally considerate and friendly customer service by utilizing the user's speech characteristics.

[0346] "User's voice data" refers to voice information recorded by a user before death, and includes individual speech characteristics and emotional expressions.

[0347] A "server" is a computer system that processes voice data, converts it, performs machine learning, generates dialogue, and stores data.

[0348] A "recording means" is a device or software that has the function of collecting and storing audio data.

[0349] "Uploading means" refers to a device or software that has the function of transmitting recorded audio data to a server via a network.

[0350] "Preprocessing" refers to the process of performing processes such as noise removal and volume normalization on audio data.

[0351] "Means for converting into text data" refers to technology for analyzing voice data and converting its contents into text format.

[0352] "Means for performing machine learning" refers to a technology that uses voice data and text data to learn a user's speech characteristics and dialogue style using a specific algorithm.

[0353] The "means for generating dialogue in real time" refers to a technology that has the function of enabling a server to generate dialogue with a visitor in real time after the death of a user and transmit the content of that dialogue to a terminal.

[0354] The "means for the terminal to play back the dialogue generated as audio" refers to a device or software that converts text data sent from the server into audio and plays it back to the visitor.

[0355] "Means for providing emotionally sensitive customer service" are technologies and systems that provide more natural and friendly interactions in response to visitors' emotions and needs.

[0356] The present invention is a system that records voice data from a user's life and realizes natural conversations that take into consideration emotions. This system is mainly composed of a user, a terminal, and a server. Specific embodiments for implementing the present invention are described below.

[0357] Audio data recording

[0358] The user starts a dedicated voice recording application and records a voice message about an episode, emotion, or event in their life. For example, they can leave a voice message such as, "I bought a new jacket today. I'm so happy I made such a good purchase." This recording becomes the basic data for generating natural dialogue based on the user's speech characteristics and emotional expressions.

[0359] Uploading and preprocessing audio data

[0360] The device converts this audio data into WAV or MP3 format and uploads it to the server. On the server side, preprocessing such as noise removal and volume normalization is first performed to generate clear audio data. This audio data is then converted into text data using speech recognition technology. A Wav2Vec2 model or similar is typically used.

[0361] Training a machine learning model

[0362] The server trains a machine learning model using voice data and the corresponding text data. The trained model learns the user's speech characteristics and dialogue style and uses the data to reproduce those characteristics. Generative AI models such as OpenAI's GPT-3 are often used for training. In addition, by combining an emotion engine, the user's emotional expressions are also included in the learning.

[0363] Generating conversations after a user dies

[0364] After a user passes away, family and friends can use the application to interact with the user in a virtual store. For example, if a family member asks, "What products do you have recommended today?", the server generates a response in real time and sends it to the device as text data.

[0365] Dialogue playback

[0366] The device converts the text data received from the server into speech and plays it back in a voice that matches the user's speech characteristics. For this purpose, the TextToSpeech module is generally used. For example, it responds naturally, saying, "We have new jackets in stock today. Please take a look."

[0367] Examples of concrete examples and prompts

[0368] As a specific example, if a family member asks in a virtual store, "What recommended items do you have today?", the server will generate a natural response such as, "Hello! We have new jackets in stock today. Please take a look," and the device will play that voice back.

[0369] Example prompt for a generative AI model:

[0370] plaintext

[0371] User: What products do you have recommended today?

[0372] Bot:

[0373] In this way, it is possible to provide visitors with emotionally sensitive and natural dialogue based on the user's voice data in life.

[0374] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0375] Step 1:

[0376] A user starts a dedicated voice recording application and records a voice message. The input is the user's speech, and the output is a recorded voice file. The recorded voice is a recording of the user's natural speech, including everyday events and emotions.

[0377] Step 2:

[0378] The device converts the recorded audio data into the appropriate format (WAV or MP3). The input is the audio file obtained in step 1, and the output is the converted audio file. This conversion makes the audio file in a format that can be uploaded to the server.

[0379] Step 3:

[0380] The device uploads the audio file to the server. The input is the converted audio file, and the output is the audio file stored on the server. The audio data is then ready for processing on the server.

[0381] Step 4:

[0382] The server preprocesses the uploaded audio data. Preprocessing includes noise reduction and volume normalization. The input is an audio file stored on the server, and the output is a preprocessed audio file. Preprocessing improves the quality of the audio data.

[0383] Step 5:

[0384] The server converts the preprocessed audio data into text data using speech recognition technology. The input is the preprocessed audio file, and the output is text data transcribed from the audio data. Models such as "Wav2Vec2" are used for speech recognition.

[0385] Step 6:

[0386] The server trains a machine learning model using voice and text data. The input is the voice data and corresponding text data, and the output is a machine learning model that has learned the user's speech characteristics. Generative AI models such as "GPT-3" are used for training.

[0387] Step 7:

[0388] After the death of the user, family and friends use the device to launch a conversation app. The input is the speech of the family and friends, and the output is the voice data sent to the server. The device records the visitor's speech and sends it to the server.

[0389] Step 8:

[0390] The server analyzes the visitor's voice data and converts it into text data. The input is the visitor's voice data, and the output is text data. The server generates a response in real time based on the converted text data.

[0391] Step 9:

[0392] The server generates a response in real time using a generative AI model. The input is the visitor's text data, and the output is the generated response text. An example of a prompt for the generative AI model is as follows:

[0393] plaintext

[0394] User: What products do you have recommended today?

[0395] Bot:

[0396] Step 10:

[0397] The terminal converts the response text received from the server into speech and plays it back to the visitor. The input is the generated response text and the output is the audio data. The terminal uses the TextToSpeech module to generate audio and plays it back to the visitor as a natural dialogue.

[0398] 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.

[0399] 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.

[0400] 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.

[0401] [Second embodiment]

[0402] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0403] 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.

[0404] 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).

[0405] 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.

[0406] 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.

[0407] 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).

[0408] 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.

[0409] 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.

[0410] 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.

[0411] 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.

[0412] 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.

[0413] 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."

[0414] This invention is a system that records the conversations and voices of a user while they are alive, and uses AI to learn from them, allowing for conversations with the user even after they have passed away. This system includes elements such as the user, a terminal, and a server, each of which plays a specific role.

[0415] Audio data recording

[0416] The user launches a dedicated recording app and leaves a voice message. For example, they can record a message like, "Today, I went on a trip with my family. It was a lot of fun." This recording reflects the user's personality and speech characteristics.

[0417] Uploading and preprocessing audio data

[0418] The device records this voice data, converts it into an appropriate format (e.g., WAV or MP3), and uploads it to a server. The uploaded voice data undergoes preprocessing, such as noise reduction and volume normalization, on the server. After preprocessing, the voice data is converted into text data using speech recognition technology.

[0419] Training a machine learning model

[0420] The server uses the voice data and corresponding text data to train a machine learning model. The model learns the user's speech characteristics and conversation style and is able to generate natural, authentic responses. The server also periodically incorporates newly uploaded data to improve the accuracy of the model.

[0421] Generating conversations after a user dies

[0422] After a user's death, a family member or friend can use the device to launch a conversation app. The device records the family member or friend's speech and sends the audio data to a server. The server analyzes the audio data and generates an appropriate response in real time. The generated response is sent to the device as text.

[0423] Dialogue playback

[0424] The device converts the text data received from the server into speech and plays back a response on behalf of the user. This allows family and friends to feel as if they are being spoken by the user. For example, if a family member asks, "How was your day?", the device will respond in a voice that sounds like the user, such as, "I went shopping today."

[0425] Specific examples

[0426] 1. The user launches the app and records a voice message saying, "I went to the park with my dog ​​today. It was fun!"

[0427] 2. The device records this audio and uploads it to the server.

[0428] 3. The server removes noise and normalizes the volume of the audio data, generating text data such as "I went to the park with my dog ​​today. It was fun!"

[0429] 4. The server uses a machine learning model to learn the user's speech characteristics.

[0430] 5. One day, a family member uses a conversation app to ask, "How's your day going?"

[0431] 6. The device sends the audio to the server, which generates a response saying, "I went for a walk today. The weather was nice."

[0432] 7. The device will play back the response as audio, providing a natural conversational experience for family members.

[0433] By combining the user's voice data from before their death with machine learning technology, the present invention makes it possible to realize natural conversations that sound like the user, even after the user has passed away.

[0434] The processing flow will be explained below.

[0435] Step 1:

[0436] The user launches a recording app and starts speaking. For example, the user might record a voice message such as, "Today, my family and I went to the park."

[0437] Step 2:

[0438] The device detects that the record button has been pressed and starts recording audio data. The recorded data is temporarily saved in local storage.

[0439] Step 3:

[0440] The device will recognize that the end recording button has been pressed and will stop recording. The recorded data will be converted to a digital format (e.g., WAV or MP3 format).

[0441] Step 4:

[0442] The device uploads the recording data to a server using a secure communication protocol (e.g., HTTPS).

[0443] Step 5:

[0444] The server receives the uploaded audio data, stores it in a database, and starts preprocessing the audio data.

[0445] Step 6:

[0446] The server performs noise reduction and volume normalization on the audio data, which improves the audio quality.

[0447] Step 7:

[0448] The server uses speech recognition technology to convert the preprocessed speech data into text data, for example, "Today, my family and I went to the park."

[0449] Step 8:

[0450] The server inputs the generated text data and corresponding audio data into a machine learning model, which begins the process of learning the user's speech characteristics.

[0451] Step 9:

[0452] The server trains the machine learning model, learning the user's speech patterns and characteristics to generate highly accurate responses.

[0453] Step 10:

[0454] The server evaluates the machine learning model and retrains it as needed, thereby maintaining the model's accuracy.

[0455] Step 11:

[0456] After the death of the user, a family member or friend can use the device to launch the conversation app and press the "Start conversation" button to begin the conversation.

[0457] Step 12:

[0458] The device records voice input from family and friends and sends the voice data to a server, for example, recording a question like "How was your day?"

[0459] Step 13:

[0460] The server analyzes the received voice data and converts it into text. Based on the converted data, it uses a machine learning model to generate an appropriate response.

[0461] Step 14:

[0462] The server sends the generated response text to the terminal, for example, generating a response such as "I went shopping today."

[0463] Step 15:

[0464] The device converts the response text received from the server into voice and plays it back to the family, enabling natural conversation.

[0465] By taking specific actions at each step, the system can provide natural dialogue that reflects the user's personality even after the user has passed away.

[0466] Example 1

[0467] 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."

[0468] There is a lack of technology to respond to the requests of family and friends who want to continue interacting with a deceased person. Furthermore, insufficient preprocessing of recorded audio data and insufficient training of machine learning models can lead to poor dialogue quality. Furthermore, it is difficult to accurately model the user's speech characteristics and generate natural dialogue.

[0469] 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.

[0470] In this invention, the server includes means for preprocessing the voice data, removing noise and normalizing the volume, and converting the voice data into text data, means for performing machine learning based on the voice data and text data to learn the user's speech characteristics, and means for the server to generate a dialogue in real time after the user's death and transmit it to the terminal, thereby enabling family and friends to experience natural dialogue even after the user's death.

[0471] "Voice data" is digital data that records a user's speech or voice message.

[0472] A "terminal" is an electronic device on which a recording application or a dialogue application used by a user is installed.

[0473] "Convert to format" refers to the process of converting audio data into an appropriate audio file format (e.g., WAV, MP3).

[0474] "Upload" refers to sending data recorded on a terminal to a server via a network.

[0475] A "server" is a central management computer that stores received data and performs various data processing and machine learning.

[0476] "Preprocessing" refers to performing data cleansing operations on the audio data, such as noise removal and volume normalization.

[0477] "Noise reduction" is a process for removing unnecessary background sounds and noise from audio data.

[0478] "Volume normalization" is a process for making the volume level of audio data uniform.

[0479] "Text data" is voice data converted into character data using voice recognition technology.

[0480] A "machine learning model" is an algorithm designed to learn from large amounts of data and perform a specific task.

[0481] "Speech characteristics" refers to the voice characteristics and speaking style of an individual user.

[0482] "Generating dialogue in real time" refers to generating and providing responses instantly as users take turns speaking.

[0483] "Converting to speech" means regenerating text data as speech using speech synthesis technology.

[0484] "Playback" means outputting the generated audio from a playback device such as a speaker.

[0485] "Family and friends" refers to people with whom the user will communicate after their death.

[0486] This invention is a system that records the conversations and voices of a user while they are alive, and uses AI to learn from them, allowing for conversations with the user even after they have passed away. This system includes elements such as the user, a terminal, and a server, each of which plays a specific role.

[0487] First, the user launches a dedicated recording app and leaves a voice message. This voice message reflects the user's personality and speech characteristics and includes specific content such as, "Today, I went on a trip with my family. It was a lot of fun."

[0488] The device then records this audio data, converts it into an appropriate format (e.g., WAV or MP3), and uploads it to a server via the Internet. The audio data format conversion is performed using software installed on the device (e.g., FFMPEG).

[0489] The server performs preprocessing on the uploaded audio data, such as noise removal and volume normalization. This preprocessing uses a noise removal algorithm (e.g., Audacity's noise removal function) and a volume normalization algorithm. After preprocessing is complete, the server converts the audio data into text data using speech recognition technology (e.g., Google Speech-to-Text API). For example, the generated text data is, "Today, I went on a trip with my family. It was a lot of fun."

[0490] The server then uses the voice data and corresponding text data to train a machine learning model (e.g., OpenAI GPT-4). The model learns the user's speech characteristics and conversational style and is able to generate natural, authentic responses. The server also periodically incorporates newly uploaded data to improve the model's accuracy.

[0491] After a user passes away, family and friends can use the device to launch a conversation app. When they speak into the microphone, the device records their voice and sends it to the server in real time. For example, if a family member asks, "How was your day?", the server receives the voice data and converts it into text using speech recognition technology. Then, a machine learning model generates a response such as, "I went for a walk today. The weather was nice."

[0492] The generated response is sent as text to the device, which then converts this text data into speech (e.g., using Amazon Polly to generate synthetic speech), and finally plays the generated speech through the device's speaker, giving family and friends the feeling that it is the user speaking.

[0493] Examples of prompt statements

[0494] Prompt: Convert the following audio data into text and use that text to generate a natural-sounding response based on the user's speaking characteristics.

[0495] Audio data: I went to the park with my dog ​​today. It was fun!

[0496] By combining a user's voice data with machine learning technology, this invention enables a user to have natural, lifelike conversations even after they have passed away. This system provides a way for family and friends to continue to relive precious memories.

[0497] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0498] Step 1:

[0499] The user launches a recording app and records a voice message. The user speaks into the microphone, saying, "Today, I went on a trip with my family. It was a lot of fun." The recording app captures this voice data in WAV format and saves it on the device.

[0500] Input: User's voice

[0501] Output: WAV format audio file

[0502] Step 2:

[0503] The device converts the stored audio data into MP3 format using software installed on the device (e.g., FFMPEG), and then uploads the converted MP3 file to a server via the Internet.

[0504] Input: WAV format audio file

[0505] Output: MP3 audio file

[0506] Step 3:

[0507] The server receives the uploaded audio data and stores it in storage. Then, preprocessing begins. Specifically, the server performs noise reduction and volume normalization on the audio data. This preprocessing uses a noise reduction algorithm (e.g., Audacity's noise reduction function) and a volume normalization algorithm.

[0508] Input: MP3 audio file

[0509] Output: Preprocessed audio file

[0510] Step 4:

[0511] The server uses the preprocessed voice data to convert it into text data using speech recognition technology (e.g., Google Speech-to-Text API). For example, the generated text data is, "Today, I went on a trip with my family. It was a lot of fun."

[0512] Input: Preprocessed audio file

[0513] Output: Text data

[0514] Step 5:

[0515] The server trains a machine learning model (e.g., OpenAI GPT-4) using the voice data and corresponding text data. Through this, the server learns the user's speech characteristics and interaction style. The server also periodically incorporates newly uploaded data and continuously retrains the model to improve its accuracy.

[0516] Input: Text and audio data

[0517] Output: A trained machine learning model

[0518] Step 6:

[0519] After the death of the user, family and friends can launch a conversation app and speak into the microphone, for example, saying, "How was your day?" The device will record the voice and send it to the server in real time.

[0520] Input: Voice of family or friends

[0521] Output: Recorded audio data

[0522] Step 7:

[0523] The server converts the received voice data into text data using speech recognition technology. For example, the text "How was your day?" is generated. Then, using a machine learning model trained on the server, it generates a response such as "I went for a walk today. The weather was nice."

[0524] Input: Recorded audio data

[0525] Output: Response text data

[0526] Step 8:

[0527] The device converts the response text data received from the server into speech, using speech synthesis technology (e.g., Amazon Polly) to generate a speech that says, "I went for a walk today. The weather was nice."

[0528] Input: Response text data

[0529] Output: Audio data

[0530] Step 9:

[0531] The device then plays the generated audio through the speaker, allowing family and friends to hear it as if you were actually speaking.

[0532] Input: Audio data

[0533] Output: Played audio

[0534] (Application example 1)

[0535] 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."

[0536] As the aging society progresses, many people want to cherish memories and conversations with their deceased loved ones. However, existing technology makes it difficult to realize natural conversations with the deceased. It is necessary to solve this problem and provide a new customer experience that enriches memories with the deceased.

[0537] 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.

[0538] In this invention, the server includes means for recording user voice data, means for uploading the recorded voice data to the server, means for preprocessing the uploaded voice data and converting it into text data, means for performing machine learning based on the voice data and text data to learn the user's speech characteristics, means for the server to generate a dialogue in real time after the user's death and transmit it to the terminal, means for the terminal to play back the generated dialogue as audio and provide a customer dialogue experience, and means for providing the customer with a dialogue experience with the deceased via a head-mounted display or smartphone, thereby allowing customers to experience a natural dialogue with the deceased and enrich their memories.

[0539] "User voice data" refers to voice information uttered by a user, and is digital data that can be recorded and analyzed.

[0540] A "server" is a computer system that stores, processes, and manages data on a network, and is responsible for managing user voice data and training machine learning models.

[0541] "Recording means" refers to the functionality of the device or software for recording and storing a user's voice data.

[0542] "Means for uploading to a server" refers to the functionality of a device or software for transmitting recorded audio data to a server via a network.

[0543] "Preprocessing" refers to the process of removing noise and normalizing the volume of uploaded audio data to make it easier to analyze.

[0544] "Means for converting into text data" refers to the technology and functions for analyzing voice data and converting it into text data.

[0545] "Means for conducting machine learning to learn a user's speech characteristics" refers to technologies and functions that use voice data and text data to train a model on a user's speech patterns and characteristics.

[0546] "Means for generating dialogue in real time" refers to technologies and functions that analyze dialogue with family and friends on the spot and instantly generate appropriate responses.

[0547] "Terminal" refers to devices used by users, such as computers, smartphones, and head-mounted displays.

[0548] "Means for playing as audio" refers to the technology or functionality for converting text data into audio and playing it back on a user device.

[0549] "Means for providing a customer interaction experience" refers to technologies and functions that provide natural and individual interactions with users via terminals and realize interactions with users.

[0550] A "head-mounted display" is a display device worn by a user that provides information visually and audibly.

[0551] A "smartphone" is a portable information terminal that combines the functions of a mobile phone with those of a personal computer, and is a device that can record, upload, and play back audio data.

[0552] The present invention provides a system for providing a conversational experience even after a user has passed away by recording the user's voice data and training a machine learning model based on the recorded voice data. This system is particularly applicable to customer conversational experiences using head-mounted displays and smartphones. The components and processing steps of this system are described in detail below.

[0553] Recording and uploading audio data

[0554] Users use a dedicated recording app to record everyday conversations and messages. This recording is done via a device such as a smartphone. The recorded voice data is saved on the device and then uploaded to a server.

[0555] Audio data preprocessing and text conversion

[0556] The server performs preprocessing on the uploaded audio data, removing noise and normalizing the volume, making the data easier to analyze. After preprocessing, the audio data is converted into text using speech recognition technology.

[0557] Training a machine learning model

[0558] The server uses the preprocessed audio data and corresponding text data to train a machine learning model. The model learns the user's speech characteristics and conversational style and generates natural-sounding responses. This allows the model to reproduce a user's conversations in the future, even after the user has passed away.

[0559] Dialogue generation and playback

[0560] After the death of a user, family and friends launch a dedicated conversation app on a head-mounted display or smartphone. The device records the family and friends' speech and sends the audio data to a server. The server analyzes the audio data in real time and generates an appropriate response. The generated response is sent as text data to the device, which converts it into audio and plays it back.

[0561] Specific examples

[0562] 1. A user uses a dedicated app to record a voice message saying, "I went to the park with my dog ​​today. It was fun!"

[0563] 2. The device uploads this audio to the server.

[0564] 3. The server performs preprocessing and generates text data such as "I went to the park with my dog ​​today. It was fun!"

[0565] 4. The server uses a machine learning model to learn the user's speech characteristics.

[0566] 5. One day, a family member uses a conversation app to ask, "How's your day going?"

[0567] 6. The device sends the audio to the server, which generates a response saying, "I went for a walk today. The weather was nice."

[0568] 7. The device will play back the response as audio, providing a natural conversational experience for family members.

[0569] Prompt Sentence Examples

[0570] "User-recorded question: 'Mom, what would you recommend today?'"

[0571] In this way, this invention combines the user's voice data from before death with machine learning technology to enable natural conversations that sound like the user even after the user has passed away. Furthermore, this system is expected to have a wide range of applications when used with head-mounted displays and smartphones.

[0572] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0573] Step 1:

[0574] The user launches a dedicated recording app and records a voice message. The input is the user's speech, and the output is the recorded voice data. Specifically, the user speaks into the smartphone's microphone and records a message such as, "Today, I went on a trip with my family. It was a lot of fun."

[0575] Step 2:

[0576] The device converts the recorded audio data into an appropriate format (e.g., WAV or MP3) and uploads it to the server. The input is the recorded audio data, and the output is the uploaded audio data. Specifically, the smartphone converts the recorded audio data and sends it to the server via the Internet.

[0577] Step 3:

[0578] The server performs preprocessing on the uploaded audio data, such as noise reduction and volume normalization. The input is the uploaded audio data, and the output is the preprocessed audio data. Specifically, the server runs the audio processing algorithm and processes the data to make it easier to analyze.

[0579] Step 4:

[0580] The server converts the preprocessed speech data into text data using speech recognition technology. The input is the preprocessed speech data, and the output is text data. The specific operation is to extract strings of characters from the speech signal using a speech recognition model (e.g., Wav2Vec2).

[0581] Step 5:

[0582] The server trains a machine learning model using voice data and the corresponding text data. The input is voice data and text data, and the output is a trained machine learning model. The specific operation is to input data into the model, and iteratively learns the user's speech characteristics.

[0583] Step 6:

[0584] After the death of a user, family and friends can launch a dialogue app and ask questions by voice. The input is the speech of the family or friend, and the output is recorded voice data. Specifically, family and friends use a smartphone or head-mounted display to record questions such as "How was your day?"

[0585] Step 7:

[0586] The device sends the audio data to the server. The input is the recorded audio data, and the output is the audio data uploaded to the server. The specific operation is that the smartphone sends the recorded audio data to the server.

[0587] Step 8:

[0588] The server analyzes the voice data in real time and generates an appropriate response. The input is the uploaded voice data, and the output is the generated text response. The specific operation is to use the speech recognition and generation AI model on the server to generate a response corresponding to the input voice data.

[0589] Step 9:

[0590] The device converts the text response received from the server into speech, recreating a natural conversational experience even after the user has passed away. The input is the generated text response, and the output is a speech response. The specific operation is that the smartphone or head-mounted display plays the text as speech using speech synthesis technology (e.g., Text-to-Speech).

[0591] 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.

[0592] This invention is a system that records a user's conversations and voices while they are alive and uses AI to learn from them, enabling conversations with the deceased person even after they have passed away. This system includes the user, terminal, and server elements, each of which plays a specific role. Furthermore, by combining it with an emotion engine, the system aims to provide natural conversations that correspond to the user's emotions.

[0593] Audio data recording

[0594] A user launches a dedicated recording app and leaves a voice message. For example, the user might record a message like, "Today, I went on a trip with my family. It was a lot of fun." This recording reflects the user's personality and speech characteristics.

[0595] Uploading and preprocessing audio data

[0596] The device records this voice data, converts it into an appropriate format (e.g., WAV or MP3), and uploads it to a server. The uploaded voice data undergoes preprocessing, such as noise reduction and volume normalization, on the server. After preprocessing, the voice data is converted into text data using speech recognition technology.

[0597] Training a machine learning model

[0598] The server uses the voice data and corresponding text data to train a machine learning model. The model learns the user's speech characteristics and conversation style and is able to generate natural, authentic responses. The server also periodically incorporates newly uploaded data to improve the accuracy of the model.

[0599] Emotion recognition by emotion engine

[0600] The server uses an emotion engine to extract emotional information from the user's voice data. This emotional information is used as additional data when the machine learning model learns the user's speech characteristics. By taking emotional information into account, more natural and emotionally relevant responses are generated.

[0601] Generating conversations after a user dies

[0602] After a user passes away, a family member or friend launches a conversation app on the device. The device records the family member or friend's speech and sends the audio data to a server. The server analyzes the audio data and converts it into text. After conversion, an emotion engine is used to analyze emotions and generate an appropriate response in real time. The generated response is sent to the device as text.

[0603] Dialogue playback

[0604] The device converts the text data received from the server into voice and plays back the response on behalf of the user. This allows family and friends to feel as if they are being spoken by the user. For example, if a family member asks, "How was your day?", the device will play back a voice response that sounds like the user, such as, "I went shopping today. The weather was nice, so it felt good."

[0605] Specific examples

[0606] 1. The user launches the app and records a voice message saying, "I went to the park with my dog ​​today. It was fun!"

[0607] 2. The device records this audio and uploads it to the server.

[0608] 3. The server removes noise and normalizes the volume of the audio data, generating text data such as "I went to the park with my dog ​​today. It was fun!"

[0609] 4. The server uses an emotion engine to extract emotion information from the voice data.

[0610] 5. The server uses a machine learning model to learn the user's emotional information along with their speech characteristics.

[0611] 6. One day, a family member uses a conversation app to ask, "How's your day going?"

[0612] 7. The device sends the audio to the server, which generates a response saying, "I went for a walk today. I was happy because the weather was nice."

[0613] 8. The device will play back the response as audio, providing a natural conversational experience for family members.

[0614] By combining the user's voice data from their lifetime with machine learning technology and an emotion engine, this invention makes it possible to realize natural conversations that reflect the user's personality and are sensitive to their emotions even after they have passed away.

[0615] The processing flow will be explained below.

[0616] Step 1:

[0617] The user starts a dedicated recording application and records a voice message. For example, the user might record a message such as, "Today, I went on a trip with my family. It was a lot of fun."

[0618] Step 2:

[0619] The device detects that the record button has been pressed and starts recording audio data. The recorded data is temporarily saved in local storage.

[0620] Step 3:

[0621] The device will recognize that the end recording button has been pressed and will stop recording. The recorded data will be converted to a digital format (e.g., WAV or MP3 format).

[0622] Step 4:

[0623] The device uploads the recording data to a server using a secure communication protocol (e.g., HTTPS).

[0624] Step 5:

[0625] The server receives the uploaded voice data, stores it in a database, confirms receipt of the data, and moves on to the next processing step.

[0626] Step 6:

[0627] The server starts pre-processing the audio data, performing noise reduction and volume normalization to improve the audio quality.

[0628] Step 7:

[0629] The server uses speech recognition technology to convert the preprocessed speech data into text data, for example, "Today, I went on a trip with my family."

[0630] Step 8:

[0631] The server uses an emotion engine to extract the user's emotional information from the voice data. For example, it recognizes positive emotions from voice characteristics such as "fun."

[0632] Step 9:

[0633] The server inputs voice data, text data, and emotional information into the machine learning model, which then learns responses based on the user's speech characteristics and emotions.

[0634] Step 10:

[0635] The server trains the machine learning model to generate natural-looking dialogue that reflects the user's emotions, and periodically ingests new data to update the model.

[0636] Step 11:

[0637] After the death of the user, a family member or friend can use the device to launch the conversation app and press the "Start conversation" button to begin the conversation.

[0638] Step 12:

[0639] The device records voice input from family and friends and sends the voice data to a server, for example, recording questions like "How was your day?"

[0640] Step 13:

[0641] The server analyzes the received voice data and converts it into text. After conversion, an emotion engine is used to extract emotional information and generate an appropriate response in real time.

[0642] Step 14:

[0643] The server sends the generated response text to the terminal, for example, "I went shopping today. The weather was nice, so it felt good."

[0644] Step 15:

[0645] The device converts the response text received from the server into voice and plays it back to the family, enabling natural conversation.

[0646] By taking specific actions at each step, the system is able to provide natural, emotionally sensitive conversations that reflect the user's personality, even after the user has passed away.

[0647] Example 2

[0648] 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."

[0649] Currently, there is no system that records a user's conversations and voices while they are alive and allows them to have a conversation with the user even after they have passed away. In particular, it is difficult to provide natural conversations that reflect the user's speech characteristics and emotions. The purpose of this invention is to provide natural conversations that reflect the user's personality and are in tune with their emotions even after the user has passed away.

[0650] 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.

[0651] In this invention, the server includes means for recording user voice data, means for converting the format of the recorded voice data by the terminal and uploading it to the server, means for preprocessing the uploaded voice data and converting it into text data, means for performing machine learning based on the voice data and text data to learn the user's speech characteristics and emotional information, means for extracting emotional information from the voice data using an emotion engine and using it as training data for a machine learning model, means for the server to generate a dialogue in real time after the user's death and send it to the terminal, and means for the terminal to play back the generated dialogue as audio. This makes it possible to provide natural dialogue that reflects the user's individuality and is in tune with their emotions.

[0652] A "user" is an individual who provides voice data to the system.

[0653] "Audio data" refers to data of a voice message generated by a user using a recording application.

[0654] A "terminal" is an electronic device that a user uses to record audio data and upload it to a server.

[0655] The "server" is a central processing unit responsible for preprocessing voice data, training machine learning models, generating dialogue, and more.

[0656] "Format conversion" is the process of changing audio data into an appropriate format, such as WAV or MP3 format.

[0657] "Noise reduction" is a process for removing unnecessary background noise contained in audio data.

[0658] "Volume normalization" is a process for making the volume level of audio data uniform.

[0659] "Text data" is voice data that has been analyzed and converted into text information.

[0660] "Machine learning" is the process of using data to train a model and learn patterns and rules from the data.

[0661] "Speech characteristics" refer to the characteristics of a user's voice and speaking style.

[0662] "Emotion information" is data that expresses the user's emotions and feelings extracted from the voice data.

[0663] An "emotion engine" is a technology or system for extracting emotional information from voice data.

[0664] "Dialogue generation" is the process of creating natural responses and dialogue based on data learned by a machine learning model.

[0665] "Real-time" refers to time characteristics that result in near-instant processing and response.

[0666] "Audio playback" means converting text data into audio and outputting it as if the user were speaking.

[0667] This invention is a system that records a user's voice data while they are alive and uses a machine learning model to learn from it, enabling conversations that sound like the user even after they have passed away. This system mainly uses the user, a terminal, and a server. The specific operation of each element is described below.

[0668] First, the user launches a dedicated recording app and records a voice message. For example, a user might record a message like, "Today, I went on a family trip. It was a lot of fun." This voice data reflects the user's personality and speech characteristics in detail. The recording app runs on a device such as a smartphone or tablet, and the user taps the "Start Recording" button to start recording and the "Stop Recording" button to end recording.

[0669] The device then converts the recorded audio data into an appropriate format (e.g., WAV or MP3). After conversion, the device uploads the audio data to the server. The server then applies a noise reduction filter to the uploaded audio data and normalizes the volume. Specifically, open-source tools such as FFmpeg are used. This preprocessing improves the quality of the audio data and optimizes it for subsequent processing.

[0670] The server converts the preprocessed voice data into text data using speech recognition technology. This process uses speech recognition services such as IBM Watson or Google Cloud Speech-to-Text. A machine learning model is then trained using the converted text data and the original voice data. Examples of models used include OpenAI's GPT-3. The model is trained using machine learning frameworks such as TensorFlow and PyTorch. During the training process, the model learns the user's speech characteristics and dialogue style.

[0671] The server then uses an emotion engine, such as IBM Watson's natural language understanding (NLU) service, to extract emotional information from the voice data. The extracted emotional information is used as additional data for machine learning models to help generate more natural and emotionally relevant responses.

[0672] After the death of a user, family and friends can use the device to launch a conversation app and ask questions such as, "How was your day?" The device records the voice and sends it to a server. The server analyzes the voice data and converts it into text. The server then uses an emotion engine to analyze emotions and generate an appropriate response in real time. This response is then sent to the device in text format.

[0673] Finally, the device converts the text data sent from the server into speech using speech synthesis technology. For example, it uses speech synthesis technology such as Google TTS (Text-to-Speech). This allows family and friends to feel as if the user is speaking. For example, if a family member asks, "How was your day?", the device will respond in a voice that sounds like the user, such as, "I went shopping today. The weather was nice, so it felt good."

[0674] Prompt Sentence Examples

[0675] "Imagine a scenario where you're speaking for a user whose family member has passed away. For example, create an example response to the question, 'How was your day?'"

[0676] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0677] Step 1:

[0678] The user starts a dedicated recording application and records a voice message.

[0679] Input: User's spoken utterance

[0680] Output: Recorded audio data (WAV or MP3 format)

[0681] Specific operation: The user launches the app, taps the "Start Recording" button to record audio, and taps the "Stop Recording" button to end the recording. For example, the user can record a voice message such as "Today, I went on a trip with my family. It was a lot of fun."

[0682] Step 2:

[0683] The device converts the recorded audio data into an appropriate format and uploads it to the server.

[0684] Input: Recorded audio data (WAV or MP3 format)

[0685] Output: Audio data uploaded to the server

[0686] Specific operation: The device converts the audio data using a format conversion tool (e.g., FFmpeg) and uploads it to a server via the Internet.

[0687] Step 3:

[0688] The server preprocesses the uploaded audio data and converts it into text data.

[0689] Input: Uploaded audio data

[0690] Output: Denoised and volume-normalised audio data, converted text data

[0691] Specific operation: The server applies a noise reduction filter, normalizes the volume, and converts the speech to text using speech recognition technology (e.g., Google Cloud Speech-to-Text). For example, speech data such as "Today, I went on a trip with my family. It was a lot of fun" is converted into text data such as "Today, I went on a trip with my family. It was a lot of fun."

[0692] Step 4:

[0693] The server uses the audio and text data to train a machine learning model.

[0694] Input: Preprocessed audio data, converted text data

[0695] Output: A trained machine learning model

[0696] How it works: The server uses the voice and text data to train a machine learning model (e.g., GPT-3) using TensorFlow or PyTorch. The model learns the user's speech characteristics and interaction style.

[0697] Step 5:

[0698] The server extracts emotion information from the voice data using an emotion engine.

[0699] Input: Audio data

[0700] Output: Extracted emotion information

[0701] What it does: The server uses sentiment analysis tools (e.g., IBM Watson NLU) to extract emotional information from the voice data and add it to the training data for the machine learning model, enabling natural, emotionally relevant responses.

[0702] Step 6:

[0703] After a user passes away, if family or friends launch a conversation app and speak to them, the device will record the audio and send it to the server.

[0704] Input: Voice utterances from family and friends

[0705] Output: Sending audio data to the server

[0706] Specific operation: When a family member or friend asks, "How was your day?", the device records the audio, preprocesses it, and sends it to the server.

[0707] Step 7:

[0708] The server analyzes the voice data, converts it into text, and then uses an emotion engine to analyze emotions and generate an appropriate response.

[0709] Input: Transmitted audio data

[0710] Output: The generated response text

[0711] How it works: The server uses speech recognition technology to convert the voice data into text, performs sentiment analysis, and uses machine learning models to generate an appropriate response, such as "I went for a walk today. I was happy because the weather was nice."

[0712] Step 8:

[0713] The device converts the generated response text into speech and plays it back to family and friends.

[0714] Input: Generated response text

[0715] Output: A spoken response

[0716] What it does: Your device uses text-to-speech technology (e.g., Google TTS) to convert the text into speech and play it back to your family or friends, providing a voice response that sounds like you.

[0717] (Application example 2)

[0718] 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."

[0719] In conventional virtual store systems, customer service staff respond mechanically and without emotion, lacking friendliness or individualized consideration for visitors. Furthermore, there was a need for a system that could retain a user's speech characteristics and enable natural conversations even after their death. Therefore, a new technology was needed that could utilize the user's voice data while they were alive to understand their emotions and provide friendly customer service in real time.

[0720] 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 recording user voice data, means for uploading the recorded voice data to the server, means for preprocessing the uploaded voice data and converting it into text data, means for performing machine learning on the voice data and text data to learn the user's speech characteristics, means for the server to generate a dialogue in real time after the user's death and transmit it to the terminal, means for the terminal to play back the generated dialogue as voice, and means for providing emotionally considerate customer service to visitors in the virtual store. This makes it possible to provide emotionally considerate and friendly customer service by utilizing the user's speech characteristics.

[0721] "User's voice data" refers to voice information recorded by a user before death, and includes individual speech characteristics and emotional expressions.

[0722] A "server" is a computer system that processes voice data, converts it, performs machine learning, generates dialogue, and stores data.

[0723] A "recording means" is a device or software that has the function of collecting and storing audio data.

[0724] "Uploading means" refers to a device or software that has the function of transmitting recorded audio data to a server via a network.

[0725] "Preprocessing" refers to the process of performing processes such as noise removal and volume normalization on audio data.

[0726] "Means for converting into text data" refers to technology for analyzing voice data and converting its contents into text format.

[0727] "Means for performing machine learning" refers to a technology that uses voice data and text data to learn a user's speech characteristics and dialogue style using a specific algorithm.

[0728] The "means for generating dialogue in real time" refers to a technology that has the function of enabling a server to generate dialogue with a visitor in real time after the death of a user and transmit the content of that dialogue to a terminal.

[0729] The "means for the terminal to play back the dialogue generated as audio" refers to a device or software that converts text data sent from the server into audio and plays it back to the visitor.

[0730] "Means for providing emotionally sensitive customer service" are technologies and systems that provide more natural and friendly interactions in response to visitors' emotions and needs.

[0731] The present invention is a system that records voice data from a user's life and realizes natural conversations that take into consideration emotions. This system is mainly composed of a user, a terminal, and a server. Specific embodiments for implementing the present invention are described below.

[0732] Audio data recording

[0733] The user starts a dedicated voice recording application and records a voice message about an episode, emotion, or event in their life. For example, they can leave a voice message such as, "I bought a new jacket today. I'm so happy I made such a good purchase." This recording becomes the basic data for generating natural dialogue based on the user's speech characteristics and emotional expressions.

[0734] Uploading and preprocessing audio data

[0735] The device converts this audio data into WAV or MP3 format and uploads it to the server. On the server side, preprocessing such as noise removal and volume normalization is first performed to generate clear audio data. This audio data is then converted into text data using speech recognition technology. A Wav2Vec2 model or similar is typically used.

[0736] Training a machine learning model

[0737] The server trains a machine learning model using voice data and the corresponding text data. The trained model learns the user's speech characteristics and dialogue style and uses the data to reproduce those characteristics. Generative AI models such as OpenAI's GPT-3 are often used for training. In addition, by combining an emotion engine, the user's emotional expressions are also included in the learning.

[0738] Generating conversations after a user dies

[0739] After a user passes away, family and friends can use the application to interact with the user in a virtual store. For example, if a family member asks, "What products do you have recommended today?", the server generates a response in real time and sends it to the device as text data.

[0740] Dialogue playback

[0741] The device converts the text data received from the server into speech and plays it back in a voice that matches the user's speech characteristics. For this purpose, the TextToSpeech module is generally used. For example, it responds naturally, saying, "We have new jackets in stock today. Please take a look."

[0742] Examples of concrete examples and prompts

[0743] As a specific example, if a family member asks in a virtual store, "What recommended items do you have today?", the server will generate a natural response such as, "Hello! We have new jackets in stock today. Please take a look," and the device will play that voice back.

[0744] Example prompt for a generative AI model:

[0745] plaintext

[0746] User: What products do you have recommended today?

[0747] Bot:

[0748] In this way, it is possible to provide visitors with emotionally sensitive and natural dialogue based on the user's voice data in life.

[0749] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0750] Step 1:

[0751] A user starts a dedicated voice recording application and records a voice message. The input is the user's speech, and the output is a recorded voice file. The recorded voice is a recording of the user's natural speech, including everyday events and emotions.

[0752] Step 2:

[0753] The device converts the recorded audio data into the appropriate format (WAV or MP3). The input is the audio file obtained in step 1, and the output is the converted audio file. This conversion makes the audio file in a format that can be uploaded to the server.

[0754] Step 3:

[0755] The device uploads the audio file to the server. The input is the converted audio file, and the output is the audio file stored on the server. The audio data is then ready for processing on the server.

[0756] Step 4:

[0757] The server preprocesses the uploaded audio data. Preprocessing includes noise reduction and volume normalization. The input is an audio file stored on the server, and the output is a preprocessed audio file. Preprocessing improves the quality of the audio data.

[0758] Step 5:

[0759] The server converts the preprocessed audio data into text data using speech recognition technology. The input is the preprocessed audio file, and the output is text data transcribed from the audio data. Models such as "Wav2Vec2" are used for speech recognition.

[0760] Step 6:

[0761] The server trains a machine learning model using voice and text data. The input is the voice data and corresponding text data, and the output is a machine learning model that has learned the user's speech characteristics. Generative AI models such as "GPT-3" are used for training.

[0762] Step 7:

[0763] After the death of the user, family and friends use the device to launch a conversation app. The input is the speech of the family and friends, and the output is the voice data sent to the server. The device records the visitor's speech and sends it to the server.

[0764] Step 8:

[0765] The server analyzes the visitor's voice data and converts it into text data. The input is the visitor's voice data, and the output is text data. The server generates a response in real time based on the converted text data.

[0766] Step 9:

[0767] The server generates a response in real time using a generative AI model. The input is the visitor's text data, and the output is the generated response text. An example of a prompt for the generative AI model is as follows:

[0768] plaintext

[0769] User: What products do you have recommended today?

[0770] Bot:

[0771] Step 10:

[0772] The terminal converts the response text received from the server into speech and plays it back to the visitor. The input is the generated response text and the output is the audio data. The terminal uses the TextToSpeech module to generate audio and plays it back to the visitor as a natural dialogue.

[0773] 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.

[0774] 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.

[0775] 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.

[0776] [Third embodiment]

[0777] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0778] 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.

[0779] 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).

[0780] 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.

[0781] 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.

[0782] 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).

[0783] 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.

[0784] 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.

[0785] 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.

[0786] 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.

[0787] 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.

[0788] 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."

[0789] This invention is a system that records the conversations and voices of a user while they are alive, and uses AI to learn from them, allowing for conversations with the user even after they have passed away. This system includes elements such as the user, a terminal, and a server, each of which plays a specific role.

[0790] Audio data recording

[0791] The user launches a dedicated recording app and leaves a voice message. For example, they can record a message like, "Today, I went on a trip with my family. It was a lot of fun." This recording reflects the user's personality and speech characteristics.

[0792] Uploading and preprocessing audio data

[0793] The device records this voice data, converts it into an appropriate format (e.g., WAV or MP3), and uploads it to a server. The uploaded voice data undergoes preprocessing, such as noise reduction and volume normalization, on the server. After preprocessing, the voice data is converted into text data using speech recognition technology.

[0794] Training a machine learning model

[0795] The server uses the voice data and corresponding text data to train a machine learning model. The model learns the user's speech characteristics and conversation style and is able to generate natural, authentic responses. The server also periodically incorporates newly uploaded data to improve the accuracy of the model.

[0796] Generating conversations after a user dies

[0797] After a user's death, a family member or friend can use the device to launch a conversation app. The device records the family member or friend's speech and sends the audio data to a server. The server analyzes the audio data and generates an appropriate response in real time. The generated response is sent to the device as text.

[0798] Dialogue playback

[0799] The device converts the text data received from the server into speech and plays back a response on behalf of the user. This allows family and friends to feel as if they are being spoken by the user. For example, if a family member asks, "How was your day?", the device will respond in a voice that sounds like the user, such as, "I went shopping today."

[0800] Specific examples

[0801] 1. The user launches the app and records a voice message saying, "I went to the park with my dog ​​today. It was fun!"

[0802] 2. The device records this audio and uploads it to the server.

[0803] 3. The server removes noise and normalizes the volume of the audio data, generating text data such as "I went to the park with my dog ​​today. It was fun!"

[0804] 4. The server uses a machine learning model to learn the user's speech characteristics.

[0805] 5. One day, a family member uses a conversation app to ask, "How's your day going?"

[0806] 6. The device sends the audio to the server, which generates a response saying, "I went for a walk today. The weather was nice."

[0807] 7. The device will play back the response as audio, providing a natural conversational experience for family members.

[0808] By combining the user's voice data from before their death with machine learning technology, the present invention makes it possible to realize natural conversations that sound like the user, even after the user has passed away.

[0809] The processing flow will be explained below.

[0810] Step 1:

[0811] The user launches a recording app and starts speaking. For example, the user might record a voice message such as, "Today, my family and I went to the park."

[0812] Step 2:

[0813] The device detects that the record button has been pressed and starts recording audio data. The recorded data is temporarily saved in local storage.

[0814] Step 3:

[0815] The device will recognize that the end recording button has been pressed and will stop recording. The recorded data will be converted to a digital format (e.g., WAV or MP3 format).

[0816] Step 4:

[0817] The device uploads the recording data to a server using a secure communication protocol (e.g., HTTPS).

[0818] Step 5:

[0819] The server receives the uploaded audio data, stores it in a database, and starts preprocessing the audio data.

[0820] Step 6:

[0821] The server performs noise reduction and volume normalization on the audio data, which improves the audio quality.

[0822] Step 7:

[0823] The server uses speech recognition technology to convert the preprocessed speech data into text data, for example, "Today, my family and I went to the park."

[0824] Step 8:

[0825] The server inputs the generated text data and corresponding audio data into a machine learning model, which begins the process of learning the user's speech characteristics.

[0826] Step 9:

[0827] The server trains the machine learning model, learning the user's speech patterns and characteristics to generate highly accurate responses.

[0828] Step 10:

[0829] The server evaluates the machine learning model and retrains it as needed, thereby maintaining the model's accuracy.

[0830] Step 11:

[0831] After the death of the user, a family member or friend can use the device to launch the conversation app and press the "Start conversation" button to begin the conversation.

[0832] Step 12:

[0833] The device records voice input from family and friends and sends the voice data to a server, for example, recording a question like "How was your day?"

[0834] Step 13:

[0835] The server analyzes the received voice data and converts it into text. Based on the converted data, it uses a machine learning model to generate an appropriate response.

[0836] Step 14:

[0837] The server sends the generated response text to the terminal, for example, generating a response such as "I went shopping today."

[0838] Step 15:

[0839] The device converts the response text received from the server into voice and plays it back to the family, enabling natural conversation.

[0840] By taking specific actions at each step, the system can provide natural dialogue that reflects the user's personality even after the user has passed away.

[0841] Example 1

[0842] 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."

[0843] There is a lack of technology to respond to the requests of family and friends who want to continue interacting with a deceased person. Furthermore, insufficient preprocessing of recorded audio data and insufficient training of machine learning models can lead to poor dialogue quality. Furthermore, it is difficult to accurately model the user's speech characteristics and generate natural dialogue.

[0844] 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.

[0845] In this invention, the server includes means for preprocessing the voice data, removing noise and normalizing the volume, and converting the voice data into text data, means for performing machine learning based on the voice data and text data to learn the user's speech characteristics, and means for the server to generate a dialogue in real time after the user's death and transmit it to the terminal, thereby enabling family and friends to experience natural dialogue even after the user's death.

[0846] "Voice data" is digital data that records a user's speech or voice message.

[0847] A "terminal" is an electronic device on which a recording application or a dialogue application used by a user is installed.

[0848] "Convert to format" refers to the process of converting audio data into an appropriate audio file format (e.g., WAV, MP3).

[0849] "Upload" refers to sending data recorded on a terminal to a server via a network.

[0850] A "server" is a central management computer that stores received data and performs various data processing and machine learning.

[0851] "Preprocessing" refers to performing data cleansing operations on the audio data, such as noise removal and volume normalization.

[0852] "Noise reduction" is a process for removing unnecessary background sounds and noise from audio data.

[0853] "Volume normalization" is a process for making the volume level of audio data uniform.

[0854] "Text data" is voice data converted into character data using voice recognition technology.

[0855] A "machine learning model" is an algorithm designed to learn from large amounts of data and perform a specific task.

[0856] "Speech characteristics" refers to the voice characteristics and speaking style of an individual user.

[0857] "Generating dialogue in real time" refers to generating and providing responses instantly as users take turns speaking.

[0858] "Converting to speech" means regenerating text data as speech using speech synthesis technology.

[0859] "Playback" means outputting the generated audio from a playback device such as a speaker.

[0860] "Family and friends" refers to people with whom the user will communicate after their death.

[0861] This invention is a system that records the conversations and voices of a user while they are alive, and uses AI to learn from them, allowing for conversations with the user even after they have passed away. This system includes elements such as the user, a terminal, and a server, each of which plays a specific role.

[0862] First, the user launches a dedicated recording app and leaves a voice message. This voice message reflects the user's personality and speech characteristics and includes specific content such as, "Today, I went on a trip with my family. It was a lot of fun."

[0863] The device then records this audio data, converts it into an appropriate format (e.g., WAV or MP3), and uploads it to a server via the Internet. The audio data format conversion is performed using software installed on the device (e.g., FFMPEG).

[0864] The server performs preprocessing on the uploaded audio data, such as noise removal and volume normalization. This preprocessing uses a noise removal algorithm (e.g., Audacity's noise removal function) and a volume normalization algorithm. After preprocessing is complete, the server converts the audio data into text data using speech recognition technology (e.g., Google Speech-to-Text API). For example, the generated text data is, "Today, I went on a trip with my family. It was a lot of fun."

[0865] The server then uses the voice data and corresponding text data to train a machine learning model (e.g., OpenAI GPT-4). The model learns the user's speech characteristics and conversational style and is able to generate natural, authentic responses. The server also periodically incorporates newly uploaded data to improve the model's accuracy.

[0866] After a user passes away, family and friends can use the device to launch a conversation app. When they speak into the microphone, the device records their voice and sends it to the server in real time. For example, if a family member asks, "How was your day?", the server receives the voice data and converts it into text using speech recognition technology. Then, a machine learning model generates a response such as, "I went for a walk today. The weather was nice."

[0867] The generated response is sent as text to the device, which then converts this text data into speech (e.g., using Amazon Polly to generate synthetic speech), and finally plays the generated speech through the device's speaker, giving family and friends the feeling that it is the user speaking.

[0868] Examples of prompt statements

[0869] Prompt: Convert the following audio data into text and use that text to generate a natural-sounding response based on the user's speaking characteristics.

[0870] Audio data: I went to the park with my dog ​​today. It was fun!

[0871] By combining a user's voice data with machine learning technology, this invention enables a user to have natural, lifelike conversations even after they have passed away. This system provides a way for family and friends to continue to relive precious memories.

[0872] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0873] Step 1:

[0874] The user launches a recording app and records a voice message. The user speaks into the microphone, saying, "Today, I went on a trip with my family. It was a lot of fun." The recording app captures this voice data in WAV format and saves it on the device.

[0875] Input: User's voice

[0876] Output: WAV format audio file

[0877] Step 2:

[0878] The device converts the stored audio data into MP3 format using software installed on the device (e.g., FFMPEG), and then uploads the converted MP3 file to a server via the Internet.

[0879] Input: WAV format audio file

[0880] Output: MP3 audio file

[0881] Step 3:

[0882] The server receives the uploaded audio data and stores it in storage. Then, preprocessing begins. Specifically, the server performs noise reduction and volume normalization on the audio data. This preprocessing uses a noise reduction algorithm (e.g., Audacity's noise reduction function) and a volume normalization algorithm.

[0883] Input: MP3 audio file

[0884] Output: Preprocessed audio file

[0885] Step 4:

[0886] The server uses the preprocessed voice data to convert it into text data using speech recognition technology (e.g., Google Speech-to-Text API). For example, the generated text data is, "Today, I went on a trip with my family. It was a lot of fun."

[0887] Input: Preprocessed audio file

[0888] Output: Text data

[0889] Step 5:

[0890] The server trains a machine learning model (e.g., OpenAI GPT-4) using the voice data and corresponding text data. Through this, the server learns the user's speech characteristics and interaction style. The server also periodically incorporates newly uploaded data and continuously retrains the model to improve its accuracy.

[0891] Input: Text and audio data

[0892] Output: A trained machine learning model

[0893] Step 6:

[0894] After the death of the user, family and friends can launch a conversation app and speak into the microphone, for example, saying, "How was your day?" The device will record the voice and send it to the server in real time.

[0895] Input: Voice of family or friends

[0896] Output: Recorded audio data

[0897] Step 7:

[0898] The server converts the received voice data into text data using speech recognition technology. For example, the text "How was your day?" is generated. Then, using a machine learning model trained on the server, it generates a response such as "I went for a walk today. The weather was nice."

[0899] Input: Recorded audio data

[0900] Output: Response text data

[0901] Step 8:

[0902] The device converts the response text data received from the server into speech, using speech synthesis technology (e.g., Amazon Polly) to generate a speech that says, "I went for a walk today. The weather was nice."

[0903] Input: Response text data

[0904] Output: Audio data

[0905] Step 9:

[0906] The device then plays the generated audio through the speaker, allowing family and friends to hear it as if you were actually speaking.

[0907] Input: Audio data

[0908] Output: Played audio

[0909] (Application example 1)

[0910] 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."

[0911] As the aging society progresses, many people want to cherish memories and conversations with their deceased loved ones. However, existing technology makes it difficult to realize natural conversations with the deceased. It is necessary to solve this problem and provide a new customer experience that enriches memories with the deceased.

[0912] 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.

[0913] In this invention, the server includes means for recording user voice data, means for uploading the recorded voice data to the server, means for preprocessing the uploaded voice data and converting it into text data, means for performing machine learning based on the voice data and text data to learn the user's speech characteristics, means for the server to generate a dialogue in real time after the user's death and transmit it to the terminal, means for the terminal to play back the generated dialogue as audio and provide a customer dialogue experience, and means for providing the customer with a dialogue experience with the deceased via a head-mounted display or smartphone, thereby allowing customers to experience a natural dialogue with the deceased and enrich their memories.

[0914] "User voice data" refers to voice information uttered by a user, and is digital data that can be recorded and analyzed.

[0915] A "server" is a computer system that stores, processes, and manages data on a network, and is responsible for managing user voice data and training machine learning models.

[0916] "Recording means" refers to the functionality of the device or software for recording and storing a user's voice data.

[0917] "Means for uploading to a server" refers to the functionality of a device or software for transmitting recorded audio data to a server via a network.

[0918] "Preprocessing" refers to the process of removing noise and normalizing the volume of uploaded audio data to make it easier to analyze.

[0919] "Means for converting into text data" refers to the technology and functions for analyzing voice data and converting it into text data.

[0920] "Means for conducting machine learning to learn a user's speech characteristics" refers to technologies and functions that use voice data and text data to train a model on a user's speech patterns and characteristics.

[0921] "Means for generating dialogue in real time" refers to technologies and functions that analyze dialogue with family and friends on the spot and instantly generate appropriate responses.

[0922] "Terminal" refers to devices used by users, such as computers, smartphones, and head-mounted displays.

[0923] "Means for playing as audio" refers to the technology or functionality for converting text data into audio and playing it back on a user device.

[0924] "Means for providing a customer interaction experience" refers to technologies and functions that provide natural and individual interactions with users via terminals and realize interactions with users.

[0925] A "head-mounted display" is a display device worn by a user that provides information visually and audibly.

[0926] A "smartphone" is a portable information terminal that combines the functions of a mobile phone with those of a personal computer, and is a device that can record, upload, and play back audio data.

[0927] The present invention provides a system for providing a conversational experience even after a user has passed away by recording the user's voice data and training a machine learning model based on the recorded voice data. This system is particularly applicable to customer conversational experiences using head-mounted displays and smartphones. The components and processing steps of this system are described in detail below.

[0928] Recording and uploading audio data

[0929] Users use a dedicated recording app to record everyday conversations and messages. This recording is done via a device such as a smartphone. The recorded voice data is saved on the device and then uploaded to a server.

[0930] Audio data preprocessing and text conversion

[0931] The server performs preprocessing on the uploaded audio data, removing noise and normalizing the volume, making the data easier to analyze. After preprocessing, the audio data is converted into text using speech recognition technology.

[0932] Training a machine learning model

[0933] The server uses the preprocessed audio data and corresponding text data to train a machine learning model. The model learns the user's speech characteristics and conversational style and generates natural-sounding responses. This allows the model to reproduce a user's conversations in the future, even after the user has passed away.

[0934] Dialogue generation and playback

[0935] After the death of a user, family and friends launch a dedicated conversation app on a head-mounted display or smartphone. The device records the family and friends' speech and sends the audio data to a server. The server analyzes the audio data in real time and generates an appropriate response. The generated response is sent as text data to the device, which converts it into audio and plays it back.

[0936] Specific examples

[0937] 1. A user uses a dedicated app to record a voice message saying, "I went to the park with my dog ​​today. It was fun!"

[0938] 2. The device uploads this audio to the server.

[0939] 3. The server performs preprocessing and generates text data such as "I went to the park with my dog ​​today. It was fun!"

[0940] 4. The server uses a machine learning model to learn the user's speech characteristics.

[0941] 5. One day, a family member uses a conversation app to ask, "How's your day going?"

[0942] 6. The device sends the audio to the server, which generates a response saying, "I went for a walk today. The weather was nice."

[0943] 7. The device will play back the response as audio, providing a natural conversational experience for family members.

[0944] Prompt Sentence Examples

[0945] "User-recorded question: 'Mom, what would you recommend today?'"

[0946] In this way, this invention combines the user's voice data from before death with machine learning technology to enable natural conversations that sound like the user even after the user has passed away. Furthermore, this system is expected to have a wide range of applications when used with head-mounted displays and smartphones.

[0947] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0948] Step 1:

[0949] The user launches a dedicated recording app and records a voice message. The input is the user's speech, and the output is the recorded voice data. Specifically, the user speaks into the smartphone's microphone and records a message such as, "Today, I went on a trip with my family. It was a lot of fun."

[0950] Step 2:

[0951] The device converts the recorded audio data into an appropriate format (e.g., WAV or MP3) and uploads it to the server. The input is the recorded audio data, and the output is the uploaded audio data. Specifically, the smartphone converts the recorded audio data and sends it to the server via the Internet.

[0952] Step 3:

[0953] The server performs preprocessing on the uploaded audio data, such as noise reduction and volume normalization. The input is the uploaded audio data, and the output is the preprocessed audio data. Specifically, the server runs the audio processing algorithm and processes the data to make it easier to analyze.

[0954] Step 4:

[0955] The server converts the preprocessed speech data into text data using speech recognition technology. The input is the preprocessed speech data, and the output is text data. The specific operation is to extract strings of characters from the speech signal using a speech recognition model (e.g., Wav2Vec2).

[0956] Step 5:

[0957] The server trains a machine learning model using voice data and the corresponding text data. The input is voice data and text data, and the output is a trained machine learning model. The specific operation is to input data into the model, and iteratively learns the user's speech characteristics.

[0958] Step 6:

[0959] After the death of a user, family and friends can launch a dialogue app and ask questions by voice. The input is the speech of the family or friend, and the output is recorded voice data. Specifically, family and friends use a smartphone or head-mounted display to record questions such as "How was your day?"

[0960] Step 7:

[0961] The device sends the audio data to the server. The input is the recorded audio data, and the output is the audio data uploaded to the server. The specific operation is that the smartphone sends the recorded audio data to the server.

[0962] Step 8:

[0963] The server analyzes the voice data in real time and generates an appropriate response. The input is the uploaded voice data, and the output is the generated text response. The specific operation is to use the speech recognition and generation AI model on the server to generate a response corresponding to the input voice data.

[0964] Step 9:

[0965] The device converts the text response received from the server into speech, recreating a natural conversational experience even after the user has passed away. The input is the generated text response, and the output is a speech response. The specific operation is that the smartphone or head-mounted display plays the text as speech using speech synthesis technology (e.g., Text-to-Speech).

[0966] 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.

[0967] This invention is a system that records a user's conversations and voices while they are alive and uses AI to learn from them, enabling conversations with the deceased person even after they have passed away. This system includes the user, terminal, and server elements, each of which plays a specific role. Furthermore, by combining it with an emotion engine, the system aims to provide natural conversations that correspond to the user's emotions.

[0968] Audio data recording

[0969] A user launches a dedicated recording app and leaves a voice message. For example, the user might record a message like, "Today, I went on a trip with my family. It was a lot of fun." This recording reflects the user's personality and speech characteristics.

[0970] Uploading and preprocessing audio data

[0971] The device records this voice data, converts it into an appropriate format (e.g., WAV or MP3), and uploads it to a server. The uploaded voice data undergoes preprocessing, such as noise reduction and volume normalization, on the server. After preprocessing, the voice data is converted into text data using speech recognition technology.

[0972] Training a machine learning model

[0973] The server uses the voice data and corresponding text data to train a machine learning model. The model learns the user's speech characteristics and conversation style and is able to generate natural, authentic responses. The server also periodically incorporates newly uploaded data to improve the accuracy of the model.

[0974] Emotion recognition by emotion engine

[0975] The server uses an emotion engine to extract emotional information from the user's voice data. This emotional information is used as additional data when the machine learning model learns the user's speech characteristics. By taking emotional information into account, more natural and emotionally relevant responses are generated.

[0976] Generating conversations after a user dies

[0977] After a user passes away, a family member or friend launches a conversation app on the device. The device records the family member or friend's speech and sends the audio data to a server. The server analyzes the audio data and converts it into text. After conversion, an emotion engine is used to analyze emotions and generate an appropriate response in real time. The generated response is sent to the device as text.

[0978] Dialogue playback

[0979] The device converts the text data received from the server into voice and plays back the response on behalf of the user. This allows family and friends to feel as if they are being spoken by the user. For example, if a family member asks, "How was your day?", the device will play back a voice response that sounds like the user, such as, "I went shopping today. The weather was nice, so it felt good."

[0980] Specific examples

[0981] 1. The user launches the app and records a voice message saying, "I went to the park with my dog ​​today. It was fun!"

[0982] 2. The device records this audio and uploads it to the server.

[0983] 3. The server removes noise and normalizes the volume of the audio data, generating text data such as "I went to the park with my dog ​​today. It was fun!"

[0984] 4. The server uses an emotion engine to extract emotion information from the voice data.

[0985] 5. The server uses a machine learning model to learn the user's emotional information along with their speech characteristics.

[0986] 6. One day, a family member uses a conversation app to ask, "How's your day going?"

[0987] 7. The device sends the audio to the server, which generates a response saying, "I went for a walk today. I was happy because the weather was nice."

[0988] 8. The device will play back the response as audio, providing a natural conversational experience for family members.

[0989] By combining the user's voice data from their lifetime with machine learning technology and an emotion engine, this invention makes it possible to realize natural conversations that reflect the user's personality and are sensitive to their emotions even after they have passed away.

[0990] The processing flow will be explained below.

[0991] Step 1:

[0992] The user starts a dedicated recording application and records a voice message. For example, the user might record a message such as, "Today, I went on a trip with my family. It was a lot of fun."

[0993] Step 2:

[0994] The device detects that the record button has been pressed and starts recording audio data. The recorded data is temporarily saved in local storage.

[0995] Step 3:

[0996] The device will recognize that the end recording button has been pressed and will stop recording. The recorded data will be converted to a digital format (e.g., WAV or MP3 format).

[0997] Step 4:

[0998] The device uploads the recording data to a server using a secure communication protocol (e.g., HTTPS).

[0999] Step 5:

[1000] The server receives the uploaded voice data, stores it in a database, confirms receipt of the data, and moves on to the next processing step.

[1001] Step 6:

[1002] The server starts pre-processing the audio data, performing noise reduction and volume normalization to improve the audio quality.

[1003] Step 7:

[1004] The server uses speech recognition technology to convert the preprocessed speech data into text data, for example, "Today, I went on a trip with my family."

[1005] Step 8:

[1006] The server uses an emotion engine to extract the user's emotional information from the voice data. For example, it recognizes positive emotions from voice characteristics such as "fun."

[1007] Step 9:

[1008] The server inputs voice data, text data, and emotional information into the machine learning model, which then learns responses based on the user's speech characteristics and emotions.

[1009] Step 10:

[1010] The server trains the machine learning model to generate natural-looking dialogue that reflects the user's emotions, and periodically ingests new data to update the model.

[1011] Step 11:

[1012] After the death of the user, a family member or friend can use the device to launch the conversation app and press the "Start conversation" button to begin the conversation.

[1013] Step 12:

[1014] The device records voice input from family and friends and sends the voice data to a server, for example, recording questions like "How was your day?"

[1015] Step 13:

[1016] The server analyzes the received voice data and converts it into text. After conversion, an emotion engine is used to extract emotional information and generate an appropriate response in real time.

[1017] Step 14:

[1018] The server sends the generated response text to the terminal, for example, "I went shopping today. The weather was nice, so it felt good."

[1019] Step 15:

[1020] The device converts the response text received from the server into voice and plays it back to the family, enabling natural conversation.

[1021] By taking specific actions at each step, the system is able to provide natural, emotionally sensitive conversations that reflect the user's personality, even after the user has passed away.

[1022] Example 2

[1023] 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."

[1024] Currently, there is no system that records a user's conversations and voices while they are alive and allows them to have a conversation with the user even after they have passed away. In particular, it is difficult to provide natural conversations that reflect the user's speech characteristics and emotions. The purpose of this invention is to provide natural conversations that reflect the user's personality and are in tune with their emotions even after the user has passed away.

[1025] 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.

[1026] In this invention, the server includes means for recording user voice data, means for converting the format of the recorded voice data by the terminal and uploading it to the server, means for preprocessing the uploaded voice data and converting it into text data, means for performing machine learning based on the voice data and text data to learn the user's speech characteristics and emotional information, means for extracting emotional information from the voice data using an emotion engine and using it as training data for a machine learning model, means for the server to generate a dialogue in real time after the user's death and send it to the terminal, and means for the terminal to play back the generated dialogue as audio. This makes it possible to provide natural dialogue that reflects the user's individuality and is in tune with their emotions.

[1027] A "user" is an individual who provides voice data to the system.

[1028] "Audio data" refers to data of a voice message generated by a user using a recording application.

[1029] A "terminal" is an electronic device that a user uses to record audio data and upload it to a server.

[1030] The "server" is a central processing unit responsible for preprocessing voice data, training machine learning models, generating dialogue, and more.

[1031] "Format conversion" is the process of changing audio data into an appropriate format, such as WAV or MP3 format.

[1032] "Noise reduction" is a process for removing unnecessary background noise contained in audio data.

[1033] "Volume normalization" is a process for making the volume level of audio data uniform.

[1034] "Text data" is voice data that has been analyzed and converted into text information.

[1035] "Machine learning" is the process of using data to train a model and learn patterns and rules from the data.

[1036] "Speech characteristics" refer to the characteristics of a user's voice and speaking style.

[1037] "Emotion information" is data that expresses the user's emotions and feelings extracted from the voice data.

[1038] An "emotion engine" is a technology or system for extracting emotional information from voice data.

[1039] "Dialogue generation" is the process of creating natural responses and dialogue based on data learned by a machine learning model.

[1040] "Real-time" refers to time characteristics that result in near-instant processing and response.

[1041] "Audio playback" means converting text data into audio and outputting it as if the user were speaking.

[1042] This invention is a system that records a user's voice data while they are alive and uses a machine learning model to learn from it, enabling conversations that sound like the user even after they have passed away. This system mainly uses the user, a terminal, and a server. The specific operation of each element is described below.

[1043] First, the user launches a dedicated recording app and records a voice message. For example, a user might record a message like, "Today, I went on a family trip. It was a lot of fun." This voice data reflects the user's personality and speech characteristics in detail. The recording app runs on a device such as a smartphone or tablet, and the user taps the "Start Recording" button to start recording and the "Stop Recording" button to end recording.

[1044] The device then converts the recorded audio data into an appropriate format (e.g., WAV or MP3). After conversion, the device uploads the audio data to the server. The server then applies a noise reduction filter to the uploaded audio data and normalizes the volume. Specifically, open-source tools such as FFmpeg are used. This preprocessing improves the quality of the audio data and optimizes it for subsequent processing.

[1045] The server converts the preprocessed voice data into text data using speech recognition technology. This process uses speech recognition services such as IBM Watson or Google Cloud Speech-to-Text. A machine learning model is then trained using the converted text data and the original voice data. Examples of models used include OpenAI's GPT-3. The model is trained using machine learning frameworks such as TensorFlow and PyTorch. During the training process, the model learns the user's speech characteristics and dialogue style.

[1046] The server then uses an emotion engine, such as IBM Watson's natural language understanding (NLU) service, to extract emotional information from the voice data. The extracted emotional information is used as additional data for machine learning models to help generate more natural and emotionally relevant responses.

[1047] After the death of a user, family and friends can use the device to launch a conversation app and ask questions such as, "How was your day?" The device records the voice and sends it to a server. The server analyzes the voice data and converts it into text. The server then uses an emotion engine to analyze emotions and generate an appropriate response in real time. This response is then sent to the device in text format.

[1048] Finally, the device converts the text data sent from the server into speech using speech synthesis technology. For example, it uses speech synthesis technology such as Google TTS (Text-to-Speech). This allows family and friends to feel as if the user is speaking. For example, if a family member asks, "How was your day?", the device will respond in a voice that sounds like the user, such as, "I went shopping today. The weather was nice, so it felt good."

[1049] Prompt Sentence Examples

[1050] "Imagine a scenario where you're speaking for a user whose family member has passed away. For example, create an example response to the question, 'How was your day?'"

[1051] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1052] Step 1:

[1053] The user starts a dedicated recording application and records a voice message.

[1054] Input: User's spoken utterance

[1055] Output: Recorded audio data (WAV or MP3 format)

[1056] Specific operation: The user launches the app, taps the "Start Recording" button to record audio, and taps the "Stop Recording" button to end the recording. For example, the user can record a voice message such as "Today, I went on a trip with my family. It was a lot of fun."

[1057] Step 2:

[1058] The device converts the recorded audio data into an appropriate format and uploads it to the server.

[1059] Input: Recorded audio data (WAV or MP3 format)

[1060] Output: Audio data uploaded to the server

[1061] Specific operation: The device converts the audio data using a format conversion tool (e.g., FFmpeg) and uploads it to a server via the Internet.

[1062] Step 3:

[1063] The server preprocesses the uploaded audio data and converts it into text data.

[1064] Input: Uploaded audio data

[1065] Output: Denoised and volume-normalised audio data, converted text data

[1066] Specific operation: The server applies a noise reduction filter, normalizes the volume, and converts the speech to text using speech recognition technology (e.g., Google Cloud Speech-to-Text). For example, speech data such as "Today, I went on a trip with my family. It was a lot of fun" is converted into text data such as "Today, I went on a trip with my family. It was a lot of fun."

[1067] Step 4:

[1068] The server uses the audio and text data to train a machine learning model.

[1069] Input: Preprocessed audio data, converted text data

[1070] Output: A trained machine learning model

[1071] How it works: The server uses the voice and text data to train a machine learning model (e.g., GPT-3) using TensorFlow or PyTorch. The model learns the user's speech characteristics and interaction style.

[1072] Step 5:

[1073] The server extracts emotion information from the voice data using an emotion engine.

[1074] Input: Audio data

[1075] Output: Extracted emotion information

[1076] What it does: The server uses sentiment analysis tools (e.g., IBM Watson NLU) to extract emotional information from the voice data and add it to the training data for the machine learning model, enabling natural, emotionally relevant responses.

[1077] Step 6:

[1078] After a user passes away, if family or friends launch a conversation app and speak to them, the device will record the audio and send it to the server.

[1079] Input: Voice utterances from family and friends

[1080] Output: Sending audio data to the server

[1081] Specific operation: When a family member or friend asks, "How was your day?", the device records the audio, preprocesses it, and sends it to the server.

[1082] Step 7:

[1083] The server analyzes the voice data, converts it into text, and then uses an emotion engine to analyze emotions and generate an appropriate response.

[1084] Input: Transmitted audio data

[1085] Output: The generated response text

[1086] How it works: The server uses speech recognition technology to convert the voice data into text, performs sentiment analysis, and uses machine learning models to generate an appropriate response, such as "I went for a walk today. I was happy because the weather was nice."

[1087] Step 8:

[1088] The device converts the generated response text into speech and plays it back to family and friends.

[1089] Input: Generated response text

[1090] Output: A spoken response

[1091] What it does: Your device uses text-to-speech technology (e.g., Google TTS) to convert the text into speech and play it back to your family or friends, providing a voice response that sounds like you.

[1092] (Application example 2)

[1093] 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."

[1094] In conventional virtual store systems, customer service staff respond mechanically and without emotion, lacking friendliness or individualized consideration for visitors. Furthermore, there was a need for a system that could retain a user's speech characteristics and enable natural conversations even after their death. Therefore, a new technology was needed that could utilize the user's voice data while they were alive to understand their emotions and provide friendly customer service in real time.

[1095] 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 recording user voice data, means for uploading the recorded voice data to the server, means for preprocessing the uploaded voice data and converting it into text data, means for performing machine learning on the voice data and text data to learn the user's speech characteristics, means for the server to generate a dialogue in real time after the user's death and transmit it to the terminal, means for the terminal to play back the generated dialogue as voice, and means for providing emotionally considerate customer service to visitors in the virtual store. This makes it possible to provide emotionally considerate and friendly customer service by utilizing the user's speech characteristics.

[1096] "User's voice data" refers to voice information recorded by a user before death, and includes individual speech characteristics and emotional expressions.

[1097] A "server" is a computer system that processes voice data, converts it, performs machine learning, generates dialogue, and stores data.

[1098] A "recording means" is a device or software that has the function of collecting and storing audio data.

[1099] "Uploading means" refers to a device or software that has the function of transmitting recorded audio data to a server via a network.

[1100] "Preprocessing" refers to the process of performing processes such as noise removal and volume normalization on audio data.

[1101] "Means for converting into text data" refers to technology for analyzing voice data and converting its contents into text format.

[1102] "Means for performing machine learning" refers to a technology that uses voice data and text data to learn a user's speech characteristics and dialogue style using a specific algorithm.

[1103] The "means for generating dialogue in real time" refers to a technology that has the function of enabling a server to generate dialogue with a visitor in real time after the death of a user and transmit the content of that dialogue to a terminal.

[1104] The "means for the terminal to play back the dialogue generated as audio" refers to a device or software that converts text data sent from the server into audio and plays it back to the visitor.

[1105] "Means for providing emotionally sensitive customer service" are technologies and systems that provide more natural and friendly interactions in response to visitors' emotions and needs.

[1106] The present invention is a system that records voice data from a user's life and realizes natural conversations that take into consideration emotions. This system is mainly composed of a user, a terminal, and a server. Specific embodiments for implementing the present invention are described below.

[1107] Audio data recording

[1108] The user starts a dedicated voice recording application and records a voice message about an episode, emotion, or event in their life. For example, they can leave a voice message such as, "I bought a new jacket today. I'm so happy I made such a good purchase." This recording becomes the basic data for generating natural dialogue based on the user's speech characteristics and emotional expressions.

[1109] Uploading and preprocessing audio data

[1110] The device converts this audio data into WAV or MP3 format and uploads it to the server. On the server side, preprocessing such as noise removal and volume normalization is first performed to generate clear audio data. This audio data is then converted into text data using speech recognition technology. A Wav2Vec2 model or similar is typically used.

[1111] Training a machine learning model

[1112] The server trains a machine learning model using voice data and the corresponding text data. The trained model learns the user's speech characteristics and dialogue style and uses the data to reproduce those characteristics. Generative AI models such as OpenAI's GPT-3 are often used for training. In addition, by combining an emotion engine, the user's emotional expressions are also included in the learning.

[1113] Generating conversations after a user dies

[1114] After a user passes away, family and friends can use the application to interact with the user in a virtual store. For example, if a family member asks, "What products do you have recommended today?", the server generates a response in real time and sends it to the device as text data.

[1115] Dialogue playback

[1116] The device converts the text data received from the server into speech and plays it back in a voice that matches the user's speech characteristics. For this purpose, the TextToSpeech module is generally used. For example, it responds naturally, saying, "We have new jackets in stock today. Please take a look."

[1117] Examples of concrete examples and prompts

[1118] As a specific example, if a family member asks in a virtual store, "What recommended items do you have today?", the server will generate a natural response such as, "Hello! We have new jackets in stock today. Please take a look," and the device will play that voice back.

[1119] Example prompt for a generative AI model:

[1120] plaintext

[1121] User: What products do you have recommended today?

[1122] Bot:

[1123] In this way, it is possible to provide visitors with emotionally sensitive and natural dialogue based on the user's voice data in life.

[1124] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1125] Step 1:

[1126] A user starts a dedicated voice recording application and records a voice message. The input is the user's speech, and the output is a recorded voice file. The recorded voice is a recording of the user's natural speech, including everyday events and emotions.

[1127] Step 2:

[1128] The device converts the recorded audio data into the appropriate format (WAV or MP3). The input is the audio file obtained in step 1, and the output is the converted audio file. This conversion makes the audio file in a format that can be uploaded to the server.

[1129] Step 3:

[1130] The device uploads the audio file to the server. The input is the converted audio file, and the output is the audio file stored on the server. The audio data is then ready for processing on the server.

[1131] Step 4:

[1132] The server preprocesses the uploaded audio data. Preprocessing includes noise reduction and volume normalization. The input is an audio file stored on the server, and the output is a preprocessed audio file. Preprocessing improves the quality of the audio data.

[1133] Step 5:

[1134] The server converts the preprocessed audio data into text data using speech recognition technology. The input is the preprocessed audio file, and the output is text data transcribed from the audio data. Models such as "Wav2Vec2" are used for speech recognition.

[1135] Step 6:

[1136] The server trains a machine learning model using voice and text data. The input is the voice data and corresponding text data, and the output is a machine learning model that has learned the user's speech characteristics. Generative AI models such as "GPT-3" are used for training.

[1137] Step 7:

[1138] After the death of the user, family and friends use the device to launch a conversation app. The input is the speech of the family and friends, and the output is the voice data sent to the server. The device records the visitor's speech and sends it to the server.

[1139] Step 8:

[1140] The server analyzes the visitor's voice data and converts it into text data. The input is the visitor's voice data, and the output is text data. The server generates a response in real time based on the converted text data.

[1141] Step 9:

[1142] The server generates a response in real time using a generative AI model. The input is the visitor's text data, and the output is the generated response text. An example of a prompt for the generative AI model is as follows:

[1143] plaintext

[1144] User: What products do you have recommended today?

[1145] Bot:

[1146] Step 10:

[1147] The terminal converts the response text received from the server into speech and plays it back to the visitor. The input is the generated response text and the output is the audio data. The terminal uses the TextToSpeech module to generate audio and plays it back to the visitor as a natural dialogue.

[1148] 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.

[1149] 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.

[1150] 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.

[1151] [Fourth embodiment]

[1152] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1153] 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.

[1154] 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).

[1155] 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.

[1156] 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.

[1157] 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).

[1158] 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.

[1159] 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.

[1160] 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.

[1161] 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.

[1162] 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.

[1163] 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.

[1164] 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."

[1165] This invention is a system that records the conversations and voices of a user while they are alive, and uses AI to learn from them, allowing for conversations with the user even after they have passed away. This system includes elements such as the user, a terminal, and a server, each of which plays a specific role.

[1166] Audio data recording

[1167] The user launches a dedicated recording app and leaves a voice message. For example, they can record a message like, "Today, I went on a trip with my family. It was a lot of fun." This recording reflects the user's personality and speech characteristics.

[1168] Uploading and preprocessing audio data

[1169] The device records this voice data, converts it into an appropriate format (e.g., WAV or MP3), and uploads it to a server. The uploaded voice data undergoes preprocessing, such as noise reduction and volume normalization, on the server. After preprocessing, the voice data is converted into text data using speech recognition technology.

[1170] Training a machine learning model

[1171] The server uses the voice data and corresponding text data to train a machine learning model. The model learns the user's speech characteristics and conversation style and is able to generate natural, authentic responses. The server also periodically incorporates newly uploaded data to improve the accuracy of the model.

[1172] Generating conversations after a user dies

[1173] After a user's death, a family member or friend can use the device to launch a conversation app. The device records the family member or friend's speech and sends the audio data to a server. The server analyzes the audio data and generates an appropriate response in real time. The generated response is sent to the device as text.

[1174] Dialogue playback

[1175] The device converts the text data received from the server into speech and plays back a response on behalf of the user. This allows family and friends to feel as if they are being spoken by the user. For example, if a family member asks, "How was your day?", the device will respond in a voice that sounds like the user, such as, "I went shopping today."

[1176] Specific examples

[1177] 1. The user launches the app and records a voice message saying, "I went to the park with my dog ​​today. It was fun!"

[1178] 2. The device records this audio and uploads it to the server.

[1179] 3. The server removes noise and normalizes the volume of the audio data, generating text data such as "I went to the park with my dog ​​today. It was fun!"

[1180] 4. The server uses a machine learning model to learn the user's speech characteristics.

[1181] 5. One day, a family member uses a conversation app to ask, "How's your day going?"

[1182] 6. The device sends the audio to the server, which generates a response saying, "I went for a walk today. The weather was nice."

[1183] 7. The device will play back the response as audio, providing a natural conversational experience for family members.

[1184] By combining the user's voice data from before their death with machine learning technology, the present invention makes it possible to realize natural conversations that sound like the user, even after the user has passed away.

[1185] The processing flow will be explained below.

[1186] Step 1:

[1187] The user launches a recording app and starts speaking. For example, the user might record a voice message such as, "Today, my family and I went to the park."

[1188] Step 2:

[1189] The device detects that the record button has been pressed and starts recording audio data. The recorded data is temporarily saved in local storage.

[1190] Step 3:

[1191] The device will recognize that the end recording button has been pressed and will stop recording. The recorded data will be converted to a digital format (e.g., WAV or MP3 format).

[1192] Step 4:

[1193] The device uploads the recording data to a server using a secure communication protocol (e.g., HTTPS).

[1194] Step 5:

[1195] The server receives the uploaded audio data, stores it in a database, and starts preprocessing the audio data.

[1196] Step 6:

[1197] The server performs noise reduction and volume normalization on the audio data, which improves the audio quality.

[1198] Step 7:

[1199] The server uses speech recognition technology to convert the preprocessed speech data into text data, for example, "Today, my family and I went to the park."

[1200] Step 8:

[1201] The server inputs the generated text data and corresponding audio data into a machine learning model, which begins the process of learning the user's speech characteristics.

[1202] Step 9:

[1203] The server trains the machine learning model, learning the user's speech patterns and characteristics to generate highly accurate responses.

[1204] Step 10:

[1205] The server evaluates the machine learning model and retrains it as needed, thereby maintaining the model's accuracy.

[1206] Step 11:

[1207] After the death of the user, a family member or friend can use the device to launch the conversation app and press the "Start conversation" button to begin the conversation.

[1208] Step 12:

[1209] The device records voice input from family and friends and sends the voice data to a server, for example, recording a question like "How was your day?"

[1210] Step 13:

[1211] The server analyzes the received voice data and converts it into text. Based on the converted data, it uses a machine learning model to generate an appropriate response.

[1212] Step 14:

[1213] The server sends the generated response text to the terminal, for example, generating a response such as "I went shopping today."

[1214] Step 15:

[1215] The device converts the response text received from the server into voice and plays it back to the family, enabling natural conversation.

[1216] By taking specific actions at each step, the system can provide natural dialogue that reflects the user's personality even after the user has passed away.

[1217] Example 1

[1218] 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."

[1219] There is a lack of technology to respond to the requests of family and friends who want to continue interacting with a deceased person. Furthermore, insufficient preprocessing of recorded audio data and insufficient training of machine learning models can lead to poor dialogue quality. Furthermore, it is difficult to accurately model the user's speech characteristics and generate natural dialogue.

[1220] 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.

[1221] In this invention, the server includes means for preprocessing the voice data, removing noise and normalizing the volume, and converting the voice data into text data, means for performing machine learning based on the voice data and text data to learn the user's speech characteristics, and means for the server to generate a dialogue in real time after the user's death and transmit it to the terminal, thereby enabling family and friends to experience natural dialogue even after the user's death.

[1222] "Voice data" is digital data that records a user's speech or voice message.

[1223] A "terminal" is an electronic device on which a recording application or a dialogue application used by a user is installed.

[1224] "Convert to format" refers to the process of converting audio data into an appropriate audio file format (e.g., WAV, MP3).

[1225] "Upload" refers to sending data recorded on a terminal to a server via a network.

[1226] A "server" is a central management computer that stores received data and performs various data processing and machine learning.

[1227] "Preprocessing" refers to performing data cleansing operations on the audio data, such as noise removal and volume normalization.

[1228] "Noise reduction" is a process for removing unnecessary background sounds and noise from audio data.

[1229] "Volume normalization" is a process for making the volume level of audio data uniform.

[1230] "Text data" is voice data converted into character data using voice recognition technology.

[1231] A "machine learning model" is an algorithm designed to learn from large amounts of data and perform a specific task.

[1232] "Speech characteristics" refers to the voice characteristics and speaking style of an individual user.

[1233] "Generating dialogue in real time" refers to generating and providing responses instantly as users take turns speaking.

[1234] "Converting to speech" means regenerating text data as speech using speech synthesis technology.

[1235] "Playback" means outputting the generated audio from a playback device such as a speaker.

[1236] "Family and friends" refers to people with whom the user will communicate after their death.

[1237] This invention is a system that records the conversations and voices of a user while they are alive, and uses AI to learn from them, allowing for conversations with the user even after they have passed away. This system includes elements such as the user, a terminal, and a server, each of which plays a specific role.

[1238] First, the user launches a dedicated recording app and leaves a voice message. This voice message reflects the user's personality and speech characteristics and includes specific content such as, "Today, I went on a trip with my family. It was a lot of fun."

[1239] The device then records this audio data, converts it into an appropriate format (e.g., WAV or MP3), and uploads it to a server via the Internet. The audio data format conversion is performed using software installed on the device (e.g., FFMPEG).

[1240] The server performs preprocessing on the uploaded audio data, such as noise removal and volume normalization. This preprocessing uses a noise removal algorithm (e.g., Audacity's noise removal function) and a volume normalization algorithm. After preprocessing is complete, the server converts the audio data into text data using speech recognition technology (e.g., Google Speech-to-Text API). For example, the generated text data is, "Today, I went on a trip with my family. It was a lot of fun."

[1241] The server then uses the voice data and corresponding text data to train a machine learning model (e.g., OpenAI GPT-4). The model learns the user's speech characteristics and conversational style and is able to generate natural, authentic responses. The server also periodically incorporates newly uploaded data to improve the model's accuracy.

[1242] After a user passes away, family and friends can use the device to launch a conversation app. When they speak into the microphone, the device records their voice and sends it to the server in real time. For example, if a family member asks, "How was your day?", the server receives the voice data and converts it into text using speech recognition technology. Then, a machine learning model generates a response such as, "I went for a walk today. The weather was nice."

[1243] The generated response is sent as text to the device, which then converts this text data into speech (e.g., using Amazon Polly to generate synthetic speech), and finally plays the generated speech through the device's speaker, giving family and friends the feeling that it is the user speaking.

[1244] Examples of prompt statements

[1245] Prompt: Convert the following audio data into text and use that text to generate a natural-sounding response based on the user's speaking characteristics.

[1246] Audio data: I went to the park with my dog ​​today. It was fun!

[1247] By combining a user's voice data with machine learning technology, this invention enables a user to have natural, lifelike conversations even after they have passed away. This system provides a way for family and friends to continue to relive precious memories.

[1248] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1249] Step 1:

[1250] The user launches a recording app and records a voice message. The user speaks into the microphone, saying, "Today, I went on a trip with my family. It was a lot of fun." The recording app captures this voice data in WAV format and saves it on the device.

[1251] Input: User's voice

[1252] Output: WAV format audio file

[1253] Step 2:

[1254] The device converts the stored audio data into MP3 format using software installed on the device (e.g., FFMPEG), and then uploads the converted MP3 file to a server via the Internet.

[1255] Input: WAV format audio file

[1256] Output: MP3 audio file

[1257] Step 3:

[1258] The server receives the uploaded audio data and stores it in storage. Then, preprocessing begins. Specifically, the server performs noise reduction and volume normalization on the audio data. This preprocessing uses a noise reduction algorithm (e.g., Audacity's noise reduction function) and a volume normalization algorithm.

[1259] Input: MP3 audio file

[1260] Output: Preprocessed audio file

[1261] Step 4:

[1262] The server uses the preprocessed voice data to convert it into text data using speech recognition technology (e.g., Google Speech-to-Text API). For example, the generated text data is, "Today, I went on a trip with my family. It was a lot of fun."

[1263] Input: Preprocessed audio file

[1264] Output: Text data

[1265] Step 5:

[1266] The server trains a machine learning model (e.g., OpenAI GPT-4) using the voice data and corresponding text data. Through this, the server learns the user's speech characteristics and interaction style. The server also periodically incorporates newly uploaded data and continuously retrains the model to improve its accuracy.

[1267] Input: Text and audio data

[1268] Output: A trained machine learning model

[1269] Step 6:

[1270] After the death of the user, family and friends can launch a conversation app and speak into the microphone, for example, saying, "How was your day?" The device will record the voice and send it to the server in real time.

[1271] Input: Voice of family or friends

[1272] Output: Recorded audio data

[1273] Step 7:

[1274] The server converts the received voice data into text data using speech recognition technology. For example, the text "How was your day?" is generated. Then, using a machine learning model trained on the server, it generates a response such as "I went for a walk today. The weather was nice."

[1275] Input: Recorded audio data

[1276] Output: Response text data

[1277] Step 8:

[1278] The device converts the response text data received from the server into speech, using speech synthesis technology (e.g., Amazon Polly) to generate a speech that says, "I went for a walk today. The weather was nice."

[1279] Input: Response text data

[1280] Output: Audio data

[1281] Step 9:

[1282] The device then plays the generated audio through the speaker, allowing family and friends to hear it as if you were actually speaking.

[1283] Input: Audio data

[1284] Output: Played audio

[1285] (Application example 1)

[1286] 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."

[1287] As the aging society progresses, many people want to cherish memories and conversations with their deceased loved ones. However, existing technology makes it difficult to realize natural conversations with the deceased. It is necessary to solve this problem and provide a new customer experience that enriches memories with the deceased.

[1288] 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.

[1289] In this invention, the server includes means for recording user voice data, means for uploading the recorded voice data to the server, means for preprocessing the uploaded voice data and converting it into text data, means for performing machine learning based on the voice data and text data to learn the user's speech characteristics, means for the server to generate a dialogue in real time after the user's death and transmit it to the terminal, means for the terminal to play back the generated dialogue as audio and provide a customer dialogue experience, and means for providing the customer with a dialogue experience with the deceased via a head-mounted display or smartphone, thereby allowing customers to experience a natural dialogue with the deceased and enrich their memories.

[1290] "User voice data" refers to voice information uttered by a user, and is digital data that can be recorded and analyzed.

[1291] A "server" is a computer system that stores, processes, and manages data on a network, and is responsible for managing user voice data and training machine learning models.

[1292] "Recording means" refers to the functionality of the device or software for recording and storing a user's voice data.

[1293] "Means for uploading to a server" refers to the functionality of a device or software for transmitting recorded audio data to a server via a network.

[1294] "Preprocessing" refers to the process of removing noise and normalizing the volume of uploaded audio data to make it easier to analyze.

[1295] "Means for converting into text data" refers to the technology and functions for analyzing voice data and converting it into text data.

[1296] "Means for conducting machine learning to learn a user's speech characteristics" refers to technologies and functions that use voice data and text data to train a model on a user's speech patterns and characteristics.

[1297] "Means for generating dialogue in real time" refers to technologies and functions that analyze dialogue with family and friends on the spot and instantly generate appropriate responses.

[1298] "Terminal" refers to devices used by users, such as computers, smartphones, and head-mounted displays.

[1299] "Means for playing as audio" refers to the technology or functionality for converting text data into audio and playing it back on a user device.

[1300] "Means for providing a customer interaction experience" refers to technologies and functions that provide natural and individual interactions with users via terminals and realize interactions with users.

[1301] A "head-mounted display" is a display device worn by a user that provides information visually and audibly.

[1302] A "smartphone" is a portable information terminal that combines the functions of a mobile phone with those of a personal computer, and is a device that can record, upload, and play back audio data.

[1303] The present invention provides a system for providing a conversational experience even after a user has passed away by recording the user's voice data and training a machine learning model based on the recorded voice data. This system is particularly applicable to customer conversational experiences using head-mounted displays and smartphones. The components and processing steps of this system are described in detail below.

[1304] Recording and uploading audio data

[1305] Users use a dedicated recording app to record everyday conversations and messages. This recording is done via a device such as a smartphone. The recorded voice data is saved on the device and then uploaded to a server.

[1306] Audio data preprocessing and text conversion

[1307] The server performs preprocessing on the uploaded audio data, removing noise and normalizing the volume, making the data easier to analyze. After preprocessing, the audio data is converted into text using speech recognition technology.

[1308] Training a machine learning model

[1309] The server uses the preprocessed audio data and corresponding text data to train a machine learning model. The model learns the user's speech characteristics and conversational style and generates natural-sounding responses. This allows the model to reproduce a user's conversations in the future, even after the user has passed away.

[1310] Dialogue generation and playback

[1311] After the death of a user, family and friends launch a dedicated conversation app on a head-mounted display or smartphone. The device records the family and friends' speech and sends the audio data to a server. The server analyzes the audio data in real time and generates an appropriate response. The generated response is sent as text data to the device, which converts it into audio and plays it back.

[1312] Specific examples

[1313] 1. A user uses a dedicated app to record a voice message saying, "I went to the park with my dog ​​today. It was fun!"

[1314] 2. The device uploads this audio to the server.

[1315] 3. The server performs preprocessing and generates text data such as "I went to the park with my dog ​​today. It was fun!"

[1316] 4. The server uses a machine learning model to learn the user's speech characteristics.

[1317] 5. One day, a family member uses a conversation app to ask, "How's your day going?"

[1318] 6. The device sends the audio to the server, which generates a response saying, "I went for a walk today. The weather was nice."

[1319] 7. The device will play back the response as audio, providing a natural conversational experience for family members.

[1320] Prompt Sentence Examples

[1321] "User-recorded question: 'Mom, what would you recommend today?'"

[1322] In this way, this invention combines the user's voice data from before death with machine learning technology to enable natural conversations that sound like the user even after the user has passed away. Furthermore, this system is expected to have a wide range of applications when used with head-mounted displays and smartphones.

[1323] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1324] Step 1:

[1325] The user launches a dedicated recording app and records a voice message. The input is the user's speech, and the output is the recorded voice data. Specifically, the user speaks into the smartphone's microphone and records a message such as, "Today, I went on a trip with my family. It was a lot of fun."

[1326] Step 2:

[1327] The device converts the recorded audio data into an appropriate format (e.g., WAV or MP3) and uploads it to the server. The input is the recorded audio data, and the output is the uploaded audio data. Specifically, the smartphone converts the recorded audio data and sends it to the server via the Internet.

[1328] Step 3:

[1329] The server performs preprocessing on the uploaded audio data, such as noise reduction and volume normalization. The input is the uploaded audio data, and the output is the preprocessed audio data. Specifically, the server runs the audio processing algorithm and processes the data to make it easier to analyze.

[1330] Step 4:

[1331] The server converts the preprocessed speech data into text data using speech recognition technology. The input is the preprocessed speech data, and the output is text data. The specific operation is to extract strings of characters from the speech signal using a speech recognition model (e.g., Wav2Vec2).

[1332] Step 5:

[1333] The server trains a machine learning model using voice data and the corresponding text data. The input is voice data and text data, and the output is a trained machine learning model. The specific operation is to input data into the model, and iteratively learns the user's speech characteristics.

[1334] Step 6:

[1335] After the death of a user, family and friends can launch a dialogue app and ask questions by voice. The input is the speech of the family or friend, and the output is recorded voice data. Specifically, family and friends use a smartphone or head-mounted display to record questions such as "How was your day?"

[1336] Step 7:

[1337] The device sends the audio data to the server. The input is the recorded audio data, and the output is the audio data uploaded to the server. The specific operation is that the smartphone sends the recorded audio data to the server.

[1338] Step 8:

[1339] The server analyzes the voice data in real time and generates an appropriate response. The input is the uploaded voice data, and the output is the generated text response. The specific operation is to use the speech recognition and generation AI model on the server to generate a response corresponding to the input voice data.

[1340] Step 9:

[1341] The device converts the text response received from the server into speech, recreating a natural conversational experience even after the user has passed away. The input is the generated text response, and the output is a speech response. The specific operation is that the smartphone or head-mounted display plays the text as speech using speech synthesis technology (e.g., Text-to-Speech).

[1342] 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.

[1343] This invention is a system that records a user's conversations and voices while they are alive and uses AI to learn from them, enabling conversations with the deceased person even after they have passed away. This system includes the user, terminal, and server elements, each of which plays a specific role. Furthermore, by combining it with an emotion engine, the system aims to provide natural conversations that correspond to the user's emotions.

[1344] Audio data recording

[1345] A user launches a dedicated recording app and leaves a voice message. For example, the user might record a message like, "Today, I went on a trip with my family. It was a lot of fun." This recording reflects the user's personality and speech characteristics.

[1346] Uploading and preprocessing audio data

[1347] The device records this voice data, converts it into an appropriate format (e.g., WAV or MP3), and uploads it to a server. The uploaded voice data undergoes preprocessing, such as noise reduction and volume normalization, on the server. After preprocessing, the voice data is converted into text data using speech recognition technology.

[1348] Training a machine learning model

[1349] The server uses the voice data and corresponding text data to train a machine learning model. The model learns the user's speech characteristics and conversation style and is able to generate natural, authentic responses. The server also periodically incorporates newly uploaded data to improve the accuracy of the model.

[1350] Emotion recognition by emotion engine

[1351] The server uses an emotion engine to extract emotional information from the user's voice data. This emotional information is used as additional data when the machine learning model learns the user's speech characteristics. By taking emotional information into account, more natural and emotionally relevant responses are generated.

[1352] Generating conversations after a user dies

[1353] After a user passes away, a family member or friend launches a conversation app on the device. The device records the family member or friend's speech and sends the audio data to a server. The server analyzes the audio data and converts it into text. After conversion, an emotion engine is used to analyze emotions and generate an appropriate response in real time. The generated response is sent to the device as text.

[1354] Dialogue playback

[1355] The device converts the text data received from the server into voice and plays back the response on behalf of the user. This allows family and friends to feel as if they are being spoken by the user. For example, if a family member asks, "How was your day?", the device will play back a voice response that sounds like the user, such as, "I went shopping today. The weather was nice, so it felt good."

[1356] Specific examples

[1357] 1. The user launches the app and records a voice message saying, "I went to the park with my dog ​​today. It was fun!"

[1358] 2. The device records this audio and uploads it to the server.

[1359] 3. The server removes noise and normalizes the volume of the audio data, generating text data such as "I went to the park with my dog ​​today. It was fun!"

[1360] 4. The server uses an emotion engine to extract emotion information from the voice data.

[1361] 5. The server uses a machine learning model to learn the user's emotional information along with their speech characteristics.

[1362] 6. One day, a family member uses a conversation app to ask, "How's your day going?"

[1363] 7. The device sends the audio to the server, which generates a response saying, "I went for a walk today. I was happy because the weather was nice."

[1364] 8. The device will play back the response as audio, providing a natural conversational experience for family members.

[1365] By combining the user's voice data from their lifetime with machine learning technology and an emotion engine, this invention makes it possible to realize natural conversations that reflect the user's personality and are sensitive to their emotions even after they have passed away.

[1366] The processing flow will be explained below.

[1367] Step 1:

[1368] The user starts a dedicated recording application and records a voice message. For example, the user might record a message such as, "Today, I went on a trip with my family. It was a lot of fun."

[1369] Step 2:

[1370] The device detects that the record button has been pressed and starts recording audio data. The recorded data is temporarily saved in local storage.

[1371] Step 3:

[1372] The device will recognize that the end recording button has been pressed and will stop recording. The recorded data will be converted to a digital format (e.g., WAV or MP3 format).

[1373] Step 4:

[1374] The device uploads the recording data to a server using a secure communication protocol (e.g., HTTPS).

[1375] Step 5:

[1376] The server receives the uploaded voice data, stores it in a database, confirms receipt of the data, and moves on to the next processing step.

[1377] Step 6:

[1378] The server starts pre-processing the audio data, performing noise reduction and volume normalization to improve the audio quality.

[1379] Step 7:

[1380] The server uses speech recognition technology to convert the preprocessed speech data into text data, for example, "Today, I went on a trip with my family."

[1381] Step 8:

[1382] The server uses an emotion engine to extract the user's emotional information from the voice data. For example, it recognizes positive emotions from voice characteristics such as "fun."

[1383] Step 9:

[1384] The server inputs voice data, text data, and emotional information into the machine learning model, which then learns responses based on the user's speech characteristics and emotions.

[1385] Step 10:

[1386] The server trains the machine learning model to generate natural-looking dialogue that reflects the user's emotions, and periodically ingests new data to update the model.

[1387] Step 11:

[1388] After the death of the user, a family member or friend can use the device to launch the conversation app and press the "Start conversation" button to begin the conversation.

[1389] Step 12:

[1390] The device records voice input from family and friends and sends the voice data to a server, for example, recording questions like "How was your day?"

[1391] Step 13:

[1392] The server analyzes the received voice data and converts it into text. After conversion, an emotion engine is used to extract emotional information and generate an appropriate response in real time.

[1393] Step 14:

[1394] The server sends the generated response text to the terminal, for example, "I went shopping today. The weather was nice, so it felt good."

[1395] Step 15:

[1396] The device converts the response text received from the server into voice and plays it back to the family, enabling natural conversation.

[1397] By taking specific actions at each step, the system is able to provide natural, emotionally sensitive conversations that reflect the user's personality, even after the user has passed away.

[1398] Example 2

[1399] 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."

[1400] Currently, there is no system that records a user's conversations and voices while they are alive and allows them to have a conversation with the user even after they have passed away. In particular, it is difficult to provide natural conversations that reflect the user's speech characteristics and emotions. The purpose of this invention is to provide natural conversations that reflect the user's personality and are in tune with their emotions even after the user has passed away.

[1401] 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.

[1402] In this invention, the server includes means for recording user voice data, means for converting the format of the recorded voice data by the terminal and uploading it to the server, means for preprocessing the uploaded voice data and converting it into text data, means for performing machine learning based on the voice data and text data to learn the user's speech characteristics and emotional information, means for extracting emotional information from the voice data using an emotion engine and using it as training data for a machine learning model, means for the server to generate a dialogue in real time after the user's death and send it to the terminal, and means for the terminal to play back the generated dialogue as audio. This makes it possible to provide natural dialogue that reflects the user's individuality and is in tune with their emotions.

[1403] A "user" is an individual who provides voice data to the system.

[1404] "Audio data" refers to data of a voice message generated by a user using a recording application.

[1405] A "terminal" is an electronic device that a user uses to record audio data and upload it to a server.

[1406] The "server" is a central processing unit responsible for preprocessing voice data, training machine learning models, generating dialogue, and more.

[1407] "Format conversion" is the process of changing audio data into an appropriate format, such as WAV or MP3 format.

[1408] "Noise reduction" is a process for removing unnecessary background noise contained in audio data.

[1409] "Volume normalization" is a process for making the volume level of audio data uniform.

[1410] "Text data" is voice data that has been analyzed and converted into text information.

[1411] "Machine learning" is the process of using data to train a model and learn patterns and rules from the data.

[1412] "Speech characteristics" refer to the characteristics of a user's voice and speaking style.

[1413] "Emotion information" is data that expresses the user's emotions and feelings extracted from the voice data.

[1414] An "emotion engine" is a technology or system for extracting emotional information from voice data.

[1415] "Dialogue generation" is the process of creating natural responses and dialogue based on data learned by a machine learning model.

[1416] "Real-time" refers to time characteristics that result in near-instant processing and response.

[1417] "Audio playback" means converting text data into audio and outputting it as if the user were speaking.

[1418] This invention is a system that records a user's voice data while they are alive and uses a machine learning model to learn from it, enabling conversations that sound like the user even after they have passed away. This system mainly uses the user, a terminal, and a server. The specific operation of each element is described below.

[1419] First, the user launches a dedicated recording app and records a voice message. For example, a user might record a message like, "Today, I went on a family trip. It was a lot of fun." This voice data reflects the user's personality and speech characteristics in detail. The recording app runs on a device such as a smartphone or tablet, and the user taps the "Start Recording" button to start recording and the "Stop Recording" button to end recording.

[1420] The device then converts the recorded audio data into an appropriate format (e.g., WAV or MP3). After conversion, the device uploads the audio data to the server. The server then applies a noise reduction filter to the uploaded audio data and normalizes the volume. Specifically, open-source tools such as FFmpeg are used. This preprocessing improves the quality of the audio data and optimizes it for subsequent processing.

[1421] The server converts the preprocessed voice data into text data using speech recognition technology. This process uses speech recognition services such as IBM Watson or Google Cloud Speech-to-Text. A machine learning model is then trained using the converted text data and the original voice data. Examples of models used include OpenAI's GPT-3. The model is trained using machine learning frameworks such as TensorFlow and PyTorch. During the training process, the model learns the user's speech characteristics and dialogue style.

[1422] The server then uses an emotion engine, such as IBM Watson's natural language understanding (NLU) service, to extract emotional information from the voice data. The extracted emotional information is used as additional data for machine learning models to help generate more natural and emotionally relevant responses.

[1423] After the death of a user, family and friends can use the device to launch a conversation app and ask questions such as, "How was your day?" The device records the voice and sends it to a server. The server analyzes the voice data and converts it into text. The server then uses an emotion engine to analyze emotions and generate an appropriate response in real time. This response is then sent to the device in text format.

[1424] Finally, the device converts the text data sent from the server into speech using speech synthesis technology. For example, it uses speech synthesis technology such as Google TTS (Text-to-Speech). This allows family and friends to feel as if the user is speaking. For example, if a family member asks, "How was your day?", the device will respond in a voice that sounds like the user, such as, "I went shopping today. The weather was nice, so it felt good."

[1425] Prompt Sentence Examples

[1426] "Imagine a scenario where you're speaking for a user whose family member has passed away. For example, create an example response to the question, 'How was your day?'"

[1427] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1428] Step 1:

[1429] The user starts a dedicated recording application and records a voice message.

[1430] Input: User's spoken utterance

[1431] Output: Recorded audio data (WAV or MP3 format)

[1432] Specific operation: The user launches the app, taps the "Start Recording" button to record audio, and taps the "Stop Recording" button to end the recording. For example, the user can record a voice message such as "Today, I went on a trip with my family. It was a lot of fun."

[1433] Step 2:

[1434] The device converts the recorded audio data into an appropriate format and uploads it to the server.

[1435] Input: Recorded audio data (WAV or MP3 format)

[1436] Output: Audio data uploaded to the server

[1437] Specific operation: The device converts the audio data using a format conversion tool (e.g., FFmpeg) and uploads it to a server via the Internet.

[1438] Step 3:

[1439] The server preprocesses the uploaded audio data and converts it into text data.

[1440] Input: Uploaded audio data

[1441] Output: Denoised and volume-normalised audio data, converted text data

[1442] Specific operation: The server applies a noise reduction filter, normalizes the volume, and converts the speech to text using speech recognition technology (e.g., Google Cloud Speech-to-Text). For example, speech data such as "Today, I went on a trip with my family. It was a lot of fun" is converted into text data such as "Today, I went on a trip with my family. It was a lot of fun."

[1443] Step 4:

[1444] The server uses the audio and text data to train a machine learning model.

[1445] Input: Preprocessed audio data, converted text data

[1446] Output: A trained machine learning model

[1447] How it works: The server uses the voice and text data to train a machine learning model (e.g., GPT-3) using TensorFlow or PyTorch. The model learns the user's speech characteristics and interaction style.

[1448] Step 5:

[1449] The server extracts emotion information from the voice data using an emotion engine.

[1450] Input: Audio data

[1451] Output: Extracted emotion information

[1452] What it does: The server uses sentiment analysis tools (e.g., IBM Watson NLU) to extract emotional information from the voice data and add it to the training data for the machine learning model, enabling natural, emotionally relevant responses.

[1453] Step 6:

[1454] After a user passes away, if family or friends launch a conversation app and speak to them, the device will record the audio and send it to the server.

[1455] Input: Voice utterances from family and friends

[1456] Output: Sending audio data to the server

[1457] Specific operation: When a family member or friend asks, "How was your day?", the device records the audio, preprocesses it, and sends it to the server.

[1458] Step 7:

[1459] The server analyzes the voice data, converts it into text, and then uses an emotion engine to analyze emotions and generate an appropriate response.

[1460] Input: Transmitted audio data

[1461] Output: The generated response text

[1462] How it works: The server uses speech recognition technology to convert the voice data into text, performs sentiment analysis, and uses machine learning models to generate an appropriate response, such as "I went for a walk today. I was happy because the weather was nice."

[1463] Step 8:

[1464] The device converts the generated response text into speech and plays it back to family and friends.

[1465] Input: Generated response text

[1466] Output: A spoken response

[1467] What it does: Your device uses text-to-speech technology (e.g., Google TTS) to convert the text into speech and play it back to your family or friends, providing a voice response that sounds like you.

[1468] (Application example 2)

[1469] 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."

[1470] In conventional virtual store systems, customer service staff respond mechanically and without emotion, lacking friendliness or individualized consideration for visitors. Furthermore, there was a need for a system that could retain a user's speech characteristics and enable natural conversations even after their death. Therefore, a new technology was needed that could utilize the user's voice data while they were alive to understand their emotions and provide friendly customer service in real time.

[1471] 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 recording user voice data, means for uploading the recorded voice data to the server, means for preprocessing the uploaded voice data and converting it into text data, means for performing machine learning on the voice data and text data to learn the user's speech characteristics, means for the server to generate a dialogue in real time after the user's death and transmit it to the terminal, means for the terminal to play back the generated dialogue as voice, and means for providing emotionally considerate customer service to visitors in the virtual store. This makes it possible to provide emotionally considerate and friendly customer service by utilizing the user's speech characteristics.

[1472] "User's voice data" refers to voice information recorded by a user before death, and includes individual speech characteristics and emotional expressions.

[1473] A "server" is a computer system that processes voice data, converts it, performs machine learning, generates dialogue, and stores data.

[1474] A "recording means" is a device or software that has the function of collecting and storing audio data.

[1475] "Uploading means" refers to a device or software that has the function of transmitting recorded audio data to a server via a network.

[1476] "Preprocessing" refers to the process of performing processes such as noise removal and volume normalization on audio data.

[1477] "Means for converting into text data" refers to technology for analyzing voice data and converting its contents into text format.

[1478] "Means for performing machine learning" refers to a technology that uses voice data and text data to learn a user's speech characteristics and dialogue style using a specific algorithm.

[1479] The "means for generating dialogue in real time" refers to a technology that has the function of enabling a server to generate dialogue with a visitor in real time after the death of a user and transmit the content of that dialogue to a terminal.

[1480] The "means for the terminal to play back the dialogue generated as audio" refers to a device or software that converts text data sent from the server into audio and plays it back to the visitor.

[1481] "Means for providing emotionally sensitive customer service" are technologies and systems that provide more natural and friendly interactions in response to visitors' emotions and needs.

[1482] The present invention is a system that records voice data from a user's life and realizes natural conversations that take into consideration emotions. This system is mainly composed of a user, a terminal, and a server. Specific embodiments for implementing the present invention are described below.

[1483] Audio data recording

[1484] The user starts a dedicated voice recording application and records a voice message about an episode, emotion, or event in their life. For example, they can leave a voice message such as, "I bought a new jacket today. I'm so happy I made such a good purchase." This recording becomes the basic data for generating natural dialogue based on the user's speech characteristics and emotional expressions.

[1485] Uploading and preprocessing audio data

[1486] The device converts this audio data into WAV or MP3 format and uploads it to the server. On the server side, preprocessing such as noise removal and volume normalization is first performed to generate clear audio data. This audio data is then converted into text data using speech recognition technology. A Wav2Vec2 model or similar is typically used.

[1487] Training a machine learning model

[1488] The server trains a machine learning model using voice data and the corresponding text data. The trained model learns the user's speech characteristics and dialogue style and uses the data to reproduce those characteristics. Generative AI models such as OpenAI's GPT-3 are often used for training. In addition, by combining an emotion engine, the user's emotional expressions are also included in the learning.

[1489] Generating conversations after a user dies

[1490] After a user passes away, family and friends can use the application to interact with the user in a virtual store. For example, if a family member asks, "What products do you have recommended today?", the server generates a response in real time and sends it to the device as text data.

[1491] Dialogue playback

[1492] The device converts the text data received from the server into speech and plays it back in a voice that matches the user's speech characteristics. For this purpose, the TextToSpeech module is generally used. For example, it responds naturally, saying, "We have new jackets in stock today. Please take a look."

[1493] Examples of concrete examples and prompts

[1494] As a specific example, if a family member asks in a virtual store, "What recommended items do you have today?", the server will generate a natural response such as, "Hello! We have new jackets in stock today. Please take a look," and the device will play that voice back.

[1495] Example prompt for a generative AI model:

[1496] plaintext

[1497] User: What products do you have recommended today?

[1498] Bot:

[1499] In this way, it is possible to provide visitors with emotionally sensitive and natural dialogue based on the user's voice data in life.

[1500] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1501] Step 1:

[1502] A user starts a dedicated voice recording application and records a voice message. The input is the user's speech, and the output is a recorded voice file. The recorded voice is a recording of the user's natural speech, including everyday events and emotions.

[1503] Step 2:

[1504] The device converts the recorded audio data into the appropriate format (WAV or MP3). The input is the audio file obtained in step 1, and the output is the converted audio file. This conversion makes the audio file in a format that can be uploaded to the server.

[1505] Step 3:

[1506] The device uploads the audio file to the server. The input is the converted audio file, and the output is the audio file stored on the server. The audio data is then ready for processing on the server.

[1507] Step 4:

[1508] The server preprocesses the uploaded audio data. Preprocessing includes noise reduction and volume normalization. The input is an audio file stored on the server, and the output is a preprocessed audio file. Preprocessing improves the quality of the audio data.

[1509] Step 5:

[1510] The server converts the preprocessed audio data into text data using speech recognition technology. The input is the preprocessed audio file, and the output is text data transcribed from the audio data. Models such as "Wav2Vec2" are used for speech recognition.

[1511] Step 6:

[1512] The server trains a machine learning model using voice and text data. The input is the voice data and corresponding text data, and the output is a machine learning model that has learned the user's speech characteristics. Generative AI models such as "GPT-3" are used for training.

[1513] Step 7:

[1514] After the death of the user, family and friends use the device to launch a conversation app. The input is the speech of the family and friends, and the output is the voice data sent to the server. The device records the visitor's speech and sends it to the server.

[1515] Step 8:

[1516] The server analyzes the visitor's voice data and converts it into text data. The input is the visitor's voice data, and the output is text data. The server generates a response in real time based on the converted text data.

[1517] Step 9:

[1518] The server generates a response in real time using a generative AI model. The input is the visitor's text data, and the output is the generated response text. An example of a prompt for the generative AI model is as follows:

[1519] plaintext

[1520] User: What products do you have recommended today?

[1521] Bot:

[1522] Step 10:

[1523] The terminal converts the response text received from the server into speech and plays it back to the visitor. The input is the generated response text and the output is the audio data. The terminal uses the TextToSpeech module to generate audio and plays it back to the visitor as a natural dialogue.

[1524] 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.

[1525] 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.

[1526] 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.

[1527] 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.

[1528] 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.

[1529] 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.

[1530] 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).

[1531] 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.

[1532] 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."

[1533] 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.

[1534] 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).

[1535] 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.

[1536] 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.

[1537] 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.

[1538] 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.

[1539] 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.

[1540] 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.

[1541] 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.

[1542] 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.

[1543] 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.

[1544] 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.

[1545] The following is further disclosed regarding the above embodiment.

[1546] (Claim 1)

[1547] means for recording user voice data;

[1548] means for uploading the recorded audio data to a server;

[1549] means for preprocessing the uploaded audio data and converting it into text data;

[1550] A means for performing machine learning based on voice data and text data to learn the user's speech characteristics;

[1551] After the user dies, the server generates a dialogue in real time and transmits it to the terminal;

[1552] The system includes means for the terminal to play back the generated dialogue as audio.

[1553] (Claim 2)

[1554] 10. The system of claim 1, further comprising means for pre-processing the audio data to perform noise removal and volume normalization.

[1555] (Claim 3)

[1556] 10. The system of claim 1, further comprising means for periodically updating the machine learning model.

[1557] "Example 1"

[1558] (Claim 1)

[1559] means for recording user voice data;

[1560] A means for converting the recorded audio data into an appropriate format on the device and uploading it to a server;

[1561] A means for preprocessing the uploaded audio data, performing noise reduction and volume normalization, and converting the audio data into text data;

[1562] A means for performing machine learning based on voice data and text data to learn the user's speech characteristics;

[1563] After the user dies, the server generates a dialogue in real time and transmits it to the terminal;

[1564] The system includes a means for converting the generated dialogue into speech and playing it back.

[1565] (Claim 2)

[1566] 10. The system of claim 1, further comprising means for periodically updating the machine learning model.

[1567] (Claim 3)

[1568] 10. The system of claim 1, further comprising means for recording the speech of the family member or friend and transmitting the recording to the server, which then analyzes the recording and generates an appropriate response.

[1569] "Application Example 1"

[1570] (Claim 1)

[1571] means for recording user voice data;

[1572] means for uploading the recorded audio data to a server;

[1573] means for preprocessing the uploaded audio data and converting it into text data;

[1574] A means for performing machine learning based on voice data and text data to learn the user's speech characteristics;

[1575] After the user dies, the server generates a dialogue in real time and transmits it to the terminal;

[1576] a means for the terminal to play the generated dialogue as audio to provide a customer dialogue experience;

[1577] A system including a means for providing a customer with an interactive experience with the deceased via a head-mounted display or smartphone.

[1578] (Claim 2)

[1579] 10. The system of claim 1, further comprising means for pre-processing the audio data to perform noise removal and volume normalization.

[1580] (Claim 3)

[1581] 10. The system of claim 1, further comprising means for periodically updating the machine learning model.

[1582] "Example 2: Combining Emotion Engines"

[1583] (Claim 1)

[1584] means for recording user voice data;

[1585] A means for converting the format of the recorded voice data by the terminal and uploading the data to a server;

[1586] means for preprocessing the uploaded audio data and converting it into text data;

[1587] A means for performing machine learning based on voice data and text data to learn the user's speech characteristics and emotional information;

[1588] A means for extracting emotional information from the speech data using an emotion engine and using the information as training data for a machine learning model;

[1589] After the user dies, the server generates a dialogue in real time and transmits it to the terminal;

[1590] The system includes means for the terminal to play back the generated dialogue as audio.

[1591] (Claim 2)

[1592] 10. The system of claim 1, further comprising means for pre-processing the audio data to perform noise removal and volume normalization.

[1593] (Claim 3)

[1594] 10. The system of claim 1, further comprising means for periodically updating the machine learning model.

[1595] "Application example 2 when combining emotion engines"

[1596] (Claim 1)

[1597] means for recording user voice data;

[1598] means for uploading the recorded audio data to a server;

[1599] means for preprocessing the uploaded audio data and converting it into text data;

[1600] A means for performing machine learning based on voice data and text data to learn the user's speech characteristics;

[1601] After the user dies, the server generates a dialogue in real time and transmits it to the terminal;

[1602] a means for the terminal to play back the generated dialogue as audio;

[1603] A system including a means for providing emotionally considerate customer service to visitors in a virtual store.

[1604] (Claim 2)

[1605] 10. The system of claim 1, further comprising means for pre-processing the audio data to perform noise removal and volume normalization.

[1606] (Claim 3)

[1607] 10. The system of claim 1, further comprising means for periodically updating the machine learning model. [Explanation of symbols]

[1608] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for recording user voice data; means for uploading the recorded audio data to a server; means for preprocessing the uploaded audio data and converting it into text data; A means for performing machine learning based on voice data and text data to learn the user's speech characteristics; After the user dies, the server generates a dialogue in real time and transmits it to the terminal; The system includes means for the terminal to play back the generated dialogue as audio.

2. 10. The system of claim 1, further comprising means for pre-processing the audio data to perform noise removal and volume normalization.

3. The system of claim 1 , further comprising means for periodically updating the machine learning model.

Images