System

Abstract

A system is provided.SOLUTION: A means for collecting communication data of a deceased person, a means for analyzing the communication data and performing preprocessing, a means for learning an AI model based on the preprocessed data, a means for integrating the learned AI model and generating an AI character similar to the deceased person, and a means for providing a digital portrait device displaying the AI character.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 modern society, an increasing number of bereaved families feel a sense of loss and loneliness due to separation from their deceased loved ones. This is due to the declining population and rising mortality rate. Traditional memorial portraits are static and do not allow for dialogue or interaction with the deceased, which is an issue and often results in insufficient psychological care for bereaved families. Therefore, new methods are needed to enable conversation and interaction with the deceased and provide psychological care for bereaved families. [Means for solving the problem]

[0005] The present invention provides a means for collecting and preprocessing communication data of the deceased. It also includes a means for training an AI model based on the preprocessed data and integrating a natural language processing (NLP) model, a voice synthesis model, and an image / video analysis model to generate an AI character that resembles the deceased. It also provides a system that includes a means for providing a digital memorial portrait device that displays this AI character, and allows users to interact with the deceased by receiving voice input from the user and generating and displaying a response using the AI ​​character. This method allows users to recreate a connection with the deceased and provide emotional care for the bereaved.

[0006] "Communication data" refers to data including chat history, call records, photos, and videos used by the deceased during their lifetime.

[0007] "Preprocessing" is the process of removing unnecessary information from collected communication data and extracting important information.

[0008] An "AI model" is an algorithm that learns from communication data and reproduces the deceased's speech, vocalizations, facial expressions, etc.

[0009] A "natural language processing (NLP) model" is a machine learning model that can analyze text data and understand its meaning.

[0010] A "voice synthesis model" is a model that is trained based on voice data and reproduces the voice of a specific person.

[0011] The "image and video analysis model" is a model that analyzes image and video data and reproduces facial expressions and gestures.

[0012] An "AI character" is a digital character that resembles a deceased person and is generated by an integrated AI model.

[0013] A "digital memorial device" is a device that displays an AI character, receives voice input from the user, and generates and displays a response. [Brief explanation of the drawings]

[0014] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0015] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

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

[0017] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0018] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0019] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0020] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0021] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0022] [First embodiment]

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

[0024] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0025] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0026] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0027] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0028] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0029] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

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

[0031] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0032] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0033] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0034] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0035] To implement this invention, we will build a system that uses the communication data of the deceased to generate an AI model and use it as a digital memorial portrait. The following explains the program processing of this system in natural language.

[0036] System Overview

[0037] The system collects communication data from the deceased, trains an AI model using preprocessed data, and displays the generated AI character as a digital portrait. Furthermore, users can talk to this digital portrait and it responds in a way that is true to the deceased's personality.

[0038] Program processing flow

[0039] 1. Data Collection Phase

[0040] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[0041] The terminal provides an interface for sending the exported data to the server.

[0042] 2. Data preprocessing phase

[0043] The server stores the received communication data and performs pre-processing on the data, which includes removing unnecessary information (e.g., spam messages, errors) and extracting important information (e.g., key phrases, important topics, and frequently occurring vocabulary).

[0044] 3. AI model training phase

[0045] The server uses the preprocessed data to train natural language processing (NLP) models, speech synthesis models, and image and video analysis models. Each model works as follows:

[0046] NLP model: Analyzes text data and learns the words and phrases of the deceased.

[0047] Speech synthesis model: Extracts clear speech and trains to reproduce the voice quality of the deceased.

[0048] Image and video analysis model: Learns facial expressions and gestures and reproduces the facial expressions and movements of the deceased.

[0049] 4. Model integration phase

[0050] The server then integrates the trained models to operate as a single AI character, a process that involves linking the outputs of each model and ensuring that the words, voices, and facial expressions match.

[0051] 5. Digital portrait generation phase

[0052] The server packages the integrated AI character and sends it to the user's device, where it can be displayed on a digital memorial device (such as a dedicated display or smartphone).

[0053] 6. Dialogue Phase

[0054] Users can talk to the digital portrait, asking questions such as "How was your day?" or "Can you tell me about an old memory?"

[0055] The device captures the user's voice with a microphone and transmits the voice data to the server.

[0056] The server analyzes the received voice using automatic speech recognition (ASR) technology and understands the content of the question.

[0057] The server uses an NLP model to generate appropriate responses and a speech synthesis model to generate responses in the voice of the deceased.

[0058] The device plays back the responses received from the server, reproducing facial expressions and gestures as needed.

[0059] Specific examples

[0060] For example, if a user speaks to a digital portrait and says, "Hello, Grandpa. How was your day?", the following happens:

[0061] 1. User: Speaks a question.

[0062] 2. Device: Captures audio and sends it to the server.

[0063] 3. Server: Analyzes the voice and understands the question.

[0064] 4. Server: Generates a response using an NLP model, such as "Hello, it was a beautiful day today. How was it for you?"

[0065] 5. Server: Generate a response in the voice of the deceased person using a speech synthesis model.

[0066] 6. Terminal: Plays back the response and reproduces the facial expression of the deceased.

[0067] In this way, the system allows the bereaved to reconnect with the deceased and provide emotional care.

[0068] The processing flow will be explained below.

[0069] Step 1:

[0070] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[0071] Step 2:

[0072] The device uploads the exported data to the server in the specified format (e.g., JSON or CSV file).

[0073] Step 3:

[0074] The server receives the uploaded communication data and stores it in a secure database.

[0075] Step 4:

[0076] The server analyzes the stored data, removes unnecessary information (such as spam messages and errors), and extracts important information (key phrases, important topics, and frequently occurring vocabulary).

[0077] Step 5:

[0078] The server uses the preprocessed data to train a natural language processing (NLP) model, which analyzes and trains the text data to memorize the words and phrases used by the deceased.

[0079] Step 6:

[0080] The server uses the audio data to train a speech synthesis model, extracting clear speech and training it to reproduce the voice quality of the deceased person.

[0081] Step 7:

[0082] The server uses image and video data to train an image and video analysis model, analyzing the facial expressions and gestures of the deceased.

[0083] Step 8:

[0084] The server integrates the trained NLP model, speech synthesis model, and image / video analysis model, allowing it to operate as a single AI character.

[0085] Step 9:

[0086] The server packages the integrated AI character and sends it to the user's device.

[0087] Step 10:

[0088] Users can talk to the AI ​​character, which is displayed as a digital portrait of the deceased, asking questions such as, "How was your day?" or "Tell me about an old trip?"

[0089] Step 11:

[0090] The device captures the user's voice with a microphone and transmits the voice data to the server.

[0091] Step 12:

[0092] The server analyzes the received voice using automatic speech recognition (ASR) technology and understands the content of the question.

[0093] Step 13:

[0094] The server uses NLP models to generate appropriate responses, such as "It was a beautiful day today. How was it for you?" in response to a user question.

[0095] Step 14:

[0096] The server uses a speech synthesis model to generate the generated response in the voice of the deceased person.

[0097] Step 15:

[0098] The device plays back the responses received from the server, reproducing facial expressions and gestures as needed.

[0099] Example 1

[0100] 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."

[0101] In order to maintain the connection between memories and emotions of the deceased, there is a need for a system that allows bereaved families to reconnect with and interact with the deceased. However, existing technologies are not sufficient to reproduce the voice, facial expressions, and language of the deceased, and there are issues with a lack of realism and the complexity of operation. This means that the current situation is one in which the emotional care of bereaved families is not being provided adequately.

[0102] 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.

[0103] In this invention, the server includes: means for collecting communication data of the deceased; means for analyzing and preprocessing the communication data; means for training an AI model based on the preprocessed data; means for training the AI ​​model using a natural language processing model, a voice synthesis model, and an image / video analysis model; means for integrating the trained models and linking the data so that the words, voice, and facial expressions match; means for packaging the integrated AI character and transmitting it to a user's device; means for providing a digital memorial portrait device that displays the AI ​​character; and means for receiving voice input from the user using the digital memorial portrait device and generating and displaying a response using the AI ​​character. This makes it possible to reproduce the appearance, voice, and speech of the deceased with high accuracy, allowing bereaved family members to naturally converse and interact with the deceased.

[0104] "Communication data" refers to information such as message history, call records, images, and videos left by the deceased.

[0105] "Preprocessing" refers to the process of removing unnecessary information from the received data and extracting key phrases, important topics, etc.

[0106] "AI model" refers to an artificial intelligence model that analyzes and learns from text data, audio data, image and video data.

[0107] "Natural language processing model" refers to an AI model used to understand and generate human language.

[0108] "Speech synthesis model" refers to an AI model used to convert text data into speech and reproduce specific speech characteristics.

[0109] "Image and video analysis model" refers to an AI model used to analyze image and video data and extract and reproduce specific facial expressions and gestures.

[0110] "Data integration" refers to the process of integrating the output of each trained AI model to generate consistent words, voices, and facial expressions.

[0111] "Packaging" refers to the process of converting the integrated AI character into a data format that can be used on the user's device.

[0112] "Digital memorial device" refers to a device that can display an AI character and generate and display responses based on voice input.

[0113] To implement this invention, it is necessary to build a system that collects communication data of the deceased, generates an AI model, and uses it as a digital memorial portrait. This system uses various hardware and software to accurately reproduce the deceased's speech, voice, and facial expressions. Specific examples are shown below.

[0114] System Overview

[0115] This system collects communication data from the deceased, preprocesses it, trains an AI model, and displays the generated AI character as a digital portrait. Furthermore, when the user speaks to this digital portrait, it can respond in a way that is true to the deceased's personality.

[0116] Hardware and software used

[0117] Hardware: User's smartphone or PC, cloud server, microphone, dedicated display.

[0118] Software: messaging applications, data transmission interface applications, cloud storage (e.g., Amazon S3), Python data processing libraries (e.g., spaCy, NLTK), speech analysis libraries (e.g., Librosa), natural language processing models (e.g., BERT, GPT-3®), speech synthesis models (e.g., Tacotron2, WaveNet), image and video analysis libraries (e.g., OpenCV, Dlib), machine learning frameworks (e.g., TENSORFLOW®, PyTorch).

[0119] Program processing flow

[0120] 1. Data collection phase:

[0121] Users log into messaging apps using their smartphones or PCs and export chat history, call logs, photos, and videos.

[0122] The device provides an interface for sending the exported data to the cloud server, for example, by clicking a "Data Upload" button in a dedicated application.

[0123] 2. Data preprocessing phase:

[0124] The server analyzes the communication data stored in the cloud storage, first filtering out spam messages and unwanted data using a Python script.

[0125] Next, we use an NLP library (e.g., spaCy) to extract key phrases and important topics, and then use topic modeling techniques (e.g., LDA) to cluster frequently occurring vocabulary.

[0126] 3. AI model training phase:

[0127] The server inputs the preprocessed text data into the BERT model, allowing it to learn the unique phrases and vocabulary of the deceased.

[0128] For the audio data, we use the Librosa library to remove noise and extract clear speech, and then use a voice synthesis model (e.g., Tacotron2) to recreate the voice quality of the deceased.

[0129] Image and video data is analyzed using OpenCV and Dlib, and the system is trained to recognize the facial expressions and gestures of the deceased.

[0130] 4. Model integration phase:

[0131] The server integrates each trained model (natural language processing model, speech synthesis model, image and video analysis model) and uses machine learning frameworks such as TensorFlow to coordinate the data so that the integrated AI character can generate consistent words, voices, and facial expressions.

[0132] 5. Digital portrait generation phase:

[0133] The server then sends the integrated AI character to the user's device, where the user can view the digital portrait using a dedicated display or smartphone.

[0134] 6. Dialogue Phase:

[0135] The user talks to the digital portrait, asking questions such as, "Hello, Grandpa. How was your day?"

[0136] The terminal captures the user's voice with a microphone and transmits the voice data to the server in real time.

[0137] The server converts the received voice into text using voice recognition technology (e.g., Google (registered trademark) Speech-to-Text API) and understands the content of the question.

[0138] Next, an NLP model is used to generate appropriate responses, and a speech synthesis model generates the responses in the voice of the deceased.

[0139] The device plays back the response received from the server and also displays animations that reproduce the facial expressions and gestures of the deceased.

[0140] Specific examples

[0141] For example, if a user says, "Hello, Grandpa. How was your day?", the process is as follows:

[0142] 1. User: Speaks a question.

[0143] 2. Device: Captures audio and sends it to the server.

[0144] 3. Server: Analyzes the voice and understands the question.

[0145] 4. Server: Generates a response using an NLP model, for example, "Hello, it was a beautiful day today. How was it for you?"

[0146] 5. Server: Generate a response in the voice of the deceased person using a speech synthesis model.

[0147] 6. Terminal: Plays back the response and reproduces the facial expression of the deceased.

[0148] Prompt Sentence Examples

[0149] "Tell me some recent memories of your grandpa."

[0150] "Tell me about your grandpa's hobbies."

[0151] "Where was your last trip?"

[0152] This allows the user to provide emotional care through intimate conversations with the deceased.

[0153] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0154] Step 1:

[0155] Input: User accesses deceased person's communication accounts and exports chat history, call logs, photos, and videos.

[0156] How it works: A user logs into a messaging app using a smartphone or PC, accesses the "Settings" menu, selects the "Export Chat History" option, and exports the relevant data.

[0157] Output: The exported data file.

[0158] Step 2:

[0159] Input: Terminal exported data file.

[0160] Specific operation: The terminal provides an interface for the user to send the exported data to the server. For example, the user clicks the "Upload Data" button in the dedicated application to send the data to the server.

[0161] Output: The data file sent to the server.

[0162] Step 3:

[0163] Input: The server receives the transmitted data file.

[0164] Specific operation: The server stores the received data in cloud storage (e.g., Amazon S3).

[0165] Output: Data stored in cloud storage.

[0166] Step 4:

[0167] Input: Data stored in cloud storage.

[0168] What it does: The server uses Python scripts to filter spam messages and error messages from the stored data, then uses NLP libraries (e.g., spaCy) to extract key phrases and important topics.

[0169] Output: Preprocessed text data.

[0170] Step 5:

[0171] Input: Preprocessed text data.

[0172] How it works: The server inputs the preprocessed text data into the BERT model, allowing it to learn the unique phrases and vocabulary of the deceased.

[0173] Output: A trained natural language processing model.

[0174] Step 6:

[0175] Input: Audio data stored in cloud storage.

[0176] How it works: The server uses the Librosa library to analyze the audio data, remove noise, and extract clear audio.

[0177] Output: Preprocessed audio data.

[0178] Step 7:

[0179] Input: Preprocessed audio data.

[0180] Specific operation: The server uses the Tacotron2 model to train the preprocessed audio data to reproduce the voice quality of the deceased.

[0181] Output: The trained speech synthesis model.

[0182] Step 8:

[0183] Input: Image and video data stored in cloud storage.

[0184] Specific operation: The server uses OpenCV and Dlib to analyze the facial expressions and gestures of the deceased in image and video data.

[0185] Output: Preprocessed image and video data.

[0186] Step 9:

[0187] Input: Preprocessed image and video data.

[0188] Specific operation: The server uses image and video analysis models to learn how to reproduce the facial expressions and movements of the deceased.

[0189] Output: A trained image and video analysis model.

[0190] Step 10:

[0191] Input: Trained natural language processing models, speech synthesis models, and image / video analysis models.

[0192] How it works: The server uses machine learning frameworks such as TensorFlow to integrate each model and generate a single AI character, coordinating verbal, vocal, and facial output to achieve results that are closer to real-life interactions.

[0193] Output: A unified AI character.

[0194] Step 11:

[0195] Input: Integrated AI character.

[0196] Specific operation: The server packages this AI character and sends it to the user's device through a dedicated API.

[0197] Output: The AI ​​character sent to the user's device.

[0198] Step 12:

[0199] Input: A question that the user speaks.

[0200] Specific Actions: The user talks to the digital portrait, asking questions such as, "Hello, Grandpa. How was your day?"

[0201] Output: The audio data captured on the device.

[0202] Step 13:

[0203] Input: Audio data captured on the device.

[0204] How it works: The device sends the captured voice data to the server, which uses the Google Speech-to-Text API to convert the voice data into text and understand the question.

[0205] Output: Parsed text data.

[0206] Step 14:

[0207] Input: Parsed text data.

[0208] What happens: The server uses the NLP model to generate an appropriate response, such as "Hello, it was a beautiful day today. How was it for you?"

[0209] Output: The generated text response.

[0210] Step 15:

[0211] Input: The generated text response.

[0212] What it does: The server uses the Tacotron2 model to vocalize the text response in the voice of the deceased.

[0213] Output: A spoken response.

[0214] Step 16:

[0215] Input: A spoken response.

[0216] What it does: The device plays back spoken responses and also displays animations that replicate the facial expressions and gestures of the deceased.

[0217] Output: Played voice response and animation.

[0218] (Application example 1)

[0219] 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."

[0220] In recent years, with the advancement of AI technology, there has been an increase in efforts to generate digital memorial portraits using the communication data of the deceased, but conventional technologies have limited interaction capabilities with users and lack security.This invention provides a system that provides real-time response guidance from an AI character that resembles the deceased, and detects and warns abnormal behavior using a score prediction model, thereby achieving a safe and comfortable interaction experience for users.

[0221] 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.

[0222] In this invention, the server includes means for collecting communication data of the deceased, means for analyzing and preprocessing the communication data, means for training an AI model based on the preprocessed data, means for integrating the trained AI model to generate an AI character resembling the deceased, means for providing a digital memorial portrait device that displays the AI ​​character, means for receiving voice input from a user using the digital memorial portrait device and generating and displaying a response using the AI ​​character, and means for detecting abnormal behavior using a score prediction model and providing a real-time audio warning. This allows for a natural conversational experience with the deceased while simultaneously alerting security guards to any abnormal behavior.

[0223] "Communication data of the deceased" refers to a collection of electronic data used by the deceased during their lifetime, and primarily includes chat history, call records, photos, and videos.

[0224] "Preprocessing" refers to the initial stage of data processing, where collected raw data is analyzed, unnecessary information is removed, and important information is extracted.

[0225] An "AI model" is an artificial intelligence program that has been trained to perform a specific task using machine learning or deep learning techniques.

[0226] An "AI character" is a digital character that integrates a trained AI model and reproduces the characteristics of a specific person (in this case, a deceased person).

[0227] A "digital memorial device" is a device that displays an AI character and enables interaction with the user, and can be a smartphone, a dedicated display, or AR-compatible smart glasses.

[0228] "Voice input from the user" refers to voice instructions or questions given by the user to the system via a microphone or the like.

[0229] A "score prediction model" is a model that uses AI technology to analyze data in real time and predict risk scores associated with specific behaviors or situations.

[0230] "Abnormal behavior detection" refers to the process of recognizing and alerting when behavior that deviates from normal patterns occurs.

[0231] "Providing a real-time warning" refers to the act of immediately issuing a warning and notifying the user of detected abnormal behavior.

[0232] This invention provides a system that collects communication data of the deceased, uses it to build an AI model, and uses it to create a digital memorial portrait. The system aims to generate an AI character that resembles the deceased and provides responses that are characteristic of the deceased through conversations with the user. It also has the ability to detect abnormal behavior and provide real-time warnings using a score prediction model.

[0233] System Configuration

[0234] 1. Hardware Configuration

[0235] Server: Responsible for storing and processing data, and training and operating AI models. A server with high-performance processing capabilities is desirable.

[0236] Device: The device used by the user, which may be a smartphone, a dedicated display, or AR-enabled smart glasses.

[0237] Camera: Used to capture footage in real time and analyze it in the score prediction model.

[0238] Microphone: Used to capture the user's voice input.

[0239] 2. Software Configuration

[0240] Data collection module: Collects communication data such as chat history, call records, photos, and videos of the deceased.

[0241] Data preprocessing module: Analyzes the collected data and extracts the necessary information.

[0242] AI model learning module: Based on the preprocessed data, natural language processing (NLP) models, speech synthesis models, image and video analysis models, and score prediction models are trained.

[0243] Integration module: Integrates each model to generate an AI character that resembles the deceased.

[0244] Digital portrait display module: Displays the generated AI character and enables interaction with the user.

[0245] Score prediction module: Detects abnormal behavior in real time based on camera footage and provides necessary warnings via voice.

[0246] Operating Procedure

[0247] 1. Data Collection: Users enter the deceased person's communication data into the system, including chat history, call records, photos, videos, etc.

[0248] 2. Data Preprocessing: The server preprocesses the received data, removing unnecessary information and extracting important information. This process includes filtering spam messages and extracting key phrases.

[0249] 3. AI model training: The server uses the preprocessed data to train the NLP model, speech synthesis model, image and video analysis model, and score prediction model, thereby acquiring knowledge to reproduce the deceased's unique speech patterns, voice quality, facial expressions, etc.

[0250] 4. Model integration: The output of each model is integrated to generate an AI character that resembles the deceased person, with matching language, voice, and facial expressions.

[0251] 5. Interactive function: The user can talk to the digital portrait, for example, asking questions like, "How was your day?" The device captures this voice and sends it to the server.

[0252] 6. Real-time warning: Camera footage is analyzed in real time, and if abnormal behavior is detected, a warning is immediately given to the user, such as an audio warning saying "Warning! Intruder detected."

[0253] Specific examples

[0254] When a user speaks to a digital portrait and asks, "Hello, how was your day?", the response is generated through the following process:

[0255] 1. User: Speaks a question.

[0256] 2. Device: Captures audio and sends it to the server.

[0257] 3. Server: Analyzes the voice and understands the question.

[0258] 4. Server: Generates a response using an NLP model, such as "It was a beautiful day today. How was it for you?"

[0259] 5. Server: Generates a response in the voice of the deceased person using a speech synthesis model and sends it to the device.

[0260] 6. Terminal: Plays back the response and reproduces the facial expression of the deceased.

[0261] Prompt Sentence Examples

[0262] Example prompt sentence:

[0263] "Warning! Unauthorized person loitering near conference room. Please investigate immediately."

[0264] "Attention! Abnormal behavior detected. Please respond immediately."

[0265] This allows users to have a natural conversation with the deceased while also supporting immediate response as a security guard.

[0266] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0267] Step 1:

[0268] Data collection

[0269] The user inputs the deceased's communication data (chat history, call records, photos, videos, etc.) into the system. The device exports this data and sends it to the server. The server receives and stores the data.

[0270] Input: Chat history, call logs, photos, videos

[0271] Output: Saved communication data

[0272] Step 2:

[0273] Data Preprocessing

[0274] The server analyzes the received communication data, removes unnecessary information (spam messages and errors), and extracts important information (key phrases, frequently occurring vocabulary, and important topics). This preprocessing process prepares the data for AI model training.

[0275] Input: Saved communication data

[0276] Output: Preprocessed data

[0277] Step 3:

[0278] Training an AI model

[0279] The server uses the preprocessed data to train various AI models (NLP model, voice synthesis model, image / video analysis model, score prediction model), which then learns the deceased's speech patterns, voice quality, facial expressions, and behavioral patterns.

[0280] Input: Preprocessed data

[0281] Output: Trained AI model (NLP model, speech synthesis model, image / video analysis model, score prediction model)

[0282] Step 4:

[0283] Model Integration

[0284] The server then integrates each trained AI model to generate an AI character that resembles the deceased. The integration process involves linking the outputs of each model and adjusting the words, voice, and facial expressions to match.

[0285] Input: Trained AI model

[0286] Output: Integrated AI character

[0287] Step 5:

[0288] Displaying a digital portrait of the deceased

[0289] The server packages the integrated AI character and sends it to the terminal, which uses the received data to display the AI ​​character on the digital portrait device.

[0290] Input: Integrated AI character

[0291] Output: Displayed AI character

[0292] Step 6:

[0293] User interaction

[0294] The user speaks to the digital portrait. The device captures the voice with a microphone and sends the captured voice data to the server. The server uses speech recognition technology to analyze and understand the question. It uses an NLP model to generate an appropriate response and a speech synthesis model to generate a response in the deceased's voice. Finally, the response is sent to the device, which plays it back. In some cases, the deceased's facial expressions and gestures are also reproduced.

[0295] Input: User's voice

[0296] Output: Vocal responses, facial expressions, and gestures of the deceased

[0297] Step 7:

[0298] Abnormal behavior detection and real-time alerts

[0299] The device's camera captures video in real time. The server analyzes the video using a score prediction model, and if abnormal behavior is detected, it immediately provides an audio warning. For example, it notifies the user with a voice message saying, "Warning! Intruder detected."

[0300] Input: Real-time video

[0301] Output: Audio warning, abnormal behavior detection results

[0302] 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.

[0303] To implement this invention, we will create an AI model using the communication data of the deceased and build a system that can be used as a digital memorial portrait. This system will also incorporate an emotion engine that recognizes the user's emotions and provides responses according to their emotions.

[0304] System Overview

[0305] The system collects communication data from the deceased, trains an AI model using preprocessed data, and displays the generated AI character as a digital portrait. Furthermore, when a user speaks to the digital portrait, it uses an emotion engine to analyze the emotions and provide appropriate responses.

[0306] Program processing flow

[0307] 1. Data Collection Phase

[0308] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[0309] The device uploads the exported data to the server in the specified format (e.g., JSON or CSV file).

[0310] 2. Data preprocessing phase

[0311] The server stores the received communication data and performs pre-processing on the data, which includes removing unnecessary information (such as spam messages and errors) and extracting important information (key phrases, important topics, and frequently occurring vocabulary).

[0312] 3. AI model training phase

[0313] The server uses the preprocessed data to train natural language processing (NLP) models, speech synthesis models, and image and video analysis models. Each model works as follows:

[0314] NLP model: Analyzes text data and learns the words and phrases of the deceased.

[0315] Speech synthesis model: Extracts clear speech and trains to reproduce the voice quality of the deceased.

[0316] Image and video analysis model: Learns facial expressions and gestures and reproduces the facial expressions and movements of the deceased.

[0317] 4. Model integration phase

[0318] The server then integrates the trained models to operate as a single AI character, a process that involves linking the outputs of each model and ensuring that the words, voices, and facial expressions match.

[0319] 5. Digital portrait generation phase

[0320] The server packages the integrated AI character and sends it to the user's device, where it can be displayed on a digital memorial device (such as a dedicated display or smartphone).

[0321] 6. Emotion recognition and response phase

[0322] The emotion engine analyzes the user's voice data and facial expression data to recognize the user's emotions (e.g., happiness, sadness, anger, surprise, etc.).

[0323] The server adapts the NLP model and speech synthesis model based on the output of the emotion engine, allowing the AI ​​character of the deceased person to generate an appropriate response based on the recognized emotion.

[0324] The server also adjusts facial expressions and tone of voice according to emotions to recreate natural conversations.

[0325] Specific examples

[0326] For example, if a user says to a digital portrait, "Hello, Grandpa. I've been feeling a bit sad today," the following happens:

[0327] 1. User: Speaks a question.

[0328] 2. Device: Captures audio and sends it to the server.

[0329] 3. Server: Analyzes the voice and understands the question.

[0330] 4. Emotion engine: Analyzes the user's voice and facial expressions to recognize when the user is feeling "sad."

[0331] 5. Server: Uses NLP models to generate responses based on emotions, such as "Hi, I'm very worried to hear that you're feeling sad. What happened?"

[0332] 6. Server: Using a speech synthesis model, the generated response is generated in the voice of the deceased.

[0333] 7. Server: Based on the emotion engine, the tone of voice and facial expressions are adjusted to interact with the user in a more natural way.

[0334] 8. The device plays back the response received from the server and reproduces the facial expression of the deceased.

[0335] In this way, the system recognizes the user's emotions and allows them to provide emotional care through dialogue with the deceased.

[0336] The processing flow will be explained below.

[0337] Step 1:

[0338] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[0339] Step 2:

[0340] The device uploads the exported data to the server in the specified format (e.g., JSON or CSV file).

[0341] Step 3:

[0342] The server receives the uploaded communication data and stores it in a secure database.

[0343] Step 4:

[0344] The server analyzes the stored data, removes unnecessary information (such as spam messages and errors), and extracts important information (key phrases, important topics, and frequently occurring vocabulary).

[0345] Step 5:

[0346] The server uses the preprocessed data to train a natural language processing (NLP) model, which analyzes and trains the text data to memorize the words and phrases used by the deceased.

[0347] Step 6:

[0348] The server uses the audio data to train a speech synthesis model, extracting clear speech and training it to reproduce the voice quality of the deceased person.

[0349] Step 7:

[0350] The server uses image and video data to train an image and video analysis model, analyzing the facial expressions and gestures of the deceased.

[0351] Step 8:

[0352] The server integrates the trained NLP model, speech synthesis model, and image / video analysis model, allowing it to operate as a single AI character.

[0353] Step 9:

[0354] The server packages the integrated AI character and sends it to the user's device.

[0355] Step 10:

[0356] Users can talk to the AI ​​character, which is displayed as a digital portrait of the deceased, asking questions such as, "How was your day?" or "Tell me about an old trip?"

[0357] Step 11:

[0358] The device captures the user's voice with a microphone and transmits the voice data to the server.

[0359] Step 12:

[0360] The server analyzes the received voice using automatic speech recognition (ASR) technology and understands the content of the question.

[0361] Step 13:

[0362] The server uses an emotion engine to analyze the user's voice data and facial expression data and recognize the user's emotions (e.g., happiness, sadness, anger, surprise, etc.).

[0363] Step 14:

[0364] The server uses NLP models to generate appropriate responses based on the recognized emotions, for example, if the user is expressing sadness, it generates a response such as "Hi, I'm very worried to hear that you're feeling sad. What happened?"

[0365] Step 15:

[0366] The server uses a speech synthesis model to generate the generated response in the voice of the deceased person, adjusting the tone of the voice based on the output of the emotion engine.

[0367] Step 16:

[0368] The server sends the generated response to the device, adjusting facial expressions and gestures as needed.

[0369] Step 17:

[0370] The device then plays back the responses received from the server and reproduces the facial expressions of the deceased, allowing the user to have a natural conversation.

[0371] Example 2

[0372] 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."

[0373] It is necessary to utilize the communication data of the deceased to create a digital memorial portrait that reflects the deceased's characteristics, and to realize a dialogue between the user and the deceased's AI character based on emotion recognition. However, conventional technologies have low accuracy in analyzing communication data and recognizing emotions, making it difficult to provide natural responses that correspond to the user's emotions. Another issue is the difficulty of integrating various models (e.g., natural language processing, speech synthesis, image and video analysis) to accurately reproduce the characteristics of the deceased.

[0374] 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.

[0375] In this invention, the server includes means for collecting communication data of the deceased, means for analyzing and preprocessing the communication data, means for training an AI model based on the preprocessed data, means for integrating the trained AI model to generate an AI character resembling the deceased, means for providing a display device for displaying the AI ​​character, means for receiving voice input from a user via the display device and generating and displaying a response using the AI ​​character, and means for recognizing the user's emotions and generating a response according to the emotion. This makes it possible to accurately reproduce the characteristics of the deceased while providing a natural and appropriate response according to the user's emotions.

[0376] "Communication data" refers to records of digital communications used by the deceased during their lifetime, such as chat history, call records, photos, and videos.

[0377] "Preprocessing" is the process of removing unnecessary information from the received data and extracting important information (such as key phrases and frequently occurring vocabulary).

[0378] An "AI model" is a collection of algorithms that use artificial intelligence techniques to analyze and learn from data and automate specific tasks.

[0379] A "natural language processing model" is a model that uses artificial intelligence technology to analyze human language and understand its meaning.

[0380] A "speech synthesis model" is a model that uses artificial intelligence technology to convert text data into speech.

[0381] The "image and video analysis model" is a model that uses artificial intelligence technology to analyze image and video data and understand its content.

[0382] An "AI character" is a digital character generated based on an AI model that reproduces the features of a deceased person.

[0383] A "display device" is a device for visually displaying an AI character to a user.

[0384] "Emotion recognition" is a technology that analyzes a user's voice data and facial expression data to identify their emotions.

[0385] To implement this invention, it is necessary to build a system that uses the communication data of the deceased to generate an AI model and use it as a digital memorial portrait. This system is equipped with an emotion engine that recognizes the user's emotions and provides responses according to the emotions.

[0386] Hardware and software used

[0387] Hardware

[0388] Server: A server for storing and analyzing data and training AI models. Use a cloud server (e.g., Amazon EC2) equipped with a high-performance processor and large memory capacity.

[0389] Terminal: A device that allows users to operate and input data. This includes smartphones, tablets, and PCs.

[0390] Display device: A device for displaying the digital portrait. A dedicated display or smartphone can be used.

[0391] software

[0392] Natural Language Processing (NLP) model: A model for analyzing text data and learning the vocabulary and phrasing of the deceased, using NLP techniques such as BERT (Bidirectional Encoder Representations from Transformers).

[0393] Speech synthesis model: A model for converting text to speech and recreating the voice quality of the deceased. Uses voice synthesis technology such as Tacotron 2.

[0394] Image and video analysis model: A model that analyzes facial expressions and gestures to recreate the movements and expressions of the deceased. It uses image and video analysis technologies such as OpenPose and Face++.

[0395] Emotion engine: A technology that analyzes the user's voice data and facial expression data to recognize the user's emotions. It uses a combination of DeepSpeech (voice recognition) and Face++ (facial expression recognition).

[0396] Specific examples

[0397] Data Collection Phase

[0398] The user exports the chat history from the deceased person's messaging app (e.g., WhatsApp), and then uploads it to the server using a dedicated app.

[0399] Data Preprocessing Phase

[0400] The server analyzes the received chat history, filters out unnecessary information, and then extracts important key phrases and frequently occurring vocabulary.

[0401] AI model training phase

[0402] The server uses the preprocessed data to train a BERT model to learn the deceased's words and phrases, then trains the model with Tacotron 2 on clear audio data to reproduce the deceased's voice, and trains the model with OpenPose and Face++ on facial expressions and gestures.

[0403] Model integration phase

[0404] The server integrates each model (natural language processing model, speech synthesis model, image / video analysis model) to generate an AI character that recreates the deceased person.

[0405] Usage example

[0406] For example, if a user says to a digital portrait, "Hello, Grandpa. I've been feeling a bit sad today," the following happens:

[0407] The user enters the question by voice.

[0408] The device captures the audio and sends it to the server.

[0409] The server analyzes the voice and understands the question.

[0410] The emotion engine analyzes the user's voice and facial expressions and recognizes that the user is feeling "sad."

[0411] The server uses an NLP model to generate a response such as, "Hello, I'm very worried to hear that you're feeling sad. What happened?"

[0412] The server uses a speech synthesis model to generate responses in the voice of the deceased, adjusting the tone of voice and facial expression depending on the emotion.

[0413] The device plays back the response received from the server and reproduces the facial expression of the deceased.

[0414] Prompt Sentence Examples

[0415] "Analyze the message data of the deceased and generate an AI model for natural conversation."

[0416] "Develop a system that recognizes and responds to user emotions. The system will use an emotion engine to engage in appropriate dialogue based on the user's emotions."

[0417] This allows the system to recognize the user's emotions and provide emotional care through dialogue with the deceased.

[0418] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0419] Step 1: Data collection

[0420] The user accesses the deceased person's messaging and social media accounts and exports chat history, call logs, photos, and videos (input: the deceased person's communication data).

[0421] The terminal uploads the exported data to the server in the specified format (e.g., JSON file or CSV file) (output: uploaded communication data).

[0422] Step 2: Data storage and preprocessing

[0423] The server receives the data sent from the terminal and stores it securely (input: uploaded communication data, output: stored communication data).

[0424] The server removes unnecessary information from the received data (specific operations: filtering spam messages and error logs) and extracts important key phrases and frequently occurring vocabulary (input: saved communication data, output: preprocessed data).

[0425] Step 3: Training the Natural Language Processing (NLP) Model

[0426] The server uses the preprocessed text data to train an NLP model (e.g., BERT) (input: preprocessed data, output: trained NLP model). Specifically, it tokenizes the raw data and trains the model to understand the structure of the text.

[0427] Step 4: Training the speech synthesis model

[0428] The server extracts the preprocessed voice data and trains it using a speech synthesis model (e.g., Tacotron 2) (input: preprocessed voice data, output: trained speech synthesis model). Specifically, it analyzes the features of the voice data and reproduces the voice quality of the deceased.

[0429] Step 5: Training the image and video analysis model

[0430] The server uses the preprocessed image and video data to train an image and video analysis model (e.g., OpenPose, Face++) (input: preprocessed image and video data, output: trained image and video analysis model). Specifically, it analyzes the facial expressions and gestures of the deceased and reproduces them.

[0431] Step 6: Model integration

[0432] The server integrates each trained model to generate a single AI character (input: trained NLP model, trained speech synthesis model, trained image and video analysis model, output: integrated AI character). Specifically, the text generated by the NLP model is input into the speech synthesis model, and the corresponding facial expression is generated by the image and video analysis model.

[0433] Step 7: Generate a digital portrait

[0434] The server packages the integrated AI character and sends it to the user's device (input: integrated AI character, output: package sent to the user's device).

[0435] The user installs the file sent to their device and displays the AI ​​character on the digital portrait device (input: sent package, output: displayed AI character).

[0436] Step 8: Emotion Recognition and Response Generation

[0437] The emotion engine analyzes the user's voice data and facial expression data and recognizes emotions (input: user's voice data, facial expression data, output: recognized emotions).

[0438] Based on the output of the emotion engine, the server adapts the NLP model and speech synthesis model to generate an appropriate response (input: recognized emotion, output: generated response).

[0439] The server generates the response in the voice of the deceased person and adjusts the tone of voice and facial expression according to the emotion (input: generated response, output: adjusted response and facial expression).

[0440] The device plays back the responses received from the server and reproduces the facial expressions of the deceased (input: adjusted responses and facial expressions, output: reproduced responses and facial expressions).

[0441] This series of processes allows the user to receive emotional care through natural conversation with the deceased.

[0442] (Application example 2)

[0443] 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."

[0444] When applying a digital memorial portrait system to a virtual store, a means of understanding the user's emotions and responding appropriately is required. Another challenge is to increase customer satisfaction by providing a customer service experience that recreates the memories and voice of the deceased.

[0445] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0446] In this invention, the server includes means for collecting communication data of the deceased, means for analyzing and preprocessing the communication data, means for training an AI model based on the preprocessed data, means for integrating the trained AI model to generate an AI character resembling the deceased, means for providing a digital portrait device that displays the AI ​​character, means for receiving voice input from a user using the digital portrait device and generating and displaying a response using the AI ​​character, means for analyzing the user's emotions using an emotion recognition engine, means for generating a response based on the user's emotions, and means for using the digital portrait device as a guide in a virtual store. This makes it possible to grasp the user's emotions and provide effective guidance in the virtual store by reproducing the deceased's responses according to those emotions.

[0447] "Communication data" refers to digital information left behind by the deceased, such as chat history, call records, photos, and videos.

[0448] "Preprocessing" refers to the process of removing unnecessary information from collected communication data and extracting important information.

[0449] An "AI model" is an artificial intelligence that is trained using technologies such as natural language processing, speech synthesis, and image and video analysis to carry out specific tasks.

[0450] An "AI character" is a digital character that resembles a deceased person and is generated by integrating a trained AI model.

[0451] A "digital memorial device" is a device that displays an AI character and allows the user to interact with it.

[0452] "Voice input" refers to the voice data that is generated when the user speaks to the digital memorial portrait device.

[0453] An "emotion recognition engine" is a system for analyzing emotions from a user's voice and images.

[0454] A "virtual store" is a store that operates in a virtual space, a place that offers products and services online.

[0455] A "guide" is a person whose role is to provide customers with product information and guidance within the store.

[0456] A "user" is a person who uses the digital memorial device to interact with an AI character of a deceased person.

[0457] To implement this invention, it is necessary to create an AI model using the communication data of the deceased and build a system that functions as a digital memorial guide in a virtual store. This system reproduces the memories and voice of the deceased and combines it with an emotion recognition engine to provide appropriate responses according to the user's emotions.

[0458] System Configuration

[0459] 1. The user uses a smartphone or smart glasses to talk to a digital portrait guide.

[0460] 2. The device captures the user's voice input and facial image data and sends them to the server.

[0461] 3. The server executes multiple methods to perform the following processes:

[0462] Data collection and preprocessing

[0463] The server collects the deceased's communication data (chat history, call records, photos, videos). The collected data is pre-processed using technologies such as NLP (natural language processing) and image and video analysis to remove unnecessary information and extract important information.

[0464] AI model training and integration

[0465] The server uses the preprocessed data to train natural language processing models, speech synthesis models, and image and video analysis models. These trained models are then integrated to generate an AI character that resembles the deceased. This AI character reproduces the deceased's speech, voice quality, facial expressions, etc., allowing users to interact with the deceased in a virtual store.

[0466] Emotion Recognition and Response Generation

[0467] The server is equipped with an emotion recognition engine that analyzes the user's emotions from their voice and facial expression data. Based on the emotion engine's analysis results, an NLP model generates an appropriate text response, and a speech synthesis model reproduces that text as the deceased's voice. This enables natural conversations that correspond to the user's emotions.

[0468] Use as a digital memorial guide

[0469] In the virtual store, users can interact with a digital memorial guide through their smartphones or smart glasses. For example, if a customer asks, "I'd like to know more about our new products," the server will analyze their voice and facial expression data and provide an answer. Specifically, it will respond with something like, "Of course. Here are our latest products, and their features are as follows."

[0470] Prompt Sentence Examples

[0471] An example prompt for generating an AI model is:

[0472] You are a digital memorial guide working in a virtual store. Respond to customers' questions and interests in a friendly and courteous manner. Strive to provide relevant and useful information while taking into consideration the customer's feelings. For example, if a customer asks, "I'd like to know more about our new products," you can respond, "Of course! Here are our latest products and their features," and then provide specific product information.

[0473] As described above, the present invention makes it possible to utilize a digital portrait of the deceased as a guide in a virtual store while being sensitive to the user's emotions.

[0474] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0475] Step 1:

[0476] The user speaks to the digital portrait guide through a smartphone or smart glasses. The user inputs questions by voice, and simultaneously captures facial image data. This input data is captured by the device, which then transmits the voice data and facial image data to a server.

[0477] Step 2:

[0478] The server analyzes the received voice data. Specifically, it converts it into text data using voice recognition technology (such as Google Speech API). At this time, voice data is input, and as a result, text data containing the user's question is output.

[0479] Step 3:

[0480] The server sends the received facial image data to an emotion recognition engine to analyze the user's emotions. The emotion recognition engine uses image analysis technology (such as OpenCV or Dlib) to analyze emotions from facial expressions. The input is facial image data, and the output is data indicating the user's emotions (e.g., happiness, sadness, anger, etc.).

[0481] Step 4:

[0482] The server uses the analyzed text data and emotion data to work with an NLP model to generate an appropriate text response. Specifically, NLP techniques are applied to the text data to generate candidate responses. The emotion data is used to adjust the response. The input is text data containing the question and emotion data, and the output is the generated appropriate response text.

[0483] Step 5:

[0484] The server sends the generated response text to a speech synthesis model, which generates audio data in the voice of the deceased. A speech synthesis model (such as Amazon Polly or Google Text-to-Speech) is used. The input is the response text, and the output is audio data that reproduces the voice of the deceased.

[0485] Step 6:

[0486] The server sends the generated voice data to the terminal, which then plays the received voice data to the user, allowing the user to interact with the digital portrait guide. The input is the voice data, and the output is the played voice response.

[0487] Step 7:

[0488] After the server or terminal has completed responding to the user's question, it waits for the next question or interaction. If the user has any new input, it returns to step 1.

[0489] Through the above processing steps, the user can have natural conversations with the digital portrait guide and is provided with guidance services in the virtual store.

[0490] 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.

[0491] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0492] 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.

[0493] [Second embodiment]

[0494] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0495] 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.

[0496] 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).

[0497] 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.

[0498] 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.

[0499] 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).

[0500] 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.

[0501] 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.

[0502] 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.

[0503] 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.

[0504] 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.

[0505] 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."

[0506] To implement this invention, we will build a system that uses the communication data of the deceased to generate an AI model and use it as a digital memorial portrait. The following explains the program processing of this system in natural language.

[0507] System Overview

[0508] The system collects communication data from the deceased, trains an AI model using preprocessed data, and displays the generated AI character as a digital portrait. Furthermore, users can talk to this digital portrait and it responds in a way that is true to the deceased's personality.

[0509] Program processing flow

[0510] 1. Data Collection Phase

[0511] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[0512] The terminal provides an interface for sending the exported data to the server.

[0513] 2. Data preprocessing phase

[0514] The server stores the received communication data and performs pre-processing on the data, which includes removing unnecessary information (e.g., spam messages, errors) and extracting important information (e.g., key phrases, important topics, and frequently occurring vocabulary).

[0515] 3. AI model training phase

[0516] The server uses the preprocessed data to train natural language processing (NLP) models, speech synthesis models, and image and video analysis models. Each model works as follows:

[0517] NLP model: Analyzes text data and learns the words and phrases of the deceased.

[0518] Speech synthesis model: Extracts clear speech and trains to reproduce the voice quality of the deceased.

[0519] Image and video analysis model: Learns facial expressions and gestures and reproduces the facial expressions and movements of the deceased.

[0520] 4. Model integration phase

[0521] The server then integrates the trained models to operate as a single AI character, a process that involves linking the outputs of each model and ensuring that the words, voices, and facial expressions match.

[0522] 5. Digital portrait generation phase

[0523] The server packages the integrated AI character and sends it to the user's device, where it can be displayed on a digital memorial device (such as a dedicated display or smartphone).

[0524] 6. Dialogue Phase

[0525] Users can talk to the digital portrait, asking questions such as "How was your day?" or "Can you tell me about an old memory?"

[0526] The device captures the user's voice with a microphone and transmits the voice data to the server.

[0527] The server analyzes the received voice using automatic speech recognition (ASR) technology and understands the content of the question.

[0528] The server uses an NLP model to generate appropriate responses and a speech synthesis model to generate responses in the voice of the deceased.

[0529] The device plays back the responses received from the server, reproducing facial expressions and gestures as needed.

[0530] Specific examples

[0531] For example, if a user speaks to a digital portrait and says, "Hello, Grandpa. How was your day?", the following happens:

[0532] 1. User: Speaks a question.

[0533] 2. Device: Captures audio and sends it to the server.

[0534] 3. Server: Analyzes the voice and understands the question.

[0535] 4. Server: Generates a response using an NLP model, such as "Hello, it was a beautiful day today. How was it for you?"

[0536] 5. Server: Generate a response in the voice of the deceased person using a speech synthesis model.

[0537] 6. Terminal: Plays back the response and reproduces the facial expression of the deceased.

[0538] In this way, the system allows the bereaved to reconnect with the deceased and provide emotional care.

[0539] The processing flow will be explained below.

[0540] Step 1:

[0541] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[0542] Step 2:

[0543] The device uploads the exported data to the server in the specified format (e.g., JSON or CSV file).

[0544] Step 3:

[0545] The server receives the uploaded communication data and stores it in a secure database.

[0546] Step 4:

[0547] The server analyzes the stored data, removes unnecessary information (such as spam messages and errors), and extracts important information (key phrases, important topics, and frequently occurring vocabulary).

[0548] Step 5:

[0549] The server uses the preprocessed data to train a natural language processing (NLP) model, which analyzes and trains the text data to memorize the words and phrases used by the deceased.

[0550] Step 6:

[0551] The server uses the audio data to train a speech synthesis model, extracting clear speech and training it to reproduce the voice quality of the deceased person.

[0552] Step 7:

[0553] The server uses image and video data to train an image and video analysis model, analyzing the facial expressions and gestures of the deceased.

[0554] Step 8:

[0555] The server integrates the trained NLP model, speech synthesis model, and image / video analysis model, allowing it to operate as a single AI character.

[0556] Step 9:

[0557] The server packages the integrated AI character and sends it to the user's device.

[0558] Step 10:

[0559] Users can talk to the AI ​​character, which is displayed as a digital portrait of the deceased, asking questions such as, "How was your day?" or "Tell me about an old trip?"

[0560] Step 11:

[0561] The device captures the user's voice with a microphone and transmits the voice data to the server.

[0562] Step 12:

[0563] The server analyzes the received voice using automatic speech recognition (ASR) technology and understands the content of the question.

[0564] Step 13:

[0565] The server uses NLP models to generate appropriate responses, such as "It was a beautiful day today. How was it for you?" in response to a user question.

[0566] Step 14:

[0567] The server uses a speech synthesis model to generate the generated response in the voice of the deceased person.

[0568] Step 15:

[0569] The device plays back the responses received from the server, reproducing facial expressions and gestures as needed.

[0570] Example 1

[0571] 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."

[0572] In order to maintain the connection between memories and emotions of the deceased, there is a need for a system that allows bereaved families to reconnect with and interact with the deceased. However, existing technologies are not sufficient to reproduce the voice, facial expressions, and language of the deceased, and there are issues with a lack of realism and the complexity of operation. This means that the current situation is one in which the emotional care of bereaved families is not being provided adequately.

[0573] 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.

[0574] In this invention, the server includes: means for collecting communication data of the deceased; means for analyzing and preprocessing the communication data; means for training an AI model based on the preprocessed data; means for training the AI ​​model using a natural language processing model, a voice synthesis model, and an image / video analysis model; means for integrating the trained models and linking the data so that the words, voice, and facial expressions match; means for packaging the integrated AI character and transmitting it to a user's device; means for providing a digital memorial portrait device that displays the AI ​​character; and means for receiving voice input from the user using the digital memorial portrait device and generating and displaying a response using the AI ​​character. This makes it possible to reproduce the appearance, voice, and speech of the deceased with high accuracy, allowing bereaved family members to naturally converse and interact with the deceased.

[0575] "Communication data" refers to information such as message history, call records, images, and videos left by the deceased.

[0576] "Preprocessing" refers to the process of removing unnecessary information from the received data and extracting key phrases, important topics, etc.

[0577] "AI model" refers to an artificial intelligence model that analyzes and learns from text data, audio data, image and video data.

[0578] "Natural language processing model" refers to an AI model used to understand and generate human language.

[0579] "Speech synthesis model" refers to an AI model used to convert text data into speech and reproduce specific speech characteristics.

[0580] "Image and video analysis model" refers to an AI model used to analyze image and video data and extract and reproduce specific facial expressions and gestures.

[0581] "Data integration" refers to the process of integrating the output of each trained AI model to generate consistent words, voices, and facial expressions.

[0582] "Packaging" refers to the process of converting the integrated AI character into a data format that can be used on the user's device.

[0583] "Digital memorial device" refers to a device that can display an AI character and generate and display responses based on voice input.

[0584] To implement this invention, it is necessary to build a system that collects communication data of the deceased, generates an AI model, and uses it as a digital memorial portrait. This system uses various hardware and software to accurately reproduce the deceased's speech, voice, and facial expressions. Specific examples are shown below.

[0585] System Overview

[0586] This system collects communication data from the deceased, preprocesses it, trains an AI model, and displays the generated AI character as a digital portrait. Furthermore, when the user speaks to this digital portrait, it can respond in a way that is true to the deceased's personality.

[0587] Hardware and software used

[0588] Hardware: User's smartphone or PC, cloud server, microphone, dedicated display.

[0589] Software: Messaging applications, data transmission interface applications, cloud storage (e.g., Amazon S3), Python data processing libraries (e.g., spaCy, NLTK), speech analysis libraries (e.g., Librosa), natural language processing models (e.g., BERT, GPT-3), speech synthesis models (e.g., Tacotron2, WaveNet), image and video analysis libraries (e.g., OpenCV, Dlib), machine learning frameworks (e.g., TensorFlow, PyTorch).

[0590] Program processing flow

[0591] 1. Data collection phase:

[0592] Users log into messaging apps using their smartphones or PCs and export chat history, call logs, photos, and videos.

[0593] The device provides an interface for sending the exported data to the cloud server, for example, by clicking a "Data Upload" button in a dedicated application.

[0594] 2. Data preprocessing phase:

[0595] The server analyzes the communication data stored in the cloud storage, first filtering out spam messages and unwanted data using a Python script.

[0596] Next, we use an NLP library (e.g., spaCy) to extract key phrases and important topics, and then use topic modeling techniques (e.g., LDA) to cluster frequently occurring vocabulary.

[0597] 3. AI model training phase:

[0598] The server inputs the preprocessed text data into the BERT model, allowing it to learn the unique phrases and vocabulary of the deceased.

[0599] For the audio data, we use the Librosa library to remove noise and extract clear speech, and then use a voice synthesis model (e.g., Tacotron2) to recreate the voice quality of the deceased.

[0600] Image and video data is analyzed using OpenCV and Dlib, and the system is trained to recognize the facial expressions and gestures of the deceased.

[0601] 4. Model integration phase:

[0602] The server integrates each trained model (natural language processing model, speech synthesis model, image and video analysis model) and uses machine learning frameworks such as TensorFlow to coordinate the data so that the integrated AI character can generate consistent words, voices, and facial expressions.

[0603] 5. Digital portrait generation phase:

[0604] The server then sends the integrated AI character to the user's device, where the user can view the digital portrait using a dedicated display or smartphone.

[0605] 6. Dialogue Phase:

[0606] The user talks to the digital portrait, asking questions such as, "Hello, Grandpa. How was your day?"

[0607] The terminal captures the user's voice with a microphone and transmits the voice data to the server in real time.

[0608] The server uses speech recognition technology (e.g., Google Speech-to-Text API) to convert the received voice into text and understand the content of the question.

[0609] Next, an NLP model is used to generate appropriate responses, and a speech synthesis model generates the responses in the voice of the deceased.

[0610] The device plays back the response received from the server and also displays animations that reproduce the facial expressions and gestures of the deceased.

[0611] Specific examples

[0612] For example, if a user says, "Hello, Grandpa. How was your day?", the process is as follows:

[0613] 1. User: Speaks a question.

[0614] 2. Device: Captures audio and sends it to the server.

[0615] 3. Server: Analyzes the voice and understands the question.

[0616] 4. Server: Generates a response using an NLP model, for example, "Hello, it was a beautiful day today. How was it for you?"

[0617] 5. Server: Generate a response in the voice of the deceased person using a speech synthesis model.

[0618] 6. Terminal: Plays back the response and reproduces the facial expression of the deceased.

[0619] Prompt Sentence Examples

[0620] "Tell me some recent memories of your grandpa."

[0621] "Tell me about your grandpa's hobbies."

[0622] "Where was your last trip?"

[0623] This allows the user to provide emotional care through intimate conversations with the deceased.

[0624] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0625] Step 1:

[0626] Input: User accesses deceased person's communication accounts and exports chat history, call logs, photos, and videos.

[0627] How it works: A user logs into a messaging app using a smartphone or PC, accesses the "Settings" menu, selects the "Export Chat History" option, and exports the relevant data.

[0628] Output: The exported data file.

[0629] Step 2:

[0630] Input: Terminal exported data file.

[0631] Specific operation: The terminal provides an interface for the user to send the exported data to the server. For example, the user clicks the "Upload Data" button in the dedicated application to send the data to the server.

[0632] Output: The data file sent to the server.

[0633] Step 3:

[0634] Input: The server receives the transmitted data file.

[0635] Specific operation: The server stores the received data in cloud storage (e.g., Amazon S3).

[0636] Output: Data stored in cloud storage.

[0637] Step 4:

[0638] Input: Data stored in cloud storage.

[0639] What it does: The server uses Python scripts to filter spam messages and error messages from the stored data, then uses NLP libraries (e.g., spaCy) to extract key phrases and important topics.

[0640] Output: Preprocessed text data.

[0641] Step 5:

[0642] Input: Preprocessed text data.

[0643] How it works: The server inputs the preprocessed text data into the BERT model, allowing it to learn the unique phrases and vocabulary of the deceased.

[0644] Output: A trained natural language processing model.

[0645] Step 6:

[0646] Input: Audio data stored in cloud storage.

[0647] How it works: The server uses the Librosa library to analyze the audio data, remove noise, and extract clear audio.

[0648] Output: Preprocessed audio data.

[0649] Step 7:

[0650] Input: Preprocessed audio data.

[0651] Specific operation: The server uses the Tacotron2 model to train the preprocessed audio data to reproduce the voice quality of the deceased.

[0652] Output: The trained speech synthesis model.

[0653] Step 8:

[0654] Input: Image and video data stored in cloud storage.

[0655] Specific operation: The server uses OpenCV and Dlib to analyze the facial expressions and gestures of the deceased in image and video data.

[0656] Output: Preprocessed image and video data.

[0657] Step 9:

[0658] Input: Preprocessed image and video data.

[0659] Specific operation: The server uses image and video analysis models to learn how to reproduce the facial expressions and movements of the deceased.

[0660] Output: A trained image and video analysis model.

[0661] Step 10:

[0662] Input: Trained natural language processing models, speech synthesis models, and image / video analysis models.

[0663] How it works: The server uses machine learning frameworks such as TensorFlow to integrate each model and generate a single AI character, coordinating verbal, vocal, and facial output to achieve results that are closer to real-life interactions.

[0664] Output: A unified AI character.

[0665] Step 11:

[0666] Input: Integrated AI character.

[0667] Specific operation: The server packages this AI character and sends it to the user's device through a dedicated API.

[0668] Output: The AI ​​character sent to the user's device.

[0669] Step 12:

[0670] Input: A question that the user speaks.

[0671] Specific Actions: The user talks to the digital portrait, asking questions such as, "Hello, Grandpa. How was your day?"

[0672] Output: The audio data captured on the device.

[0673] Step 13:

[0674] Input: Audio data captured on the device.

[0675] How it works: The device sends the captured voice data to the server, which uses the Google Speech-to-Text API to convert the voice data into text and understand the question.

[0676] Output: Parsed text data.

[0677] Step 14:

[0678] Input: Parsed text data.

[0679] What happens: The server uses the NLP model to generate an appropriate response, such as "Hello, it was a beautiful day today. How was it for you?"

[0680] Output: The generated text response.

[0681] Step 15:

[0682] Input: The generated text response.

[0683] What it does: The server uses the Tacotron2 model to vocalize the text response in the voice of the deceased.

[0684] Output: A spoken response.

[0685] Step 16:

[0686] Input: A spoken response.

[0687] What it does: The device plays back spoken responses and also displays animations that replicate the facial expressions and gestures of the deceased.

[0688] Output: Played voice response and animation.

[0689] (Application example 1)

[0690] 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."

[0691] In recent years, with the advancement of AI technology, there has been an increase in efforts to generate digital memorial portraits using the communication data of the deceased, but conventional technologies have limited interaction capabilities with users and lack security.This invention provides a system that provides real-time response guidance from an AI character that resembles the deceased, and detects and warns abnormal behavior using a score prediction model, thereby achieving a safe and comfortable interaction experience for users.

[0692] 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.

[0693] In this invention, the server includes means for collecting communication data of the deceased, means for analyzing and preprocessing the communication data, means for training an AI model based on the preprocessed data, means for integrating the trained AI model to generate an AI character resembling the deceased, means for providing a digital memorial portrait device that displays the AI ​​character, means for receiving voice input from a user using the digital memorial portrait device and generating and displaying a response using the AI ​​character, and means for detecting abnormal behavior using a score prediction model and providing a real-time audio warning. This allows for a natural conversational experience with the deceased while simultaneously alerting security guards to any abnormal behavior.

[0694] "Communication data of the deceased" refers to a collection of electronic data used by the deceased during their lifetime, and primarily includes chat history, call records, photos, and videos.

[0695] "Preprocessing" refers to the initial stage of data processing, where collected raw data is analyzed, unnecessary information is removed, and important information is extracted.

[0696] An "AI model" is an artificial intelligence program that has been trained to perform a specific task using machine learning or deep learning techniques.

[0697] An "AI character" is a digital character that integrates a trained AI model and reproduces the characteristics of a specific person (in this case, a deceased person).

[0698] A "digital memorial device" is a device that displays an AI character and enables interaction with the user, and can be a smartphone, a dedicated display, or AR-compatible smart glasses.

[0699] "Voice input from the user" refers to voice instructions or questions given by the user to the system via a microphone or the like.

[0700] A "score prediction model" is a model that uses AI technology to analyze data in real time and predict risk scores associated with specific behaviors or situations.

[0701] "Abnormal behavior detection" refers to the process of recognizing and alerting when behavior that deviates from normal patterns occurs.

[0702] "Providing a real-time warning" refers to the act of immediately issuing a warning and notifying the user of detected abnormal behavior.

[0703] This invention provides a system that collects communication data of the deceased, uses it to build an AI model, and uses it to create a digital memorial portrait. The system aims to generate an AI character that resembles the deceased and provides responses that are characteristic of the deceased through conversations with the user. It also has the ability to detect abnormal behavior and provide real-time warnings using a score prediction model.

[0704] System Configuration

[0705] 1. Hardware Configuration

[0706] Server: Responsible for storing and processing data, and training and operating AI models. A server with high-performance processing capabilities is desirable.

[0707] Device: The device used by the user, which may be a smartphone, a dedicated display, or AR-enabled smart glasses.

[0708] Camera: Used to capture footage in real time and analyze it in the score prediction model.

[0709] Microphone: Used to capture the user's voice input.

[0710] 2. Software Configuration

[0711] Data collection module: Collects communication data such as chat history, call records, photos, and videos of the deceased.

[0712] Data preprocessing module: Analyzes the collected data and extracts the necessary information.

[0713] AI model learning module: Based on the preprocessed data, natural language processing (NLP) models, speech synthesis models, image and video analysis models, and score prediction models are trained.

[0714] Integration module: Integrates each model to generate an AI character that resembles the deceased.

[0715] Digital portrait display module: Displays the generated AI character and enables interaction with the user.

[0716] Score prediction module: Detects abnormal behavior in real time based on camera footage and provides necessary warnings via voice.

[0717] Operating Procedure

[0718] 1. Data Collection: Users enter the deceased person's communication data into the system, including chat history, call records, photos, videos, etc.

[0719] 2. Data Preprocessing: The server preprocesses the received data, removing unnecessary information and extracting important information. This process includes filtering spam messages and extracting key phrases.

[0720] 3. AI model training: The server uses the preprocessed data to train the NLP model, speech synthesis model, image and video analysis model, and score prediction model, thereby acquiring knowledge to reproduce the deceased's unique speech patterns, voice quality, facial expressions, etc.

[0721] 4. Model integration: The output of each model is integrated to generate an AI character that resembles the deceased person, with matching language, voice, and facial expressions.

[0722] 5. Interactive function: The user can talk to the digital portrait, for example, asking questions like, "How was your day?" The device captures this voice and sends it to the server.

[0723] 6. Real-time warning: Camera footage is analyzed in real time, and if abnormal behavior is detected, a warning is immediately given to the user, such as an audio warning saying "Warning! Intruder detected."

[0724] Specific examples

[0725] When a user speaks to a digital portrait and asks, "Hello, how was your day?", the response is generated through the following process:

[0726] 1. User: Speaks a question.

[0727] 2. Device: Captures audio and sends it to the server.

[0728] 3. Server: Analyzes the voice and understands the question.

[0729] 4. Server: Generates a response using an NLP model, such as "It was a beautiful day today. How was it for you?"

[0730] 5. Server: Generates a response in the voice of the deceased person using a speech synthesis model and sends it to the device.

[0731] 6. Terminal: Plays back the response and reproduces the facial expression of the deceased.

[0732] Prompt Sentence Examples

[0733] Example prompt sentence:

[0734] "Warning! Unauthorized person loitering near conference room. Please investigate immediately."

[0735] "Attention! Abnormal behavior detected. Please respond immediately."

[0736] This allows users to have a natural conversation with the deceased while also supporting immediate response as a security guard.

[0737] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0738] Step 1:

[0739] Data collection

[0740] The user inputs the deceased's communication data (chat history, call records, photos, videos, etc.) into the system. The device exports this data and sends it to the server. The server receives and stores the data.

[0741] Input: Chat history, call logs, photos, videos

[0742] Output: Saved communication data

[0743] Step 2:

[0744] Data Preprocessing

[0745] The server analyzes the received communication data, removes unnecessary information (spam messages and errors), and extracts important information (key phrases, frequently occurring vocabulary, and important topics). This preprocessing process prepares the data for AI model training.

[0746] Input: Saved communication data

[0747] Output: Preprocessed data

[0748] Step 3:

[0749] Training an AI model

[0750] The server uses the preprocessed data to train various AI models (NLP model, voice synthesis model, image / video analysis model, score prediction model), which then learns the deceased's speech patterns, voice quality, facial expressions, and behavioral patterns.

[0751] Input: Preprocessed data

[0752] Output: Trained AI model (NLP model, speech synthesis model, image / video analysis model, score prediction model)

[0753] Step 4:

[0754] Model Integration

[0755] The server then integrates each trained AI model to generate an AI character that resembles the deceased. The integration process involves linking the outputs of each model and adjusting the words, voice, and facial expressions to match.

[0756] Input: Trained AI model

[0757] Output: Integrated AI character

[0758] Step 5:

[0759] Displaying a digital portrait of the deceased

[0760] The server packages the integrated AI character and sends it to the terminal, which uses the received data to display the AI ​​character on the digital portrait device.

[0761] Input: Integrated AI character

[0762] Output: Displayed AI character

[0763] Step 6:

[0764] User interaction

[0765] The user speaks to the digital portrait. The device captures the voice with a microphone and sends the captured voice data to the server. The server uses speech recognition technology to analyze and understand the question. It uses an NLP model to generate an appropriate response and a speech synthesis model to generate a response in the deceased's voice. Finally, the response is sent to the device, which plays it back. In some cases, the deceased's facial expressions and gestures are also reproduced.

[0766] Input: User's voice

[0767] Output: Vocal responses, facial expressions, and gestures of the deceased

[0768] Step 7:

[0769] Abnormal behavior detection and real-time alerts

[0770] The device's camera captures video in real time. The server analyzes the video using a score prediction model, and if abnormal behavior is detected, it immediately provides an audio warning. For example, it notifies the user with a voice message saying, "Warning! Intruder detected."

[0771] Input: Real-time video

[0772] Output: Audio warning, abnormal behavior detection results

[0773] 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.

[0774] To implement this invention, we will create an AI model using the communication data of the deceased and build a system that can be used as a digital memorial portrait. This system will also incorporate an emotion engine that recognizes the user's emotions and provides responses according to their emotions.

[0775] System Overview

[0776] The system collects communication data from the deceased, trains an AI model using preprocessed data, and displays the generated AI character as a digital portrait. Furthermore, when a user speaks to the digital portrait, it uses an emotion engine to analyze the emotions and provide appropriate responses.

[0777] Program processing flow

[0778] 1. Data Collection Phase

[0779] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[0780] The device uploads the exported data to the server in the specified format (e.g., JSON or CSV file).

[0781] 2. Data preprocessing phase

[0782] The server stores the received communication data and performs pre-processing on the data, which includes removing unnecessary information (such as spam messages and errors) and extracting important information (key phrases, important topics, and frequently occurring vocabulary).

[0783] 3. AI model training phase

[0784] The server uses the preprocessed data to train natural language processing (NLP) models, speech synthesis models, and image and video analysis models. Each model works as follows:

[0785] NLP model: Analyzes text data and learns the words and phrases of the deceased.

[0786] Speech synthesis model: Extracts clear speech and trains to reproduce the voice quality of the deceased.

[0787] Image and video analysis model: Learns facial expressions and gestures and reproduces the facial expressions and movements of the deceased.

[0788] 4. Model integration phase

[0789] The server then integrates the trained models to operate as a single AI character, a process that involves linking the outputs of each model and ensuring that the words, voices, and facial expressions match.

[0790] 5. Digital portrait generation phase

[0791] The server packages the integrated AI character and sends it to the user's device, where it can be displayed on a digital memorial device (such as a dedicated display or smartphone).

[0792] 6. Emotion recognition and response phase

[0793] The emotion engine analyzes the user's voice data and facial expression data to recognize the user's emotions (e.g., happiness, sadness, anger, surprise, etc.).

[0794] The server adapts the NLP model and speech synthesis model based on the output of the emotion engine, allowing the AI ​​character of the deceased person to generate an appropriate response based on the recognized emotion.

[0795] The server also adjusts facial expressions and tone of voice according to emotions to recreate natural conversations.

[0796] Specific examples

[0797] For example, if a user says to a digital portrait, "Hello, Grandpa. I've been feeling a bit sad today," the following happens:

[0798] 1. User: Speaks a question.

[0799] 2. Device: Captures audio and sends it to the server.

[0800] 3. Server: Analyzes the voice and understands the question.

[0801] 4. Emotion engine: Analyzes the user's voice and facial expressions to recognize when the user is feeling "sad."

[0802] 5. Server: Uses NLP models to generate responses based on emotions, such as "Hi, I'm very worried to hear that you're feeling sad. What happened?"

[0803] 6. Server: Using a speech synthesis model, the generated response is generated in the voice of the deceased.

[0804] 7. Server: Based on the emotion engine, the tone of voice and facial expressions are adjusted to interact with the user in a more natural way.

[0805] 8. The device plays back the response received from the server and reproduces the facial expression of the deceased.

[0806] In this way, the system recognizes the user's emotions and allows them to provide emotional care through dialogue with the deceased.

[0807] The processing flow will be explained below.

[0808] Step 1:

[0809] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[0810] Step 2:

[0811] The device uploads the exported data to the server in the specified format (e.g., JSON or CSV file).

[0812] Step 3:

[0813] The server receives the uploaded communication data and stores it in a secure database.

[0814] Step 4:

[0815] The server analyzes the stored data, removes unnecessary information (such as spam messages and errors), and extracts important information (key phrases, important topics, and frequently occurring vocabulary).

[0816] Step 5:

[0817] The server uses the preprocessed data to train a natural language processing (NLP) model, which analyzes and trains the text data to memorize the words and phrases used by the deceased.

[0818] Step 6:

[0819] The server uses the audio data to train a speech synthesis model, extracting clear speech and training it to reproduce the voice quality of the deceased person.

[0820] Step 7:

[0821] The server uses image and video data to train an image and video analysis model, analyzing the facial expressions and gestures of the deceased.

[0822] Step 8:

[0823] The server integrates the trained NLP model, speech synthesis model, and image / video analysis model, allowing it to operate as a single AI character.

[0824] Step 9:

[0825] The server packages the integrated AI character and sends it to the user's device.

[0826] Step 10:

[0827] Users can talk to the AI ​​character, which is displayed as a digital portrait of the deceased, asking questions such as, "How was your day?" or "Tell me about an old trip?"

[0828] Step 11:

[0829] The device captures the user's voice with a microphone and transmits the voice data to the server.

[0830] Step 12:

[0831] The server analyzes the received voice using automatic speech recognition (ASR) technology and understands the content of the question.

[0832] Step 13:

[0833] The server uses an emotion engine to analyze the user's voice data and facial expression data and recognize the user's emotions (e.g., happiness, sadness, anger, surprise, etc.).

[0834] Step 14:

[0835] The server uses NLP models to generate appropriate responses based on the recognized emotions, for example, if the user is expressing sadness, it generates a response such as "Hi, I'm very worried to hear that you're feeling sad. What happened?"

[0836] Step 15:

[0837] The server uses a speech synthesis model to generate the generated response in the voice of the deceased person, adjusting the tone of the voice based on the output of the emotion engine.

[0838] Step 16:

[0839] The server sends the generated response to the device, adjusting facial expressions and gestures as needed.

[0840] Step 17:

[0841] The device then plays back the responses received from the server and reproduces the facial expressions of the deceased, allowing the user to have a natural conversation.

[0842] Example 2

[0843] 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."

[0844] It is necessary to utilize the communication data of the deceased to create a digital memorial portrait that reflects the deceased's characteristics, and to realize a dialogue between the user and the deceased's AI character based on emotion recognition. However, conventional technologies have low accuracy in analyzing communication data and recognizing emotions, making it difficult to provide natural responses that correspond to the user's emotions. Another issue is the difficulty of integrating various models (e.g., natural language processing, speech synthesis, image and video analysis) to accurately reproduce the characteristics of the deceased.

[0845] 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.

[0846] In this invention, the server includes means for collecting communication data of the deceased, means for analyzing and preprocessing the communication data, means for training an AI model based on the preprocessed data, means for integrating the trained AI model to generate an AI character resembling the deceased, means for providing a display device for displaying the AI ​​character, means for receiving voice input from a user via the display device and generating and displaying a response using the AI ​​character, and means for recognizing the user's emotions and generating a response according to the emotion. This makes it possible to accurately reproduce the characteristics of the deceased while providing a natural and appropriate response according to the user's emotions.

[0847] "Communication data" refers to records of digital communications used by the deceased during their lifetime, such as chat history, call records, photos, and videos.

[0848] "Preprocessing" is the process of removing unnecessary information from the received data and extracting important information (such as key phrases and frequently occurring vocabulary).

[0849] An "AI model" is a collection of algorithms that use artificial intelligence techniques to analyze and learn from data and automate specific tasks.

[0850] A "natural language processing model" is a model that uses artificial intelligence technology to analyze human language and understand its meaning.

[0851] A "speech synthesis model" is a model that uses artificial intelligence technology to convert text data into speech.

[0852] The "image and video analysis model" is a model that uses artificial intelligence technology to analyze image and video data and understand its content.

[0853] An "AI character" is a digital character generated based on an AI model that reproduces the features of a deceased person.

[0854] A "display device" is a device for visually displaying an AI character to a user.

[0855] "Emotion recognition" is a technology that analyzes a user's voice data and facial expression data to identify their emotions.

[0856] To implement this invention, it is necessary to build a system that uses the communication data of the deceased to generate an AI model and use it as a digital memorial portrait. This system is equipped with an emotion engine that recognizes the user's emotions and provides responses according to the emotions.

[0857] Hardware and software used

[0858] Hardware

[0859] Server: A server for storing and analyzing data and training AI models. Use a cloud server (e.g., Amazon EC2) equipped with a high-performance processor and large memory capacity.

[0860] Terminal: A device that allows users to operate and input data. This includes smartphones, tablets, and PCs.

[0861] Display device: A device for displaying the digital portrait. A dedicated display or smartphone can be used.

[0862] software

[0863] Natural Language Processing (NLP) model: A model for analyzing text data and learning the vocabulary and phrasing of the deceased, using NLP techniques such as BERT (Bidirectional Encoder Representations from Transformers).

[0864] Speech synthesis model: A model for converting text to speech and recreating the voice quality of the deceased. Uses voice synthesis technology such as Tacotron 2.

[0865] Image and video analysis model: A model that analyzes facial expressions and gestures to recreate the movements and expressions of the deceased. It uses image and video analysis technologies such as OpenPose and Face++.

[0866] Emotion engine: A technology that analyzes the user's voice data and facial expression data to recognize the user's emotions. It uses a combination of DeepSpeech (voice recognition) and Face++ (facial expression recognition).

[0867] Specific examples

[0868] Data Collection Phase

[0869] The user exports the chat history from the deceased person's messaging app (e.g., WhatsApp), and then uploads it to the server using a dedicated app.

[0870] Data Preprocessing Phase

[0871] The server analyzes the received chat history, filters out unnecessary information, and then extracts important key phrases and frequently occurring vocabulary.

[0872] AI model training phase

[0873] The server uses the preprocessed data to train a BERT model to learn the deceased's words and phrases, then trains the model with Tacotron 2 on clear audio data to reproduce the deceased's voice, and trains the model with OpenPose and Face++ on facial expressions and gestures.

[0874] Model integration phase

[0875] The server integrates each model (natural language processing model, speech synthesis model, image / video analysis model) to generate an AI character that recreates the deceased person.

[0876] Usage example

[0877] For example, if a user says to a digital portrait, "Hello, Grandpa. I've been feeling a bit sad today," the following happens:

[0878] The user enters the question by voice.

[0879] The device captures the audio and sends it to the server.

[0880] The server analyzes the voice and understands the question.

[0881] The emotion engine analyzes the user's voice and facial expressions and recognizes that the user is feeling "sad."

[0882] The server uses an NLP model to generate a response such as, "Hello, I'm very worried to hear that you're feeling sad. What happened?"

[0883] The server uses a speech synthesis model to generate responses in the voice of the deceased, adjusting the tone of voice and facial expression depending on the emotion.

[0884] The device plays back the response received from the server and reproduces the facial expression of the deceased.

[0885] Prompt Sentence Examples

[0886] "Analyze the message data of the deceased and generate an AI model for natural conversation."

[0887] "Develop a system that recognizes and responds to user emotions. The system will use an emotion engine to engage in appropriate dialogue based on the user's emotions."

[0888] This allows the system to recognize the user's emotions and provide emotional care through dialogue with the deceased.

[0889] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0890] Step 1: Data collection

[0891] The user accesses the deceased person's messaging and social media accounts and exports chat history, call logs, photos, and videos (input: the deceased person's communication data).

[0892] The terminal uploads the exported data to the server in the specified format (e.g., JSON file or CSV file) (output: uploaded communication data).

[0893] Step 2: Data storage and preprocessing

[0894] The server receives the data sent from the terminal and stores it securely (input: uploaded communication data, output: stored communication data).

[0895] The server removes unnecessary information from the received data (specific operations: filtering spam messages and error logs) and extracts important key phrases and frequently occurring vocabulary (input: saved communication data, output: preprocessed data).

[0896] Step 3: Training the Natural Language Processing (NLP) Model

[0897] The server uses the preprocessed text data to train an NLP model (e.g., BERT) (input: preprocessed data, output: trained NLP model). Specifically, it tokenizes the raw data and trains the model to understand the structure of the text.

[0898] Step 4: Training the speech synthesis model

[0899] The server extracts the preprocessed voice data and trains it using a speech synthesis model (e.g., Tacotron 2) (input: preprocessed voice data, output: trained speech synthesis model). Specifically, it analyzes the features of the voice data and reproduces the voice quality of the deceased.

[0900] Step 5: Training the image and video analysis model

[0901] The server uses the preprocessed image and video data to train an image and video analysis model (e.g., OpenPose, Face++) (input: preprocessed image and video data, output: trained image and video analysis model). Specifically, it analyzes the facial expressions and gestures of the deceased and reproduces them.

[0902] Step 6: Model integration

[0903] The server integrates each trained model to generate a single AI character (input: trained NLP model, trained speech synthesis model, trained image and video analysis model, output: integrated AI character). Specifically, the text generated by the NLP model is input into the speech synthesis model, and the corresponding facial expression is generated by the image and video analysis model.

[0904] Step 7: Generate a digital portrait

[0905] The server packages the integrated AI character and sends it to the user's device (input: integrated AI character, output: package sent to the user's device).

[0906] The user installs the file sent to their device and displays the AI ​​character on the digital portrait device (input: sent package, output: displayed AI character).

[0907] Step 8: Emotion Recognition and Response Generation

[0908] The emotion engine analyzes the user's voice data and facial expression data and recognizes emotions (input: user's voice data, facial expression data, output: recognized emotions).

[0909] Based on the output of the emotion engine, the server adapts the NLP model and speech synthesis model to generate an appropriate response (input: recognized emotion, output: generated response).

[0910] The server generates the response in the voice of the deceased person and adjusts the tone of voice and facial expression according to the emotion (input: generated response, output: adjusted response and facial expression).

[0911] The device plays back the responses received from the server and reproduces the facial expressions of the deceased (input: adjusted responses and facial expressions, output: reproduced responses and facial expressions).

[0912] This series of processes allows the user to receive emotional care through natural conversation with the deceased.

[0913] (Application example 2)

[0914] 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."

[0915] When applying a digital memorial portrait system to a virtual store, a means of understanding the user's emotions and responding appropriately is required. Another challenge is to increase customer satisfaction by providing a customer service experience that recreates the memories and voice of the deceased.

[0916] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0917] In this invention, the server includes means for collecting communication data of the deceased, means for analyzing and preprocessing the communication data, means for training an AI model based on the preprocessed data, means for integrating the trained AI model to generate an AI character resembling the deceased, means for providing a digital portrait device that displays the AI ​​character, means for receiving voice input from a user using the digital portrait device and generating and displaying a response using the AI ​​character, means for analyzing the user's emotions using an emotion recognition engine, means for generating a response based on the user's emotions, and means for using the digital portrait device as a guide in a virtual store. This makes it possible to grasp the user's emotions and provide effective guidance in the virtual store by reproducing the deceased's responses according to those emotions.

[0918] "Communication data" refers to digital information left behind by the deceased, such as chat history, call records, photos, and videos.

[0919] "Preprocessing" refers to the process of removing unnecessary information from collected communication data and extracting important information.

[0920] An "AI model" is an artificial intelligence that is trained using technologies such as natural language processing, speech synthesis, and image and video analysis to carry out specific tasks.

[0921] An "AI character" is a digital character that resembles a deceased person and is generated by integrating a trained AI model.

[0922] A "digital memorial device" is a device that displays an AI character and allows the user to interact with it.

[0923] "Voice input" refers to the voice data that is generated when the user speaks to the digital memorial portrait device.

[0924] An "emotion recognition engine" is a system for analyzing emotions from a user's voice and images.

[0925] A "virtual store" is a store that operates in a virtual space, a place that offers products and services online.

[0926] A "guide" is a person whose role is to provide customers with product information and guidance within the store.

[0927] A "user" is a person who uses the digital memorial device to interact with an AI character of a deceased person.

[0928] To implement this invention, it is necessary to create an AI model using the communication data of the deceased and build a system that functions as a digital memorial guide in a virtual store. This system reproduces the memories and voice of the deceased and combines it with an emotion recognition engine to provide appropriate responses according to the user's emotions.

[0929] System Configuration

[0930] 1. The user uses a smartphone or smart glasses to talk to a digital portrait guide.

[0931] 2. The device captures the user's voice input and facial image data and sends them to the server.

[0932] 3. The server executes multiple methods to perform the following processes:

[0933] Data collection and preprocessing

[0934] The server collects the deceased's communication data (chat history, call records, photos, videos). The collected data is pre-processed using technologies such as NLP (natural language processing) and image and video analysis to remove unnecessary information and extract important information.

[0935] AI model training and integration

[0936] The server uses the preprocessed data to train natural language processing models, speech synthesis models, and image and video analysis models. These trained models are then integrated to generate an AI character that resembles the deceased. This AI character reproduces the deceased's speech, voice quality, facial expressions, etc., allowing users to interact with the deceased in a virtual store.

[0937] Emotion Recognition and Response Generation

[0938] The server is equipped with an emotion recognition engine that analyzes the user's emotions from their voice and facial expression data. Based on the emotion engine's analysis results, an NLP model generates an appropriate text response, and a speech synthesis model reproduces that text as the deceased's voice. This enables natural conversations that correspond to the user's emotions.

[0939] Use as a digital memorial guide

[0940] In the virtual store, users can interact with a digital memorial guide through their smartphones or smart glasses. For example, if a customer asks, "I'd like to know more about our new products," the server will analyze their voice and facial expression data and provide an answer. Specifically, it will respond with something like, "Of course. Here are our latest products, and their features are as follows."

[0941] Prompt Sentence Examples

[0942] An example prompt for generating an AI model is:

[0943] You are a digital memorial guide working in a virtual store. Respond to customers' questions and interests in a friendly and courteous manner. Strive to provide relevant and useful information while taking into consideration the customer's feelings. For example, if a customer asks, "I'd like to know more about our new products," you can respond, "Of course! Here are our latest products and their features," and then provide specific product information.

[0944] As described above, the present invention makes it possible to utilize a digital portrait of the deceased as a guide in a virtual store while being sensitive to the user's emotions.

[0945] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0946] Step 1:

[0947] The user speaks to the digital portrait guide through a smartphone or smart glasses. The user inputs questions by voice, and simultaneously captures facial image data. This input data is captured by the device, which then transmits the voice data and facial image data to a server.

[0948] Step 2:

[0949] The server analyzes the received voice data. Specifically, it converts it into text data using voice recognition technology (such as Google Speech API). At this time, voice data is input, and as a result, text data containing the user's question is output.

[0950] Step 3:

[0951] The server sends the received facial image data to an emotion recognition engine to analyze the user's emotions. The emotion recognition engine uses image analysis technology (such as OpenCV or Dlib) to analyze emotions from facial expressions. The input is facial image data, and the output is data indicating the user's emotions (e.g., happiness, sadness, anger, etc.).

[0952] Step 4:

[0953] The server uses the analyzed text data and emotion data to work with an NLP model to generate an appropriate text response. Specifically, NLP techniques are applied to the text data to generate candidate responses. The emotion data is used to adjust the response. The input is text data containing the question and emotion data, and the output is the generated appropriate response text.

[0954] Step 5:

[0955] The server sends the generated response text to a speech synthesis model, which generates audio data in the voice of the deceased. A speech synthesis model (such as Amazon Polly or Google Text-to-Speech) is used. The input is the response text, and the output is audio data that reproduces the voice of the deceased.

[0956] Step 6:

[0957] The server sends the generated voice data to the terminal, which then plays the received voice data to the user, allowing the user to interact with the digital portrait guide. The input is the voice data, and the output is the played voice response.

[0958] Step 7:

[0959] After the server or terminal has completed responding to the user's question, it waits for the next question or interaction. If the user has any new input, it returns to step 1.

[0960] Through the above processing steps, the user can have natural conversations with the digital portrait guide and is provided with guidance services in the virtual store.

[0961] 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.

[0962] 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.

[0963] 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.

[0964] [Third embodiment]

[0965] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0966] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0967] 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).

[0968] 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.

[0969] 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.

[0970] 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).

[0971] 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.

[0972] 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.

[0973] 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.

[0974] 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.

[0975] 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.

[0976] 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."

[0977] To implement this invention, we will build a system that uses the communication data of the deceased to generate an AI model and use it as a digital memorial portrait. The following explains the program processing of this system in natural language.

[0978] System Overview

[0979] The system collects communication data from the deceased, trains an AI model using preprocessed data, and displays the generated AI character as a digital portrait. Furthermore, users can talk to this digital portrait and it responds in a way that is true to the deceased's personality.

[0980] Program processing flow

[0981] 1. Data Collection Phase

[0982] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[0983] The terminal provides an interface for sending the exported data to the server.

[0984] 2. Data preprocessing phase

[0985] The server stores the received communication data and performs pre-processing on the data, which includes removing unnecessary information (e.g., spam messages, errors) and extracting important information (e.g., key phrases, important topics, and frequently occurring vocabulary).

[0986] 3. AI model training phase

[0987] The server uses the preprocessed data to train natural language processing (NLP) models, speech synthesis models, and image and video analysis models. Each model works as follows:

[0988] NLP model: Analyzes text data and learns the words and phrases of the deceased.

[0989] Speech synthesis model: Extracts clear speech and trains to reproduce the voice quality of the deceased.

[0990] Image and video analysis model: Learns facial expressions and gestures and reproduces the facial expressions and movements of the deceased.

[0991] 4. Model integration phase

[0992] The server then integrates the trained models to operate as a single AI character, a process that involves linking the outputs of each model and ensuring that the words, voices, and facial expressions match.

[0993] 5. Digital portrait generation phase

[0994] The server packages the integrated AI character and sends it to the user's device, where it can be displayed on a digital memorial device (such as a dedicated display or smartphone).

[0995] 6. Dialogue Phase

[0996] Users can talk to the digital portrait, asking questions such as "How was your day?" or "Can you tell me about an old memory?"

[0997] The device captures the user's voice with a microphone and transmits the voice data to the server.

[0998] The server analyzes the received voice using automatic speech recognition (ASR) technology and understands the content of the question.

[0999] The server uses an NLP model to generate appropriate responses and a speech synthesis model to generate responses in the voice of the deceased.

[1000] The device plays back the responses received from the server, reproducing facial expressions and gestures as needed.

[1001] Specific examples

[1002] For example, if a user speaks to a digital portrait and says, "Hello, Grandpa. How was your day?", the following happens:

[1003] 1. User: Speaks a question.

[1004] 2. Device: Captures audio and sends it to the server.

[1005] 3. Server: Analyzes the voice and understands the question.

[1006] 4. Server: Generates a response using an NLP model, such as "Hello, it was a beautiful day today. How was it for you?"

[1007] 5. Server: Generate a response in the voice of the deceased person using a speech synthesis model.

[1008] 6. Terminal: Plays back the response and reproduces the facial expression of the deceased.

[1009] In this way, the system allows the bereaved to reconnect with the deceased and provide emotional care.

[1010] The processing flow will be explained below.

[1011] Step 1:

[1012] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[1013] Step 2:

[1014] The device uploads the exported data to the server in the specified format (e.g., JSON or CSV file).

[1015] Step 3:

[1016] The server receives the uploaded communication data and stores it in a secure database.

[1017] Step 4:

[1018] The server analyzes the stored data, removes unnecessary information (such as spam messages and errors), and extracts important information (key phrases, important topics, and frequently occurring vocabulary).

[1019] Step 5:

[1020] The server uses the preprocessed data to train a natural language processing (NLP) model, which analyzes and trains the text data to memorize the words and phrases used by the deceased.

[1021] Step 6:

[1022] The server uses the audio data to train a speech synthesis model, extracting clear speech and training it to reproduce the voice quality of the deceased person.

[1023] Step 7:

[1024] The server uses image and video data to train an image and video analysis model, analyzing the facial expressions and gestures of the deceased.

[1025] Step 8:

[1026] The server integrates the trained NLP model, speech synthesis model, and image / video analysis model, allowing it to operate as a single AI character.

[1027] Step 9:

[1028] The server packages the integrated AI character and sends it to the user's device.

[1029] Step 10:

[1030] Users can talk to the AI ​​character, which is displayed as a digital portrait of the deceased, asking questions such as, "How was your day?" or "Tell me about an old trip?"

[1031] Step 11:

[1032] The device captures the user's voice with a microphone and transmits the voice data to the server.

[1033] Step 12:

[1034] The server analyzes the received voice using automatic speech recognition (ASR) technology and understands the content of the question.

[1035] Step 13:

[1036] The server uses NLP models to generate appropriate responses, such as "It was a beautiful day today. How was it for you?" in response to a user question.

[1037] Step 14:

[1038] The server uses a speech synthesis model to generate the generated response in the voice of the deceased person.

[1039] Step 15:

[1040] The device plays back the responses received from the server, reproducing facial expressions and gestures as needed.

[1041] Example 1

[1042] 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."

[1043] In order to maintain the connection between memories and emotions of the deceased, there is a need for a system that allows bereaved families to reconnect with and interact with the deceased. However, existing technologies are not sufficient to reproduce the voice, facial expressions, and language of the deceased, and there are issues with a lack of realism and the complexity of operation. This means that the current situation is one in which the emotional care of bereaved families is not being provided adequately.

[1044] 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.

[1045] In this invention, the server includes: means for collecting communication data of the deceased; means for analyzing and preprocessing the communication data; means for training an AI model based on the preprocessed data; means for training the AI ​​model using a natural language processing model, a voice synthesis model, and an image / video analysis model; means for integrating the trained models and linking the data so that the words, voice, and facial expressions match; means for packaging the integrated AI character and transmitting it to a user's device; means for providing a digital memorial portrait device that displays the AI ​​character; and means for receiving voice input from the user using the digital memorial portrait device and generating and displaying a response using the AI ​​character. This makes it possible to reproduce the appearance, voice, and speech of the deceased with high accuracy, allowing bereaved family members to naturally converse and interact with the deceased.

[1046] "Communication data" refers to information such as message history, call records, images, and videos left by the deceased.

[1047] "Preprocessing" refers to the process of removing unnecessary information from the received data and extracting key phrases, important topics, etc.

[1048] "AI model" refers to an artificial intelligence model that analyzes and learns from text data, audio data, image and video data.

[1049] "Natural language processing model" refers to an AI model used to understand and generate human language.

[1050] "Speech synthesis model" refers to an AI model used to convert text data into speech and reproduce specific speech characteristics.

[1051] "Image and video analysis model" refers to an AI model used to analyze image and video data and extract and reproduce specific facial expressions and gestures.

[1052] "Data integration" refers to the process of integrating the output of each trained AI model to generate consistent words, voices, and facial expressions.

[1053] "Packaging" refers to the process of converting the integrated AI character into a data format that can be used on the user's device.

[1054] "Digital memorial device" refers to a device that can display an AI character and generate and display responses based on voice input.

[1055] To implement this invention, it is necessary to build a system that collects communication data of the deceased, generates an AI model, and uses it as a digital memorial portrait. This system uses various hardware and software to accurately reproduce the deceased's speech, voice, and facial expressions. Specific examples are shown below.

[1056] System Overview

[1057] This system collects communication data from the deceased, preprocesses it, trains an AI model, and displays the generated AI character as a digital portrait. Furthermore, when the user speaks to this digital portrait, it can respond in a way that is true to the deceased's personality.

[1058] Hardware and software used

[1059] Hardware: User's smartphone or PC, cloud server, microphone, dedicated display.

[1060] Software: Messaging applications, data transmission interface applications, cloud storage (e.g., Amazon S3), Python data processing libraries (e.g., spaCy, NLTK), speech analysis libraries (e.g., Librosa), natural language processing models (e.g., BERT, GPT-3), speech synthesis models (e.g., Tacotron2, WaveNet), image and video analysis libraries (e.g., OpenCV, Dlib), machine learning frameworks (e.g., TensorFlow, PyTorch).

[1061] Program processing flow

[1062] 1. Data collection phase:

[1063] Users log into messaging apps using their smartphones or PCs and export chat history, call logs, photos, and videos.

[1064] The device provides an interface for sending the exported data to the cloud server, for example, by clicking a "Data Upload" button in a dedicated application.

[1065] 2. Data preprocessing phase:

[1066] The server analyzes the communication data stored in the cloud storage, first filtering out spam messages and unwanted data using a Python script.

[1067] Next, we use an NLP library (e.g., spaCy) to extract key phrases and important topics, and then use topic modeling techniques (e.g., LDA) to cluster frequently occurring vocabulary.

[1068] 3. AI model training phase:

[1069] The server inputs the preprocessed text data into the BERT model, allowing it to learn the unique phrases and vocabulary of the deceased.

[1070] For the audio data, we use the Librosa library to remove noise and extract clear speech, and then use a voice synthesis model (e.g., Tacotron2) to recreate the voice quality of the deceased.

[1071] Image and video data is analyzed using OpenCV and Dlib, and the system is trained to recognize the facial expressions and gestures of the deceased.

[1072] 4. Model integration phase:

[1073] The server integrates each trained model (natural language processing model, speech synthesis model, image and video analysis model) and uses machine learning frameworks such as TensorFlow to coordinate the data so that the integrated AI character can generate consistent words, voices, and facial expressions.

[1074] 5. Digital portrait generation phase:

[1075] The server then sends the integrated AI character to the user's device, where the user can view the digital portrait using a dedicated display or smartphone.

[1076] 6. Dialogue Phase:

[1077] The user talks to the digital portrait, asking questions such as, "Hello, Grandpa. How was your day?"

[1078] The terminal captures the user's voice with a microphone and transmits the voice data to the server in real time.

[1079] The server uses speech recognition technology (e.g., Google Speech-to-Text API) to convert the received voice into text and understand the content of the question.

[1080] Next, an NLP model is used to generate appropriate responses, and a speech synthesis model generates the responses in the voice of the deceased.

[1081] The device plays back the response received from the server and also displays animations that reproduce the facial expressions and gestures of the deceased.

[1082] Specific examples

[1083] For example, if a user says, "Hello, Grandpa. How was your day?", the process is as follows:

[1084] 1. User: Speaks a question.

[1085] 2. Device: Captures audio and sends it to the server.

[1086] 3. Server: Analyzes the voice and understands the question.

[1087] 4. Server: Generates a response using an NLP model, for example, "Hello, it was a beautiful day today. How was it for you?"

[1088] 5. Server: Generate a response in the voice of the deceased person using a speech synthesis model.

[1089] 6. Terminal: Plays back the response and reproduces the facial expression of the deceased.

[1090] Prompt Sentence Examples

[1091] "Tell me some recent memories of your grandpa."

[1092] "Tell me about your grandpa's hobbies."

[1093] "Where was your last trip?"

[1094] This allows the user to provide emotional care through intimate conversations with the deceased.

[1095] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1096] Step 1:

[1097] Input: User accesses deceased person's communication accounts and exports chat history, call logs, photos, and videos.

[1098] How it works: A user logs into a messaging app using a smartphone or PC, accesses the "Settings" menu, selects the "Export Chat History" option, and exports the relevant data.

[1099] Output: The exported data file.

[1100] Step 2:

[1101] Input: Terminal exported data file.

[1102] Specific operation: The terminal provides an interface for the user to send the exported data to the server. For example, the user clicks the "Upload Data" button in the dedicated application to send the data to the server.

[1103] Output: The data file sent to the server.

[1104] Step 3:

[1105] Input: The server receives the transmitted data file.

[1106] Specific operation: The server stores the received data in cloud storage (e.g., Amazon S3).

[1107] Output: Data stored in cloud storage.

[1108] Step 4:

[1109] Input: Data stored in cloud storage.

[1110] What it does: The server uses Python scripts to filter spam messages and error messages from the stored data, then uses NLP libraries (e.g., spaCy) to extract key phrases and important topics.

[1111] Output: Preprocessed text data.

[1112] Step 5:

[1113] Input: Preprocessed text data.

[1114] How it works: The server inputs the preprocessed text data into the BERT model, allowing it to learn the unique phrases and vocabulary of the deceased.

[1115] Output: A trained natural language processing model.

[1116] Step 6:

[1117] Input: Audio data stored in cloud storage.

[1118] How it works: The server uses the Librosa library to analyze the audio data, remove noise, and extract clear audio.

[1119] Output: Preprocessed audio data.

[1120] Step 7:

[1121] Input: Preprocessed audio data.

[1122] Specific operation: The server uses the Tacotron2 model to train the preprocessed audio data to reproduce the voice quality of the deceased.

[1123] Output: The trained speech synthesis model.

[1124] Step 8:

[1125] Input: Image and video data stored in cloud storage.

[1126] Specific operation: The server uses OpenCV and Dlib to analyze the facial expressions and gestures of the deceased in image and video data.

[1127] Output: Preprocessed image and video data.

[1128] Step 9:

[1129] Input: Preprocessed image and video data.

[1130] Specific operation: The server uses image and video analysis models to learn how to reproduce the facial expressions and movements of the deceased.

[1131] Output: A trained image and video analysis model.

[1132] Step 10:

[1133] Input: Trained natural language processing models, speech synthesis models, and image / video analysis models.

[1134] How it works: The server uses machine learning frameworks such as TensorFlow to integrate each model and generate a single AI character, coordinating verbal, vocal, and facial output to achieve results that are closer to real-life interactions.

[1135] Output: A unified AI character.

[1136] Step 11:

[1137] Input: Integrated AI character.

[1138] Specific operation: The server packages this AI character and sends it to the user's device through a dedicated API.

[1139] Output: The AI ​​character sent to the user's device.

[1140] Step 12:

[1141] Input: A question that the user speaks.

[1142] Specific Actions: The user talks to the digital portrait, asking questions such as, "Hello, Grandpa. How was your day?"

[1143] Output: The audio data captured on the device.

[1144] Step 13:

[1145] Input: Audio data captured on the device.

[1146] How it works: The device sends the captured voice data to the server, which uses the Google Speech-to-Text API to convert the voice data into text and understand the question.

[1147] Output: Parsed text data.

[1148] Step 14:

[1149] Input: Parsed text data.

[1150] What happens: The server uses the NLP model to generate an appropriate response, such as "Hello, it was a beautiful day today. How was it for you?"

[1151] Output: The generated text response.

[1152] Step 15:

[1153] Input: The generated text response.

[1154] What it does: The server uses the Tacotron2 model to vocalize the text response in the voice of the deceased.

[1155] Output: A spoken response.

[1156] Step 16:

[1157] Input: A spoken response.

[1158] What it does: The device plays back spoken responses and also displays animations that replicate the facial expressions and gestures of the deceased.

[1159] Output: Played voice response and animation.

[1160] (Application example 1)

[1161] 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."

[1162] In recent years, with the advancement of AI technology, there has been an increase in efforts to generate digital memorial portraits using the communication data of the deceased, but conventional technologies have limited interaction capabilities with users and lack security.This invention provides a system that provides real-time response guidance from an AI character that resembles the deceased, and detects and warns abnormal behavior using a score prediction model, thereby achieving a safe and comfortable interaction experience for users.

[1163] 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.

[1164] In this invention, the server includes means for collecting communication data of the deceased, means for analyzing and preprocessing the communication data, means for training an AI model based on the preprocessed data, means for integrating the trained AI model to generate an AI character resembling the deceased, means for providing a digital memorial portrait device that displays the AI ​​character, means for receiving voice input from a user using the digital memorial portrait device and generating and displaying a response using the AI ​​character, and means for detecting abnormal behavior using a score prediction model and providing a real-time audio warning. This allows for a natural conversational experience with the deceased while simultaneously alerting security guards to any abnormal behavior.

[1165] "Communication data of the deceased" refers to a collection of electronic data used by the deceased during their lifetime, and primarily includes chat history, call records, photos, and videos.

[1166] "Preprocessing" refers to the initial stage of data processing, where collected raw data is analyzed, unnecessary information is removed, and important information is extracted.

[1167] An "AI model" is an artificial intelligence program that has been trained to perform a specific task using machine learning or deep learning techniques.

[1168] An "AI character" is a digital character that integrates a trained AI model and reproduces the characteristics of a specific person (in this case, a deceased person).

[1169] A "digital memorial device" is a device that displays an AI character and enables interaction with the user, and can be a smartphone, a dedicated display, or AR-compatible smart glasses.

[1170] "Voice input from the user" refers to voice instructions or questions given by the user to the system via a microphone or the like.

[1171] A "score prediction model" is a model that uses AI technology to analyze data in real time and predict risk scores associated with specific behaviors or situations.

[1172] "Abnormal behavior detection" refers to the process of recognizing and alerting when behavior that deviates from normal patterns occurs.

[1173] "Providing a real-time warning" refers to the act of immediately issuing a warning and notifying the user of detected abnormal behavior.

[1174] This invention provides a system that collects communication data of the deceased, uses it to build an AI model, and uses it to create a digital memorial portrait. The system aims to generate an AI character that resembles the deceased and provides responses that are characteristic of the deceased through conversations with the user. It also has the ability to detect abnormal behavior and provide real-time warnings using a score prediction model.

[1175] System Configuration

[1176] 1. Hardware Configuration

[1177] Server: Responsible for storing and processing data, and training and operating AI models. A server with high-performance processing capabilities is desirable.

[1178] Device: The device used by the user, which may be a smartphone, a dedicated display, or AR-enabled smart glasses.

[1179] Camera: Used to capture footage in real time and analyze it in the score prediction model.

[1180] Microphone: Used to capture the user's voice input.

[1181] 2. Software Configuration

[1182] Data collection module: Collects communication data such as chat history, call records, photos, and videos of the deceased.

[1183] Data preprocessing module: Analyzes the collected data and extracts the necessary information.

[1184] AI model learning module: Based on the preprocessed data, natural language processing (NLP) models, speech synthesis models, image and video analysis models, and score prediction models are trained.

[1185] Integration module: Integrates each model to generate an AI character that resembles the deceased.

[1186] Digital portrait display module: Displays the generated AI character and enables interaction with the user.

[1187] Score prediction module: Detects abnormal behavior in real time based on camera footage and provides necessary warnings via voice.

[1188] Operating Procedure

[1189] 1. Data Collection: Users enter the deceased person's communication data into the system, including chat history, call records, photos, videos, etc.

[1190] 2. Data Preprocessing: The server preprocesses the received data, removing unnecessary information and extracting important information. This process includes filtering spam messages and extracting key phrases.

[1191] 3. AI model training: The server uses the preprocessed data to train the NLP model, speech synthesis model, image and video analysis model, and score prediction model, thereby acquiring knowledge to reproduce the deceased's unique speech patterns, voice quality, facial expressions, etc.

[1192] 4. Model integration: The output of each model is integrated to generate an AI character that resembles the deceased person, with matching language, voice, and facial expressions.

[1193] 5. Interactive function: The user can talk to the digital portrait, for example, asking questions like, "How was your day?" The device captures this voice and sends it to the server.

[1194] 6. Real-time warning: Camera footage is analyzed in real time, and if abnormal behavior is detected, a warning is immediately given to the user, such as an audio warning saying "Warning! Intruder detected."

[1195] Specific examples

[1196] When a user speaks to a digital portrait and asks, "Hello, how was your day?", the response is generated through the following process:

[1197] 1. User: Speaks a question.

[1198] 2. Device: Captures audio and sends it to the server.

[1199] 3. Server: Analyzes the voice and understands the question.

[1200] 4. Server: Generates a response using an NLP model, such as "It was a beautiful day today. How was it for you?"

[1201] 5. Server: Generates a response in the voice of the deceased person using a speech synthesis model and sends it to the device.

[1202] 6. Terminal: Plays back the response and reproduces the facial expression of the deceased.

[1203] Prompt Sentence Examples

[1204] Example prompt sentence:

[1205] "Warning! Unauthorized person loitering near conference room. Please investigate immediately."

[1206] "Attention! Abnormal behavior detected. Please respond immediately."

[1207] This allows users to have a natural conversation with the deceased while also supporting immediate response as a security guard.

[1208] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1209] Step 1:

[1210] Data collection

[1211] The user inputs the deceased's communication data (chat history, call records, photos, videos, etc.) into the system. The device exports this data and sends it to the server. The server receives and stores the data.

[1212] Input: Chat history, call logs, photos, videos

[1213] Output: Saved communication data

[1214] Step 2:

[1215] Data Preprocessing

[1216] The server analyzes the received communication data, removes unnecessary information (spam messages and errors), and extracts important information (key phrases, frequently occurring vocabulary, and important topics). This preprocessing process prepares the data for AI model training.

[1217] Input: Saved communication data

[1218] Output: Preprocessed data

[1219] Step 3:

[1220] Training an AI model

[1221] The server uses the preprocessed data to train various AI models (NLP model, voice synthesis model, image / video analysis model, score prediction model), which then learns the deceased's speech patterns, voice quality, facial expressions, and behavioral patterns.

[1222] Input: Preprocessed data

[1223] Output: Trained AI model (NLP model, speech synthesis model, image / video analysis model, score prediction model)

[1224] Step 4:

[1225] Model Integration

[1226] The server then integrates each trained AI model to generate an AI character that resembles the deceased. The integration process involves linking the outputs of each model and adjusting the words, voice, and facial expressions to match.

[1227] Input: Trained AI model

[1228] Output: Integrated AI character

[1229] Step 5:

[1230] Displaying a digital portrait of the deceased

[1231] The server packages the integrated AI character and sends it to the terminal, which uses the received data to display the AI ​​character on the digital portrait device.

[1232] Input: Integrated AI character

[1233] Output: Displayed AI character

[1234] Step 6:

[1235] User interaction

[1236] The user speaks to the digital portrait. The device captures the voice with a microphone and sends the captured voice data to the server. The server uses speech recognition technology to analyze and understand the question. It uses an NLP model to generate an appropriate response and a speech synthesis model to generate a response in the deceased's voice. Finally, the response is sent to the device, which plays it back. In some cases, the deceased's facial expressions and gestures are also reproduced.

[1237] Input: User's voice

[1238] Output: Vocal responses, facial expressions, and gestures of the deceased

[1239] Step 7:

[1240] Abnormal behavior detection and real-time alerts

[1241] The device's camera captures video in real time. The server analyzes the video using a score prediction model, and if abnormal behavior is detected, it immediately provides an audio warning. For example, it notifies the user with a voice message saying, "Warning! Intruder detected."

[1242] Input: Real-time video

[1243] Output: Audio warning, abnormal behavior detection results

[1244] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1245] To implement this invention, we will create an AI model using the communication data of the deceased and build a system that can be used as a digital memorial portrait. This system will also incorporate an emotion engine that recognizes the user's emotions and provides responses according to their emotions.

[1246] System Overview

[1247] The system collects communication data from the deceased, trains an AI model using preprocessed data, and displays the generated AI character as a digital portrait. Furthermore, when a user speaks to the digital portrait, it uses an emotion engine to analyze the emotions and provide appropriate responses.

[1248] Program processing flow

[1249] 1. Data Collection Phase

[1250] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[1251] The device uploads the exported data to the server in the specified format (e.g., JSON or CSV file).

[1252] 2. Data preprocessing phase

[1253] The server stores the received communication data and performs pre-processing on the data, which includes removing unnecessary information (such as spam messages and errors) and extracting important information (key phrases, important topics, and frequently occurring vocabulary).

[1254] 3. AI model training phase

[1255] The server uses the preprocessed data to train natural language processing (NLP) models, speech synthesis models, and image and video analysis models. Each model works as follows:

[1256] NLP model: Analyzes text data and learns the words and phrases of the deceased.

[1257] Speech synthesis model: Extracts clear speech and trains to reproduce the voice quality of the deceased.

[1258] Image and video analysis model: Learns facial expressions and gestures and reproduces the facial expressions and movements of the deceased.

[1259] 4. Model integration phase

[1260] The server then integrates the trained models to operate as a single AI character, a process that involves linking the outputs of each model and ensuring that the words, voices, and facial expressions match.

[1261] 5. Digital portrait generation phase

[1262] The server packages the integrated AI character and sends it to the user's device, where it can be displayed on a digital memorial device (such as a dedicated display or smartphone).

[1263] 6. Emotion recognition and response phase

[1264] The emotion engine analyzes the user's voice data and facial expression data to recognize the user's emotions (e.g., happiness, sadness, anger, surprise, etc.).

[1265] The server adapts the NLP model and speech synthesis model based on the output of the emotion engine, allowing the AI ​​character of the deceased person to generate an appropriate response based on the recognized emotion.

[1266] The server also adjusts facial expressions and tone of voice according to emotions to recreate natural conversations.

[1267] Specific examples

[1268] For example, if a user says to a digital portrait, "Hello, Grandpa. I've been feeling a bit sad today," the following happens:

[1269] 1. User: Speaks a question.

[1270] 2. Device: Captures audio and sends it to the server.

[1271] 3. Server: Analyzes the voice and understands the question.

[1272] 4. Emotion engine: Analyzes the user's voice and facial expressions to recognize when the user is feeling "sad."

[1273] 5. Server: Uses NLP models to generate responses based on emotions, such as "Hi, I'm very worried to hear that you're feeling sad. What happened?"

[1274] 6. Server: Using a speech synthesis model, the generated response is generated in the voice of the deceased.

[1275] 7. Server: Based on the emotion engine, the tone of voice and facial expressions are adjusted to interact with the user in a more natural way.

[1276] 8. The device plays back the response received from the server and reproduces the facial expression of the deceased.

[1277] In this way, the system recognizes the user's emotions and allows them to provide emotional care through dialogue with the deceased.

[1278] The processing flow will be explained below.

[1279] Step 1:

[1280] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[1281] Step 2:

[1282] The device uploads the exported data to the server in the specified format (e.g., JSON or CSV file).

[1283] Step 3:

[1284] The server receives the uploaded communication data and stores it in a secure database.

[1285] Step 4:

[1286] The server analyzes the stored data, removes unnecessary information (such as spam messages and errors), and extracts important information (key phrases, important topics, and frequently occurring vocabulary).

[1287] Step 5:

[1288] The server uses the preprocessed data to train a natural language processing (NLP) model, which analyzes and trains the text data to memorize the words and phrases used by the deceased.

[1289] Step 6:

[1290] The server uses the audio data to train a speech synthesis model, extracting clear speech and training it to reproduce the voice quality of the deceased person.

[1291] Step 7:

[1292] The server uses image and video data to train an image and video analysis model, analyzing the facial expressions and gestures of the deceased.

[1293] Step 8:

[1294] The server integrates the trained NLP model, speech synthesis model, and image / video analysis model, allowing it to operate as a single AI character.

[1295] Step 9:

[1296] The server packages the integrated AI character and sends it to the user's device.

[1297] Step 10:

[1298] Users can talk to the AI ​​character, which is displayed as a digital portrait of the deceased, asking questions such as, "How was your day?" or "Tell me about an old trip?"

[1299] Step 11:

[1300] The device captures the user's voice with a microphone and transmits the voice data to the server.

[1301] Step 12:

[1302] The server analyzes the received voice using automatic speech recognition (ASR) technology and understands the content of the question.

[1303] Step 13:

[1304] The server uses an emotion engine to analyze the user's voice data and facial expression data and recognize the user's emotions (e.g., happiness, sadness, anger, surprise, etc.).

[1305] Step 14:

[1306] The server uses NLP models to generate appropriate responses based on the recognized emotions, for example, if the user is expressing sadness, it generates a response such as "Hi, I'm very worried to hear that you're feeling sad. What happened?"

[1307] Step 15:

[1308] The server uses a speech synthesis model to generate the generated response in the voice of the deceased person, adjusting the tone of the voice based on the output of the emotion engine.

[1309] Step 16:

[1310] The server sends the generated response to the device, adjusting facial expressions and gestures as needed.

[1311] Step 17:

[1312] The device then plays back the responses received from the server and reproduces the facial expressions of the deceased, allowing the user to have a natural conversation.

[1313] Example 2

[1314] 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."

[1315] It is necessary to utilize the communication data of the deceased to create a digital memorial portrait that reflects the deceased's characteristics, and to realize a dialogue between the user and the deceased's AI character based on emotion recognition. However, conventional technologies have low accuracy in analyzing communication data and recognizing emotions, making it difficult to provide natural responses that correspond to the user's emotions. Another issue is the difficulty of integrating various models (e.g., natural language processing, speech synthesis, image and video analysis) to accurately reproduce the characteristics of the deceased.

[1316] 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.

[1317] In this invention, the server includes means for collecting communication data of the deceased, means for analyzing and preprocessing the communication data, means for training an AI model based on the preprocessed data, means for integrating the trained AI model to generate an AI character resembling the deceased, means for providing a display device for displaying the AI ​​character, means for receiving voice input from a user via the display device and generating and displaying a response using the AI ​​character, and means for recognizing the user's emotions and generating a response according to the emotion. This makes it possible to accurately reproduce the characteristics of the deceased while providing a natural and appropriate response according to the user's emotions.

[1318] "Communication data" refers to records of digital communications used by the deceased during their lifetime, such as chat history, call records, photos, and videos.

[1319] "Preprocessing" is the process of removing unnecessary information from the received data and extracting important information (such as key phrases and frequently occurring vocabulary).

[1320] An "AI model" is a collection of algorithms that use artificial intelligence techniques to analyze and learn from data and automate specific tasks.

[1321] A "natural language processing model" is a model that uses artificial intelligence technology to analyze human language and understand its meaning.

[1322] A "speech synthesis model" is a model that uses artificial intelligence technology to convert text data into speech.

[1323] The "image and video analysis model" is a model that uses artificial intelligence technology to analyze image and video data and understand its content.

[1324] An "AI character" is a digital character generated based on an AI model that reproduces the features of a deceased person.

[1325] A "display device" is a device for visually displaying an AI character to a user.

[1326] "Emotion recognition" is a technology that analyzes a user's voice data and facial expression data to identify their emotions.

[1327] To implement this invention, it is necessary to build a system that uses the communication data of the deceased to generate an AI model and use it as a digital memorial portrait. This system is equipped with an emotion engine that recognizes the user's emotions and provides responses according to the emotions.

[1328] Hardware and software used

[1329] Hardware

[1330] Server: A server for storing and analyzing data and training AI models. Use a cloud server (e.g., Amazon EC2) equipped with a high-performance processor and large memory capacity.

[1331] Terminal: A device that allows users to operate and input data. This includes smartphones, tablets, and PCs.

[1332] Display device: A device for displaying the digital portrait. A dedicated display or smartphone can be used.

[1333] software

[1334] Natural Language Processing (NLP) model: A model for analyzing text data and learning the vocabulary and phrasing of the deceased, using NLP techniques such as BERT (Bidirectional Encoder Representations from Transformers).

[1335] Speech synthesis model: A model for converting text to speech and recreating the voice quality of the deceased. Uses voice synthesis technology such as Tacotron 2.

[1336] Image and video analysis model: A model that analyzes facial expressions and gestures to recreate the movements and expressions of the deceased. It uses image and video analysis technologies such as OpenPose and Face++.

[1337] Emotion engine: A technology that analyzes the user's voice data and facial expression data to recognize the user's emotions. It uses a combination of DeepSpeech (voice recognition) and Face++ (facial expression recognition).

[1338] Specific examples

[1339] Data Collection Phase

[1340] The user exports the chat history from the deceased person's messaging app (e.g., WhatsApp), and then uploads it to the server using a dedicated app.

[1341] Data Preprocessing Phase

[1342] The server analyzes the received chat history, filters out unnecessary information, and then extracts important key phrases and frequently occurring vocabulary.

[1343] AI model training phase

[1344] The server uses the preprocessed data to train a BERT model to learn the deceased's words and phrases, then trains the model with Tacotron 2 on clear audio data to reproduce the deceased's voice, and trains the model with OpenPose and Face++ on facial expressions and gestures.

[1345] Model integration phase

[1346] The server integrates each model (natural language processing model, speech synthesis model, image / video analysis model) to generate an AI character that recreates the deceased person.

[1347] Usage example

[1348] For example, if a user says to a digital portrait, "Hello, Grandpa. I've been feeling a bit sad today," the following happens:

[1349] The user enters the question by voice.

[1350] The device captures the audio and sends it to the server.

[1351] The server analyzes the voice and understands the question.

[1352] The emotion engine analyzes the user's voice and facial expressions and recognizes that the user is feeling "sad."

[1353] The server uses an NLP model to generate a response such as, "Hello, I'm very worried to hear that you're feeling sad. What happened?"

[1354] The server uses a speech synthesis model to generate responses in the voice of the deceased, adjusting the tone of voice and facial expression depending on the emotion.

[1355] The device plays back the response received from the server and reproduces the facial expression of the deceased.

[1356] Prompt Sentence Examples

[1357] "Analyze the message data of the deceased and generate an AI model for natural conversation."

[1358] "Develop a system that recognizes and responds to user emotions. The system will use an emotion engine to engage in appropriate dialogue based on the user's emotions."

[1359] This allows the system to recognize the user's emotions and provide emotional care through dialogue with the deceased.

[1360] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1361] Step 1: Data collection

[1362] The user accesses the deceased person's messaging and social media accounts and exports chat history, call logs, photos, and videos (input: the deceased person's communication data).

[1363] The terminal uploads the exported data to the server in the specified format (e.g., JSON file or CSV file) (output: uploaded communication data).

[1364] Step 2: Data storage and preprocessing

[1365] The server receives the data sent from the terminal and stores it securely (input: uploaded communication data, output: stored communication data).

[1366] The server removes unnecessary information from the received data (specific operations: filtering spam messages and error logs) and extracts important key phrases and frequently occurring vocabulary (input: saved communication data, output: preprocessed data).

[1367] Step 3: Training the Natural Language Processing (NLP) Model

[1368] The server uses the preprocessed text data to train an NLP model (e.g., BERT) (input: preprocessed data, output: trained NLP model). Specifically, it tokenizes the raw data and trains the model to understand the structure of the text.

[1369] Step 4: Training the speech synthesis model

[1370] The server extracts the preprocessed voice data and trains it using a speech synthesis model (e.g., Tacotron 2) (input: preprocessed voice data, output: trained speech synthesis model). Specifically, it analyzes the features of the voice data and reproduces the voice quality of the deceased.

[1371] Step 5: Training the image and video analysis model

[1372] The server uses the preprocessed image and video data to train an image and video analysis model (e.g., OpenPose, Face++) (input: preprocessed image and video data, output: trained image and video analysis model). Specifically, it analyzes the facial expressions and gestures of the deceased and reproduces them.

[1373] Step 6: Model integration

[1374] The server integrates each trained model to generate a single AI character (input: trained NLP model, trained speech synthesis model, trained image and video analysis model, output: integrated AI character). Specifically, the text generated by the NLP model is input into the speech synthesis model, and the corresponding facial expression is generated by the image and video analysis model.

[1375] Step 7: Generate a digital portrait

[1376] The server packages the integrated AI character and sends it to the user's device (input: integrated AI character, output: package sent to the user's device).

[1377] The user installs the file sent to their device and displays the AI ​​character on the digital portrait device (input: sent package, output: displayed AI character).

[1378] Step 8: Emotion Recognition and Response Generation

[1379] The emotion engine analyzes the user's voice data and facial expression data and recognizes emotions (input: user's voice data, facial expression data, output: recognized emotions).

[1380] Based on the output of the emotion engine, the server adapts the NLP model and speech synthesis model to generate an appropriate response (input: recognized emotion, output: generated response).

[1381] The server generates the response in the voice of the deceased person and adjusts the tone of voice and facial expression according to the emotion (input: generated response, output: adjusted response and facial expression).

[1382] The device plays back the responses received from the server and reproduces the facial expressions of the deceased (input: adjusted responses and facial expressions, output: reproduced responses and facial expressions).

[1383] This series of processes allows the user to receive emotional care through natural conversation with the deceased.

[1384] (Application example 2)

[1385] 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."

[1386] When applying a digital memorial portrait system to a virtual store, a means of understanding the user's emotions and responding appropriately is required. Another challenge is to increase customer satisfaction by providing a customer service experience that recreates the memories and voice of the deceased.

[1387] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1388] In this invention, the server includes means for collecting communication data of the deceased, means for analyzing and preprocessing the communication data, means for training an AI model based on the preprocessed data, means for integrating the trained AI model to generate an AI character resembling the deceased, means for providing a digital portrait device that displays the AI ​​character, means for receiving voice input from a user using the digital portrait device and generating and displaying a response using the AI ​​character, means for analyzing the user's emotions using an emotion recognition engine, means for generating a response based on the user's emotions, and means for using the digital portrait device as a guide in a virtual store. This makes it possible to grasp the user's emotions and provide effective guidance in the virtual store by reproducing the deceased's responses according to those emotions.

[1389] "Communication data" refers to digital information left behind by the deceased, such as chat history, call records, photos, and videos.

[1390] "Preprocessing" refers to the process of removing unnecessary information from collected communication data and extracting important information.

[1391] An "AI model" is an artificial intelligence that is trained using technologies such as natural language processing, speech synthesis, and image and video analysis to carry out specific tasks.

[1392] An "AI character" is a digital character that resembles a deceased person and is generated by integrating a trained AI model.

[1393] A "digital memorial device" is a device that displays an AI character and allows the user to interact with it.

[1394] "Voice input" refers to the voice data that is generated when the user speaks to the digital memorial portrait device.

[1395] An "emotion recognition engine" is a system for analyzing emotions from a user's voice and images.

[1396] A "virtual store" is a store that operates in a virtual space, a place that offers products and services online.

[1397] A "guide" is a person whose role is to provide customers with product information and guidance within the store.

[1398] A "user" is a person who uses the digital memorial device to interact with an AI character of a deceased person.

[1399] To implement this invention, it is necessary to create an AI model using the communication data of the deceased and build a system that functions as a digital memorial guide in a virtual store. This system reproduces the memories and voice of the deceased and combines it with an emotion recognition engine to provide appropriate responses according to the user's emotions.

[1400] System Configuration

[1401] 1. The user uses a smartphone or smart glasses to talk to a digital portrait guide.

[1402] 2. The device captures the user's voice input and facial image data and sends them to the server.

[1403] 3. The server executes multiple methods to perform the following processes:

[1404] Data collection and preprocessing

[1405] The server collects the deceased's communication data (chat history, call records, photos, videos). The collected data is pre-processed using technologies such as NLP (natural language processing) and image and video analysis to remove unnecessary information and extract important information.

[1406] AI model training and integration

[1407] The server uses the preprocessed data to train natural language processing models, speech synthesis models, and image and video analysis models. These trained models are then integrated to generate an AI character that resembles the deceased. This AI character reproduces the deceased's speech, voice quality, facial expressions, etc., allowing users to interact with the deceased in a virtual store.

[1408] Emotion Recognition and Response Generation

[1409] The server is equipped with an emotion recognition engine that analyzes the user's emotions from their voice and facial expression data. Based on the emotion engine's analysis results, an NLP model generates an appropriate text response, and a speech synthesis model reproduces that text as the deceased's voice. This enables natural conversations that correspond to the user's emotions.

[1410] Use as a digital memorial guide

[1411] In the virtual store, users can interact with a digital memorial guide through their smartphones or smart glasses. For example, if a customer asks, "I'd like to know more about our new products," the server will analyze their voice and facial expression data and provide an answer. Specifically, it will respond with something like, "Of course. Here are our latest products, and their features are as follows."

[1412] Prompt Sentence Examples

[1413] An example prompt for generating an AI model is:

[1414] You are a digital memorial guide working in a virtual store. Respond to customers' questions and interests in a friendly and courteous manner. Strive to provide relevant and useful information while taking into consideration the customer's feelings. For example, if a customer asks, "I'd like to know more about our new products," you can respond, "Of course! Here are our latest products and their features," and then provide specific product information.

[1415] As described above, the present invention makes it possible to utilize a digital portrait of the deceased as a guide in a virtual store while being sensitive to the user's emotions.

[1416] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1417] Step 1:

[1418] The user speaks to the digital portrait guide through a smartphone or smart glasses. The user inputs questions by voice, and simultaneously captures facial image data. This input data is captured by the device, which then transmits the voice data and facial image data to a server.

[1419] Step 2:

[1420] The server analyzes the received voice data. Specifically, it converts it into text data using voice recognition technology (such as Google Speech API). At this time, voice data is input, and as a result, text data containing the user's question is output.

[1421] Step 3:

[1422] The server sends the received facial image data to an emotion recognition engine to analyze the user's emotions. The emotion recognition engine uses image analysis technology (such as OpenCV or Dlib) to analyze emotions from facial expressions. The input is facial image data, and the output is data indicating the user's emotions (e.g., happiness, sadness, anger, etc.).

[1423] Step 4:

[1424] The server uses the analyzed text data and emotion data to work with an NLP model to generate an appropriate text response. Specifically, NLP techniques are applied to the text data to generate candidate responses. The emotion data is used to adjust the response. The input is text data containing the question and emotion data, and the output is the generated appropriate response text.

[1425] Step 5:

[1426] The server sends the generated response text to a speech synthesis model, which generates audio data in the voice of the deceased. A speech synthesis model (such as Amazon Polly or Google Text-to-Speech) is used. The input is the response text, and the output is audio data that reproduces the voice of the deceased.

[1427] Step 6:

[1428] The server sends the generated voice data to the terminal, which then plays the received voice data to the user, allowing the user to interact with the digital portrait guide. The input is the voice data, and the output is the played voice response.

[1429] Step 7:

[1430] After the server or terminal has completed responding to the user's question, it waits for the next question or interaction. If the user has any new input, it returns to step 1.

[1431] Through the above processing steps, the user can have natural conversations with the digital portrait guide and is provided with guidance services in the virtual store.

[1432] 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.

[1433] 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.

[1434] 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.

[1435] [Fourth embodiment]

[1436] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1437] 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.

[1438] 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).

[1439] 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.

[1440] 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.

[1441] 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).

[1442] 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.

[1443] 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.

[1444] 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.

[1445] 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.

[1446] 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.

[1447] 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.

[1448] 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."

[1449] To implement this invention, we will build a system that uses the communication data of the deceased to generate an AI model and use it as a digital memorial portrait. The following explains the program processing of this system in natural language.

[1450] System Overview

[1451] The system collects communication data from the deceased, trains an AI model using preprocessed data, and displays the generated AI character as a digital portrait. Furthermore, users can talk to this digital portrait and it responds in a way that is true to the deceased's personality.

[1452] Program processing flow

[1453] 1. Data Collection Phase

[1454] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[1455] The terminal provides an interface for sending the exported data to the server.

[1456] 2. Data preprocessing phase

[1457] The server stores the received communication data and performs pre-processing on the data, which includes removing unnecessary information (e.g., spam messages, errors) and extracting important information (e.g., key phrases, important topics, and frequently occurring vocabulary).

[1458] 3. AI model training phase

[1459] The server uses the preprocessed data to train natural language processing (NLP) models, speech synthesis models, and image and video analysis models. Each model works as follows:

[1460] NLP model: Analyzes text data and learns the words and phrases of the deceased.

[1461] Speech synthesis model: Extracts clear speech and trains to reproduce the voice quality of the deceased.

[1462] Image and video analysis model: Learns facial expressions and gestures and reproduces the facial expressions and movements of the deceased.

[1463] 4. Model integration phase

[1464] The server then integrates the trained models to operate as a single AI character, a process that involves linking the outputs of each model and ensuring that the words, voices, and facial expressions match.

[1465] 5. Digital portrait generation phase

[1466] The server packages the integrated AI character and sends it to the user's device, where it can be displayed on a digital memorial device (such as a dedicated display or smartphone).

[1467] 6. Dialogue Phase

[1468] Users can talk to the digital portrait, asking questions such as "How was your day?" or "Can you tell me about an old memory?"

[1469] The device captures the user's voice with a microphone and transmits the voice data to the server.

[1470] The server analyzes the received voice using automatic speech recognition (ASR) technology and understands the content of the question.

[1471] The server uses an NLP model to generate appropriate responses and a speech synthesis model to generate responses in the voice of the deceased.

[1472] The device plays back the responses received from the server, reproducing facial expressions and gestures as needed.

[1473] Specific examples

[1474] For example, if a user speaks to a digital portrait and says, "Hello, Grandpa. How was your day?", the following happens:

[1475] 1. User: Speaks a question.

[1476] 2. Device: Captures audio and sends it to the server.

[1477] 3. Server: Analyzes the voice and understands the question.

[1478] 4. Server: Generates a response using an NLP model, such as "Hello, it was a beautiful day today. How was it for you?"

[1479] 5. Server: Generate a response in the voice of the deceased person using a speech synthesis model.

[1480] 6. Terminal: Plays back the response and reproduces the facial expression of the deceased.

[1481] In this way, the system allows the bereaved to reconnect with the deceased and provide emotional care.

[1482] The processing flow will be explained below.

[1483] Step 1:

[1484] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[1485] Step 2:

[1486] The device uploads the exported data to the server in the specified format (e.g., JSON or CSV file).

[1487] Step 3:

[1488] The server receives the uploaded communication data and stores it in a secure database.

[1489] Step 4:

[1490] The server analyzes the stored data, removes unnecessary information (such as spam messages and errors), and extracts important information (key phrases, important topics, and frequently occurring vocabulary).

[1491] Step 5:

[1492] The server uses the preprocessed data to train a natural language processing (NLP) model, which analyzes and trains the text data to memorize the words and phrases used by the deceased.

[1493] Step 6:

[1494] The server uses the audio data to train a speech synthesis model, extracting clear speech and training it to reproduce the voice quality of the deceased person.

[1495] Step 7:

[1496] The server uses image and video data to train an image and video analysis model, analyzing the facial expressions and gestures of the deceased.

[1497] Step 8:

[1498] The server integrates the trained NLP model, speech synthesis model, and image / video analysis model, allowing it to operate as a single AI character.

[1499] Step 9:

[1500] The server packages the integrated AI character and sends it to the user's device.

[1501] Step 10:

[1502] Users can talk to the AI ​​character, which is displayed as a digital portrait of the deceased, asking questions such as, "How was your day?" or "Tell me about an old trip?"

[1503] Step 11:

[1504] The device captures the user's voice with a microphone and transmits the voice data to the server.

[1505] Step 12:

[1506] The server analyzes the received voice using automatic speech recognition (ASR) technology and understands the content of the question.

[1507] Step 13:

[1508] The server uses NLP models to generate appropriate responses, such as "It was a beautiful day today. How was it for you?" in response to a user question.

[1509] Step 14:

[1510] The server uses a speech synthesis model to generate the generated response in the voice of the deceased person.

[1511] Step 15:

[1512] The device plays back the responses received from the server, reproducing facial expressions and gestures as needed.

[1513] Example 1

[1514] 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."

[1515] In order to maintain the connection between memories and emotions of the deceased, there is a need for a system that allows bereaved families to reconnect with and interact with the deceased. However, existing technologies are not sufficient to reproduce the voice, facial expressions, and language of the deceased, and there are issues with a lack of realism and the complexity of operation. This means that the current situation is one in which the emotional care of bereaved families is not being provided adequately.

[1516] 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.

[1517] In this invention, the server includes: means for collecting communication data of the deceased; means for analyzing and preprocessing the communication data; means for training an AI model based on the preprocessed data; means for training the AI ​​model using a natural language processing model, a voice synthesis model, and an image / video analysis model; means for integrating the trained models and linking the data so that the words, voice, and facial expressions match; means for packaging the integrated AI character and transmitting it to a user's device; means for providing a digital memorial portrait device that displays the AI ​​character; and means for receiving voice input from the user using the digital memorial portrait device and generating and displaying a response using the AI ​​character. This makes it possible to reproduce the appearance, voice, and speech of the deceased with high accuracy, allowing bereaved family members to naturally converse and interact with the deceased.

[1518] "Communication data" refers to information such as message history, call records, images, and videos left by the deceased.

[1519] "Preprocessing" refers to the process of removing unnecessary information from the received data and extracting key phrases, important topics, etc.

[1520] "AI model" refers to an artificial intelligence model that analyzes and learns from text data, audio data, image and video data.

[1521] "Natural language processing model" refers to an AI model used to understand and generate human language.

[1522] "Speech synthesis model" refers to an AI model used to convert text data into speech and reproduce specific speech characteristics.

[1523] "Image and video analysis model" refers to an AI model used to analyze image and video data and extract and reproduce specific facial expressions and gestures.

[1524] "Data integration" refers to the process of integrating the output of each trained AI model to generate consistent words, voices, and facial expressions.

[1525] "Packaging" refers to the process of converting the integrated AI character into a data format that can be used on the user's device.

[1526] "Digital memorial device" refers to a device that can display an AI character and generate and display responses based on voice input.

[1527] To implement this invention, it is necessary to build a system that collects communication data of the deceased, generates an AI model, and uses it as a digital memorial portrait. This system uses various hardware and software to accurately reproduce the deceased's speech, voice, and facial expressions. Specific examples are shown below.

[1528] System Overview

[1529] This system collects communication data from the deceased, preprocesses it, trains an AI model, and displays the generated AI character as a digital portrait. Furthermore, when the user speaks to this digital portrait, it can respond in a way that is true to the deceased's personality.

[1530] Hardware and software used

[1531] Hardware: User's smartphone or PC, cloud server, microphone, dedicated display.

[1532] Software: Messaging applications, data transmission interface applications, cloud storage (e.g., Amazon S3), Python data processing libraries (e.g., spaCy, NLTK), speech analysis libraries (e.g., Librosa), natural language processing models (e.g., BERT, GPT-3), speech synthesis models (e.g., Tacotron2, WaveNet), image and video analysis libraries (e.g., OpenCV, Dlib), machine learning frameworks (e.g., TensorFlow, PyTorch).

[1533] Program processing flow

[1534] 1. Data collection phase:

[1535] Users log into messaging apps using their smartphones or PCs and export chat history, call logs, photos, and videos.

[1536] The device provides an interface for sending the exported data to the cloud server, for example, by clicking a "Data Upload" button in a dedicated application.

[1537] 2. Data preprocessing phase:

[1538] The server analyzes the communication data stored in the cloud storage, first filtering out spam messages and unwanted data using a Python script.

[1539] Next, we use an NLP library (e.g., spaCy) to extract key phrases and important topics, and then use topic modeling techniques (e.g., LDA) to cluster frequently occurring vocabulary.

[1540] 3. AI model training phase:

[1541] The server inputs the preprocessed text data into the BERT model, allowing it to learn the unique phrases and vocabulary of the deceased.

[1542] For the audio data, we use the Librosa library to remove noise and extract clear speech, and then use a voice synthesis model (e.g., Tacotron2) to recreate the voice quality of the deceased.

[1543] Image and video data is analyzed using OpenCV and Dlib, and the system is trained to recognize the facial expressions and gestures of the deceased.

[1544] 4. Model integration phase:

[1545] The server integrates each trained model (natural language processing model, speech synthesis model, image and video analysis model) and uses machine learning frameworks such as TensorFlow to coordinate the data so that the integrated AI character can generate consistent words, voices, and facial expressions.

[1546] 5. Digital portrait generation phase:

[1547] The server then sends the integrated AI character to the user's device, where the user can view the digital portrait using a dedicated display or smartphone.

[1548] 6. Dialogue Phase:

[1549] The user talks to the digital portrait, asking questions such as, "Hello, Grandpa. How was your day?"

[1550] The terminal captures the user's voice with a microphone and transmits the voice data to the server in real time.

[1551] The server uses speech recognition technology (e.g., Google Speech-to-Text API) to convert the received voice into text and understand the content of the question.

[1552] Next, an NLP model is used to generate appropriate responses, and a speech synthesis model generates the responses in the voice of the deceased.

[1553] The device plays back the response received from the server and also displays animations that reproduce the facial expressions and gestures of the deceased.

[1554] Specific examples

[1555] For example, if a user says, "Hello, Grandpa. How was your day?", the process is as follows:

[1556] 1. User: Speaks a question.

[1557] 2. Device: Captures audio and sends it to the server.

[1558] 3. Server: Analyzes the voice and understands the question.

[1559] 4. Server: Generates a response using an NLP model, for example, "Hello, it was a beautiful day today. How was it for you?"

[1560] 5. Server: Generate a response in the voice of the deceased person using a speech synthesis model.

[1561] 6. Terminal: Plays back the response and reproduces the facial expression of the deceased.

[1562] Prompt Sentence Examples

[1563] "Tell me some recent memories of your grandpa."

[1564] "Tell me about your grandpa's hobbies."

[1565] "Where was your last trip?"

[1566] This allows the user to provide emotional care through intimate conversations with the deceased.

[1567] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1568] Step 1:

[1569] Input: User accesses deceased person's communication accounts and exports chat history, call logs, photos, and videos.

[1570] How it works: A user logs into a messaging app using a smartphone or PC, accesses the "Settings" menu, selects the "Export Chat History" option, and exports the relevant data.

[1571] Output: The exported data file.

[1572] Step 2:

[1573] Input: Terminal exported data file.

[1574] Specific operation: The terminal provides an interface for the user to send the exported data to the server. For example, the user clicks the "Upload Data" button in the dedicated application to send the data to the server.

[1575] Output: The data file sent to the server.

[1576] Step 3:

[1577] Input: The server receives the transmitted data file.

[1578] Specific operation: The server stores the received data in cloud storage (e.g., Amazon S3).

[1579] Output: Data stored in cloud storage.

[1580] Step 4:

[1581] Input: Data stored in cloud storage.

[1582] What it does: The server uses Python scripts to filter spam messages and error messages from the stored data, then uses NLP libraries (e.g., spaCy) to extract key phrases and important topics.

[1583] Output: Preprocessed text data.

[1584] Step 5:

[1585] Input: Preprocessed text data.

[1586] How it works: The server inputs the preprocessed text data into the BERT model, allowing it to learn the unique phrases and vocabulary of the deceased.

[1587] Output: A trained natural language processing model.

[1588] Step 6:

[1589] Input: Audio data stored in cloud storage.

[1590] How it works: The server uses the Librosa library to analyze the audio data, remove noise, and extract clear audio.

[1591] Output: Preprocessed audio data.

[1592] Step 7:

[1593] Input: Preprocessed audio data.

[1594] Specific operation: The server uses the Tacotron2 model to train the preprocessed audio data to reproduce the voice quality of the deceased.

[1595] Output: The trained speech synthesis model.

[1596] Step 8:

[1597] Input: Image and video data stored in cloud storage.

[1598] Specific operation: The server uses OpenCV and Dlib to analyze the facial expressions and gestures of the deceased in image and video data.

[1599] Output: Preprocessed image and video data.

[1600] Step 9:

[1601] Input: Preprocessed image and video data.

[1602] Specific operation: The server uses image and video analysis models to learn how to reproduce the facial expressions and movements of the deceased.

[1603] Output: A trained image and video analysis model.

[1604] Step 10:

[1605] Input: Trained natural language processing models, speech synthesis models, and image / video analysis models.

[1606] How it works: The server uses machine learning frameworks such as TensorFlow to integrate each model and generate a single AI character, coordinating verbal, vocal, and facial output to achieve results that are closer to real-life interactions.

[1607] Output: A unified AI character.

[1608] Step 11:

[1609] Input: Integrated AI character.

[1610] Specific operation: The server packages this AI character and sends it to the user's device through a dedicated API.

[1611] Output: The AI ​​character sent to the user's device.

[1612] Step 12:

[1613] Input: A question that the user speaks.

[1614] Specific Actions: The user talks to the digital portrait, asking questions such as, "Hello, Grandpa. How was your day?"

[1615] Output: The audio data captured on the device.

[1616] Step 13:

[1617] Input: Audio data captured on the device.

[1618] How it works: The device sends the captured voice data to the server, which uses the Google Speech-to-Text API to convert the voice data into text and understand the question.

[1619] Output: Parsed text data.

[1620] Step 14:

[1621] Input: Parsed text data.

[1622] What happens: The server uses the NLP model to generate an appropriate response, such as "Hello, it was a beautiful day today. How was it for you?"

[1623] Output: The generated text response.

[1624] Step 15:

[1625] Input: The generated text response.

[1626] What it does: The server uses the Tacotron2 model to vocalize the text response in the voice of the deceased.

[1627] Output: A spoken response.

[1628] Step 16:

[1629] Input: A spoken response.

[1630] What it does: The device plays back spoken responses and also displays animations that replicate the facial expressions and gestures of the deceased.

[1631] Output: Played voice response and animation.

[1632] (Application example 1)

[1633] 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."

[1634] In recent years, with the advancement of AI technology, there has been an increase in efforts to generate digital memorial portraits using the communication data of the deceased, but conventional technologies have limited interaction capabilities with users and lack security.This invention provides a system that provides real-time response guidance from an AI character that resembles the deceased, and detects and warns abnormal behavior using a score prediction model, thereby achieving a safe and comfortable interaction experience for users.

[1635] 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.

[1636] In this invention, the server includes means for collecting communication data of the deceased, means for analyzing and preprocessing the communication data, means for training an AI model based on the preprocessed data, means for integrating the trained AI model to generate an AI character resembling the deceased, means for providing a digital memorial portrait device that displays the AI ​​character, means for receiving voice input from a user using the digital memorial portrait device and generating and displaying a response using the AI ​​character, and means for detecting abnormal behavior using a score prediction model and providing a real-time audio warning. This allows for a natural conversational experience with the deceased while simultaneously alerting security guards to any abnormal behavior.

[1637] "Communication data of the deceased" refers to a collection of electronic data used by the deceased during their lifetime, and primarily includes chat history, call records, photos, and videos.

[1638] "Preprocessing" refers to the initial stage of data processing, where collected raw data is analyzed, unnecessary information is removed, and important information is extracted.

[1639] An "AI model" is an artificial intelligence program that has been trained to perform a specific task using machine learning or deep learning techniques.

[1640] An "AI character" is a digital character that integrates a trained AI model and reproduces the characteristics of a specific person (in this case, a deceased person).

[1641] A "digital memorial device" is a device that displays an AI character and enables interaction with the user, and can be a smartphone, a dedicated display, or AR-compatible smart glasses.

[1642] "Voice input from the user" refers to voice instructions or questions given by the user to the system via a microphone or the like.

[1643] A "score prediction model" is a model that uses AI technology to analyze data in real time and predict risk scores associated with specific behaviors or situations.

[1644] "Abnormal behavior detection" refers to the process of recognizing and alerting when behavior that deviates from normal patterns occurs.

[1645] "Providing a real-time warning" refers to the act of immediately issuing a warning and notifying the user of detected abnormal behavior.

[1646] This invention provides a system that collects communication data of the deceased, uses it to build an AI model, and uses it to create a digital memorial portrait. The system aims to generate an AI character that resembles the deceased and provides responses that are characteristic of the deceased through conversations with the user. It also has the ability to detect abnormal behavior and provide real-time warnings using a score prediction model.

[1647] System Configuration

[1648] 1. Hardware Configuration

[1649] Server: Responsible for storing and processing data, and training and operating AI models. A server with high-performance processing capabilities is desirable.

[1650] Device: The device used by the user, which may be a smartphone, a dedicated display, or AR-enabled smart glasses.

[1651] Camera: Used to capture footage in real time and analyze it in the score prediction model.

[1652] Microphone: Used to capture the user's voice input.

[1653] 2. Software Configuration

[1654] Data collection module: Collects communication data such as chat history, call records, photos, and videos of the deceased.

[1655] Data preprocessing module: Analyzes the collected data and extracts the necessary information.

[1656] AI model learning module: Based on the preprocessed data, natural language processing (NLP) models, speech synthesis models, image and video analysis models, and score prediction models are trained.

[1657] Integration module: Integrates each model to generate an AI character that resembles the deceased.

[1658] Digital portrait display module: Displays the generated AI character and enables interaction with the user.

[1659] Score prediction module: Detects abnormal behavior in real time based on camera footage and provides necessary warnings via voice.

[1660] Operating Procedure

[1661] 1. Data Collection: Users enter the deceased person's communication data into the system, including chat history, call records, photos, videos, etc.

[1662] 2. Data Preprocessing: The server preprocesses the received data, removing unnecessary information and extracting important information. This process includes filtering spam messages and extracting key phrases.

[1663] 3. AI model training: The server uses the preprocessed data to train the NLP model, speech synthesis model, image and video analysis model, and score prediction model, thereby acquiring knowledge to reproduce the deceased's unique speech patterns, voice quality, facial expressions, etc.

[1664] 4. Model integration: The output of each model is integrated to generate an AI character that resembles the deceased person, with matching language, voice, and facial expressions.

[1665] 5. Interactive function: The user can talk to the digital portrait, for example, asking questions like, "How was your day?" The device captures this voice and sends it to the server.

[1666] 6. Real-time warning: Camera footage is analyzed in real time, and if abnormal behavior is detected, a warning is immediately given to the user, such as an audio warning saying "Warning! Intruder detected."

[1667] Specific examples

[1668] When a user speaks to a digital portrait and asks, "Hello, how was your day?", the response is generated through the following process:

[1669] 1. User: Speaks a question.

[1670] 2. Device: Captures audio and sends it to the server.

[1671] 3. Server: Analyzes the voice and understands the question.

[1672] 4. Server: Generates a response using an NLP model, such as "It was a beautiful day today. How was it for you?"

[1673] 5. Server: Generates a response in the voice of the deceased person using a speech synthesis model and sends it to the device.

[1674] 6. Terminal: Plays back the response and reproduces the facial expression of the deceased.

[1675] Prompt Sentence Examples

[1676] Example prompt sentence:

[1677] "Warning! Unauthorized person loitering near conference room. Please investigate immediately."

[1678] "Attention! Abnormal behavior detected. Please respond immediately."

[1679] This allows users to have a natural conversation with the deceased while also supporting immediate response as a security guard.

[1680] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1681] Step 1:

[1682] Data collection

[1683] The user inputs the deceased's communication data (chat history, call records, photos, videos, etc.) into the system. The device exports this data and sends it to the server. The server receives and stores the data.

[1684] Input: Chat history, call logs, photos, videos

[1685] Output: Saved communication data

[1686] Step 2:

[1687] Data Preprocessing

[1688] The server analyzes the received communication data, removes unnecessary information (spam messages and errors), and extracts important information (key phrases, frequently occurring vocabulary, and important topics). This preprocessing process prepares the data for AI model training.

[1689] Input: Saved communication data

[1690] Output: Preprocessed data

[1691] Step 3:

[1692] Training an AI model

[1693] The server uses the preprocessed data to train various AI models (NLP model, voice synthesis model, image / video analysis model, score prediction model), which then learns the deceased's speech patterns, voice quality, facial expressions, and behavioral patterns.

[1694] Input: Preprocessed data

[1695] Output: Trained AI model (NLP model, speech synthesis model, image / video analysis model, score prediction model)

[1696] Step 4:

[1697] Model Integration

[1698] The server then integrates each trained AI model to generate an AI character that resembles the deceased. The integration process involves linking the outputs of each model and adjusting the words, voice, and facial expressions to match.

[1699] Input: Trained AI model

[1700] Output: Integrated AI character

[1701] Step 5:

[1702] Displaying a digital portrait of the deceased

[1703] The server packages the integrated AI character and sends it to the terminal, which uses the received data to display the AI ​​character on the digital portrait device.

[1704] Input: Integrated AI character

[1705] Output: Displayed AI character

[1706] Step 6:

[1707] User interaction

[1708] The user speaks to the digital portrait. The device captures the voice with a microphone and sends the captured voice data to the server. The server uses speech recognition technology to analyze and understand the question. It uses an NLP model to generate an appropriate response and a speech synthesis model to generate a response in the deceased's voice. Finally, the response is sent to the device, which plays it back. In some cases, the deceased's facial expressions and gestures are also reproduced.

[1709] Input: User's voice

[1710] Output: Vocal responses, facial expressions, and gestures of the deceased

[1711] Step 7:

[1712] Abnormal behavior detection and real-time alerts

[1713] The device's camera captures video in real time. The server analyzes the video using a score prediction model, and if abnormal behavior is detected, it immediately provides an audio warning. For example, it notifies the user with a voice message saying, "Warning! Intruder detected."

[1714] Input: Real-time video

[1715] Output: Audio warning, abnormal behavior detection results

[1716] 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.

[1717] To implement this invention, we will create an AI model using the communication data of the deceased and build a system that can be used as a digital memorial portrait. This system will also incorporate an emotion engine that recognizes the user's emotions and provides responses according to their emotions.

[1718] System Overview

[1719] The system collects communication data from the deceased, trains an AI model using preprocessed data, and displays the generated AI character as a digital portrait. Furthermore, when a user speaks to the digital portrait, it uses an emotion engine to analyze the emotions and provide appropriate responses.

[1720] Program processing flow

[1721] 1. Data Collection Phase

[1722] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[1723] The device uploads the exported data to the server in the specified format (e.g., JSON or CSV file).

[1724] 2. Data preprocessing phase

[1725] The server stores the received communication data and performs pre-processing on the data, which includes removing unnecessary information (such as spam messages and errors) and extracting important information (key phrases, important topics, and frequently occurring vocabulary).

[1726] 3. AI model training phase

[1727] The server uses the preprocessed data to train natural language processing (NLP) models, speech synthesis models, and image and video analysis models. Each model works as follows:

[1728] NLP model: Analyzes text data and learns the words and phrases of the deceased.

[1729] Speech synthesis model: Extracts clear speech and trains to reproduce the voice quality of the deceased.

[1730] Image and video analysis model: Learns facial expressions and gestures and reproduces the facial expressions and movements of the deceased.

[1731] 4. Model integration phase

[1732] The server then integrates the trained models to operate as a single AI character, a process that involves linking the outputs of each model and ensuring that the words, voices, and facial expressions match.

[1733] 5. Digital portrait generation phase

[1734] The server packages the integrated AI character and sends it to the user's device, where it can be displayed on a digital memorial device (such as a dedicated display or smartphone).

[1735] 6. Emotion recognition and response phase

[1736] The emotion engine analyzes the user's voice data and facial expression data to recognize the user's emotions (e.g., happiness, sadness, anger, surprise, etc.).

[1737] The server adapts the NLP model and speech synthesis model based on the output of the emotion engine, allowing the AI ​​character of the deceased person to generate an appropriate response based on the recognized emotion.

[1738] The server also adjusts facial expressions and tone of voice according to emotions to recreate natural conversations.

[1739] Specific examples

[1740] For example, if a user says to a digital portrait, "Hello, Grandpa. I've been feeling a bit sad today," the following happens:

[1741] 1. User: Speaks a question.

[1742] 2. Device: Captures audio and sends it to the server.

[1743] 3. Server: Analyzes the voice and understands the question.

[1744] 4. Emotion engine: Analyzes the user's voice and facial expressions to recognize when the user is feeling "sad."

[1745] 5. Server: Uses NLP models to generate responses based on emotions, such as "Hi, I'm very worried to hear that you're feeling sad. What happened?"

[1746] 6. Server: Using a speech synthesis model, the generated response is generated in the voice of the deceased.

[1747] 7. Server: Based on the emotion engine, the tone of voice and facial expressions are adjusted to interact with the user in a more natural way.

[1748] 8. The device plays back the response received from the server and reproduces the facial expression of the deceased.

[1749] In this way, the system recognizes the user's emotions and allows them to provide emotional care through dialogue with the deceased.

[1750] The processing flow will be explained below.

[1751] Step 1:

[1752] The user accesses the deceased person's communication accounts (e.g., messaging apps) and exports chat history, call logs, photos, and videos.

[1753] Step 2:

[1754] The device uploads the exported data to the server in the specified format (e.g., JSON or CSV file).

[1755] Step 3:

[1756] The server receives the uploaded communication data and stores it in a secure database.

[1757] Step 4:

[1758] The server analyzes the stored data, removes unnecessary information (such as spam messages and errors), and extracts important information (key phrases, important topics, and frequently occurring vocabulary).

[1759] Step 5:

[1760] The server uses the preprocessed data to train a natural language processing (NLP) model, which analyzes and trains the text data to memorize the words and phrases used by the deceased.

[1761] Step 6:

[1762] The server uses the audio data to train a speech synthesis model, extracting clear speech and training it to reproduce the voice quality of the deceased person.

[1763] Step 7:

[1764] The server uses image and video data to train an image and video analysis model, analyzing the facial expressions and gestures of the deceased.

[1765] Step 8:

[1766] The server integrates the trained NLP model, speech synthesis model, and image / video analysis model, allowing it to operate as a single AI character.

[1767] Step 9:

[1768] The server packages the integrated AI character and sends it to the user's device.

[1769] Step 10:

[1770] Users can talk to the AI ​​character, which is displayed as a digital portrait of the deceased, asking questions such as, "How was your day?" or "Tell me about an old trip?"

[1771] Step 11:

[1772] The device captures the user's voice with a microphone and transmits the voice data to the server.

[1773] Step 12:

[1774] The server analyzes the received voice using automatic speech recognition (ASR) technology and understands the content of the question.

[1775] Step 13:

[1776] The server uses an emotion engine to analyze the user's voice data and facial expression data and recognize the user's emotions (e.g., happiness, sadness, anger, surprise, etc.).

[1777] Step 14:

[1778] The server uses NLP models to generate appropriate responses based on the recognized emotions, for example, if the user is expressing sadness, it generates a response such as "Hi, I'm very worried to hear that you're feeling sad. What happened?"

[1779] Step 15:

[1780] The server uses a speech synthesis model to generate the generated response in the voice of the deceased person, adjusting the tone of the voice based on the output of the emotion engine.

[1781] Step 16:

[1782] The server sends the generated response to the device, adjusting facial expressions and gestures as needed.

[1783] Step 17:

[1784] The device then plays back the responses received from the server and reproduces the facial expressions of the deceased, allowing the user to have a natural conversation.

[1785] Example 2

[1786] 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."

[1787] It is necessary to utilize the communication data of the deceased to create a digital memorial portrait that reflects the deceased's characteristics, and to realize a dialogue between the user and the deceased's AI character based on emotion recognition. However, conventional technologies have low accuracy in analyzing communication data and recognizing emotions, making it difficult to provide natural responses that correspond to the user's emotions. Another issue is the difficulty of integrating various models (e.g., natural language processing, speech synthesis, image and video analysis) to accurately reproduce the characteristics of the deceased.

[1788] 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.

[1789] In this invention, the server includes means for collecting communication data of the deceased, means for analyzing and preprocessing the communication data, means for training an AI model based on the preprocessed data, means for integrating the trained AI model to generate an AI character resembling the deceased, means for providing a display device for displaying the AI ​​character, means for receiving voice input from a user via the display device and generating and displaying a response using the AI ​​character, and means for recognizing the user's emotions and generating a response according to the emotion. This makes it possible to accurately reproduce the characteristics of the deceased while providing a natural and appropriate response according to the user's emotions.

[1790] "Communication data" refers to records of digital communications used by the deceased during their lifetime, such as chat history, call records, photos, and videos.

[1791] "Preprocessing" is the process of removing unnecessary information from the received data and extracting important information (such as key phrases and frequently occurring vocabulary).

[1792] An "AI model" is a collection of algorithms that use artificial intelligence techniques to analyze and learn from data and automate specific tasks.

[1793] A "natural language processing model" is a model that uses artificial intelligence technology to analyze human language and understand its meaning.

[1794] A "speech synthesis model" is a model that uses artificial intelligence technology to convert text data into speech.

[1795] The "image and video analysis model" is a model that uses artificial intelligence technology to analyze image and video data and understand its content.

[1796] An "AI character" is a digital character generated based on an AI model that reproduces the features of a deceased person.

[1797] A "display device" is a device for visually displaying an AI character to a user.

[1798] "Emotion recognition" is a technology that analyzes a user's voice data and facial expression data to identify their emotions.

[1799] To implement this invention, it is necessary to build a system that uses the communication data of the deceased to generate an AI model and use it as a digital memorial portrait. This system is equipped with an emotion engine that recognizes the user's emotions and provides responses according to the emotions.

[1800] Hardware and software used

[1801] Hardware

[1802] Server: A server for storing and analyzing data and training AI models. Use a cloud server (e.g., Amazon EC2) equipped with a high-performance processor and large memory capacity.

[1803] Terminal: A device that allows users to operate and input data. This includes smartphones, tablets, and PCs.

[1804] Display device: A device for displaying the digital portrait. A dedicated display or smartphone can be used.

[1805] software

[1806] Natural Language Processing (NLP) model: A model for analyzing text data and learning the vocabulary and phrasing of the deceased, using NLP techniques such as BERT (Bidirectional Encoder Representations from Transformers).

[1807] Speech synthesis model: A model for converting text to speech and recreating the voice quality of the deceased. Uses voice synthesis technology such as Tacotron 2.

[1808] Image and video analysis model: A model that analyzes facial expressions and gestures to recreate the movements and expressions of the deceased. It uses image and video analysis technologies such as OpenPose and Face++.

[1809] Emotion engine: A technology that analyzes the user's voice data and facial expression data to recognize the user's emotions. It uses a combination of DeepSpeech (voice recognition) and Face++ (facial expression recognition).

[1810] Specific examples

[1811] Data Collection Phase

[1812] The user exports the chat history from the deceased person's messaging app (e.g., WhatsApp), and then uploads it to the server using a dedicated app.

[1813] Data Preprocessing Phase

[1814] The server analyzes the received chat history, filters out unnecessary information, and then extracts important key phrases and frequently occurring vocabulary.

[1815] AI model training phase

[1816] The server uses the preprocessed data to train a BERT model to learn the deceased's words and phrases, then trains the model with Tacotron 2 on clear audio data to reproduce the deceased's voice, and trains the model with OpenPose and Face++ on facial expressions and gestures.

[1817] Model integration phase

[1818] The server integrates each model (natural language processing model, speech synthesis model, image / video analysis model) to generate an AI character that recreates the deceased person.

[1819] Usage example

[1820] For example, if a user says to a digital portrait, "Hello, Grandpa. I've been feeling a bit sad today," the following happens:

[1821] The user enters the question by voice.

[1822] The device captures the audio and sends it to the server.

[1823] The server analyzes the voice and understands the question.

[1824] The emotion engine analyzes the user's voice and facial expressions and recognizes that the user is feeling "sad."

[1825] The server uses an NLP model to generate a response such as, "Hello, I'm very worried to hear that you're feeling sad. What happened?"

[1826] The server uses a speech synthesis model to generate responses in the voice of the deceased, adjusting the tone of voice and facial expression depending on the emotion.

[1827] The device plays back the response received from the server and reproduces the facial expression of the deceased.

[1828] Prompt Sentence Examples

[1829] "Analyze the message data of the deceased and generate an AI model for natural conversation."

[1830] "Develop a system that recognizes and responds to user emotions. The system will use an emotion engine to engage in appropriate dialogue based on the user's emotions."

[1831] This allows the system to recognize the user's emotions and provide emotional care through dialogue with the deceased.

[1832] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1833] Step 1: Data collection

[1834] The user accesses the deceased person's messaging and social media accounts and exports chat history, call logs, photos, and videos (input: the deceased person's communication data).

[1835] The terminal uploads the exported data to the server in the specified format (e.g., JSON file or CSV file) (output: uploaded communication data).

[1836] Step 2: Data storage and preprocessing

[1837] The server receives the data sent from the terminal and stores it securely (input: uploaded communication data, output: stored communication data).

[1838] The server removes unnecessary information from the received data (specific operations: filtering spam messages and error logs) and extracts important key phrases and frequently occurring vocabulary (input: saved communication data, output: preprocessed data).

[1839] Step 3: Training the Natural Language Processing (NLP) Model

[1840] The server uses the preprocessed text data to train an NLP model (e.g., BERT) (input: preprocessed data, output: trained NLP model). Specifically, it tokenizes the raw data and trains the model to understand the structure of the text.

[1841] Step 4: Training the speech synthesis model

[1842] The server extracts the preprocessed voice data and trains it using a speech synthesis model (e.g., Tacotron 2) (input: preprocessed voice data, output: trained speech synthesis model). Specifically, it analyzes the features of the voice data and reproduces the voice quality of the deceased.

[1843] Step 5: Training the image and video analysis model

[1844] The server uses the preprocessed image and video data to train an image and video analysis model (e.g., OpenPose, Face++) (input: preprocessed image and video data, output: trained image and video analysis model). Specifically, it analyzes the facial expressions and gestures of the deceased and reproduces them.

[1845] Step 6: Model integration

[1846] The server integrates each trained model to generate a single AI character (input: trained NLP model, trained speech synthesis model, trained image and video analysis model, output: integrated AI character). Specifically, the text generated by the NLP model is input into the speech synthesis model, and the corresponding facial expression is generated by the image and video analysis model.

[1847] Step 7: Generate a digital portrait

[1848] The server packages the integrated AI character and sends it to the user's device (input: integrated AI character, output: package sent to the user's device).

[1849] The user installs the file sent to their device and displays the AI ​​character on the digital portrait device (input: sent package, output: displayed AI character).

[1850] Step 8: Emotion Recognition and Response Generation

[1851] The emotion engine analyzes the user's voice data and facial expression data and recognizes emotions (input: user's voice data, facial expression data, output: recognized emotions).

[1852] Based on the output of the emotion engine, the server adapts the NLP model and speech synthesis model to generate an appropriate response (input: recognized emotion, output: generated response).

[1853] The server generates the response in the voice of the deceased person and adjusts the tone of voice and facial expression according to the emotion (input: generated response, output: adjusted response and facial expression).

[1854] The device plays back the responses received from the server and reproduces the facial expressions of the deceased (input: adjusted responses and facial expressions, output: reproduced responses and facial expressions).

[1855] This series of processes allows the user to receive emotional care through natural conversation with the deceased.

[1856] (Application example 2)

[1857] 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."

[1858] When applying a digital memorial portrait system to a virtual store, a means of understanding the user's emotions and responding appropriately is required. Another challenge is to increase customer satisfaction by providing a customer service experience that recreates the memories and voice of the deceased.

[1859] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1860] In this invention, the server includes means for collecting communication data of the deceased, means for analyzing and preprocessing the communication data, means for training an AI model based on the preprocessed data, means for integrating the trained AI model to generate an AI character resembling the deceased, means for providing a digital portrait device that displays the AI ​​character, means for receiving voice input from a user using the digital portrait device and generating and displaying a response using the AI ​​character, means for analyzing the user's emotions using an emotion recognition engine, means for generating a response based on the user's emotions, and means for using the digital portrait device as a guide in a virtual store. This makes it possible to grasp the user's emotions and provide effective guidance in the virtual store by reproducing the deceased's responses according to those emotions.

[1861] "Communication data" refers to digital information left behind by the deceased, such as chat history, call records, photos, and videos.

[1862] "Preprocessing" refers to the process of removing unnecessary information from collected communication data and extracting important information.

[1863] An "AI model" is an artificial intelligence that is trained using technologies such as natural language processing, speech synthesis, and image and video analysis to carry out specific tasks.

[1864] An "AI character" is a digital character that resembles a deceased person and is generated by integrating a trained AI model.

[1865] A "digital memorial device" is a device that displays an AI character and allows the user to interact with it.

[1866] "Voice input" refers to the voice data that is generated when the user speaks to the digital memorial portrait device.

[1867] An "emotion recognition engine" is a system for analyzing emotions from a user's voice and images.

[1868] A "virtual store" is a store that operates in a virtual space, a place that offers products and services online.

[1869] A "guide" is a person whose role is to provide customers with product information and guidance within the store.

[1870] A "user" is a person who uses the digital memorial device to interact with an AI character of a deceased person.

[1871] To implement this invention, it is necessary to create an AI model using the communication data of the deceased and build a system that functions as a digital memorial guide in a virtual store. This system reproduces the memories and voice of the deceased and combines it with an emotion recognition engine to provide appropriate responses according to the user's emotions.

[1872] System Configuration

[1873] 1. The user uses a smartphone or smart glasses to talk to a digital portrait guide.

[1874] 2. The device captures the user's voice input and facial image data and sends them to the server.

[1875] 3. The server executes multiple methods to perform the following processes:

[1876] Data collection and preprocessing

[1877] The server collects the deceased's communication data (chat history, call records, photos, videos). The collected data is pre-processed using technologies such as NLP (natural language processing) and image and video analysis to remove unnecessary information and extract important information.

[1878] AI model training and integration

[1879] The server uses the preprocessed data to train natural language processing models, speech synthesis models, and image and video analysis models. These trained models are then integrated to generate an AI character that resembles the deceased. This AI character reproduces the deceased's speech, voice quality, facial expressions, etc., allowing users to interact with the deceased in a virtual store.

[1880] Emotion Recognition and Response Generation

[1881] The server is equipped with an emotion recognition engine that analyzes the user's emotions from their voice and facial expression data. Based on the emotion engine's analysis results, an NLP model generates an appropriate text response, and a speech synthesis model reproduces that text as the deceased's voice. This enables natural conversations that correspond to the user's emotions.

[1882] Use as a digital memorial guide

[1883] In the virtual store, users can interact with a digital memorial guide through their smartphones or smart glasses. For example, if a customer asks, "I'd like to know more about our new products," the server will analyze their voice and facial expression data and provide an answer. Specifically, it will respond with something like, "Of course. Here are our latest products, and their features are as follows."

[1884] Prompt Sentence Examples

[1885] An example prompt for generating an AI model is:

[1886] You are a digital memorial guide working in a virtual store. Respond to customers' questions and interests in a friendly and courteous manner. Strive to provide relevant and useful information while taking into consideration the customer's feelings. For example, if a customer asks, "I'd like to know more about our new products," you can respond, "Of course! Here are our latest products and their features," and then provide specific product information.

[1887] As described above, the present invention makes it possible to utilize a digital portrait of the deceased as a guide in a virtual store while being sensitive to the user's emotions.

[1888] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1889] Step 1:

[1890] The user speaks to the digital portrait guide through a smartphone or smart glasses. The user inputs questions by voice, and simultaneously captures facial image data. This input data is captured by the device, which then transmits the voice data and facial image data to a server.

[1891] Step 2:

[1892] The server analyzes the received voice data. Specifically, it converts it into text data using voice recognition technology (such as Google Speech API). At this time, voice data is input, and as a result, text data containing the user's question is output.

[1893] Step 3:

[1894] The server sends the received facial image data to an emotion recognition engine to analyze the user's emotions. The emotion recognition engine uses image analysis technology (such as OpenCV or Dlib) to analyze emotions from facial expressions. The input is facial image data, and the output is data indicating the user's emotions (e.g., happiness, sadness, anger, etc.).

[1895] Step 4:

[1896] The server uses the analyzed text data and emotion data to work with an NLP model to generate an appropriate text response. Specifically, NLP techniques are applied to the text data to generate candidate responses. The emotion data is used to adjust the response. The input is text data containing the question and emotion data, and the output is the generated appropriate response text.

[1897] Step 5:

[1898] The server sends the generated response text to a speech synthesis model, which generates audio data in the voice of the deceased. A speech synthesis model (such as Amazon Polly or Google Text-to-Speech) is used. The input is the response text, and the output is audio data that reproduces the voice of the deceased.

[1899] Step 6:

[1900] The server sends the generated voice data to the terminal, which then plays the received voice data to the user, allowing the user to interact with the digital portrait guide. The input is the voice data, and the output is the played voice response.

[1901] Step 7:

[1902] After the server or terminal has completed responding to the user's question, it waits for the next question or interaction. If the user has any new input, it returns to step 1.

[1903] Through the above processing steps, the user can have natural conversations with the digital portrait guide and is provided with guidance services in the virtual store.

[1904] 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.

[1905] 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.

[1906] 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.

[1907] 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.

[1908] 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.

[1909] 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.

[1910] 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).

[1911] 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.

[1912] 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."

[1913] 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.

[1914] 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).

[1915] 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.

[1916] 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.

[1917] 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.

[1918] 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.

[1919] 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.

[1920] 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.

[1921] 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.

[1922] 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.

[1923] 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.

[1924] 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.

[1925] The following is further disclosed regarding the above embodiment.

[1926] (Claim 1)

[1927] a means of collecting the deceased person's communications data;

[1928] means for analyzing and preprocessing the communication data;

[1929] A means for training an AI model based on the preprocessed data;

[1930] A means for integrating the learned AI model to generate an AI character that resembles the deceased person;

[1931] A means for providing a digital memorial portrait device that displays the AI ​​character;

[1932] means for receiving voice input from a user via said digital portrait device and generating and displaying a response using said AI character;

[1933] A system including:

[1934] (Claim 2)

[1935] The system of claim 1, wherein the communication data includes chat history, call logs, photos, and videos.

[1936] (Claim 3)

[1937] The system of claim 1, wherein the AI ​​models include a natural language processing (NLP) model, a speech synthesis model, and an image / video analysis model.

[1938] "Example 1"

[1939] (Claim 1)

[1940] a means of collecting the deceased person's communications data;

[1941] means for analyzing and preprocessing the communication data;

[1942] A means for training an AI model based on the preprocessed data;

[1943] A means of training an AI model using natural language processing models, speech synthesis models, and image / video analysis models;

[1944] A means to integrate each trained model and link data so that words, voices, and facial expressions match;

[1945] A means for packaging and transmitting the integrated AI character to a user's device;

[1946] A means for providing a digital memorial portrait device that displays the AI ​​character;

[1947] means for receiving voice input from a user via said digital portrait device and generating and displaying a response using said AI character;

[1948] A system including:

[1949] (Claim 2)

[1950] The system of claim 1, wherein the communication data includes message history, call logs, images, and videos.

[1951] (Claim 3)

[1952] The system of claim 1, wherein the AI ​​models include a natural language processing model, a speech synthesis model, and an image / video analysis model.

[1953] "Application Example 1"

[1954] (Claim 1)

[1955] a means of collecting the deceased person's communications data;

[1956] means for analyzing and preprocessing the communication data;

[1957] A means for training an AI model based on the preprocessed data;

[1958] A means for integrating the learned AI model to generate an AI character that resembles the deceased person;

[1959] A means for providing a digital memorial portrait device that displays the AI ​​character;

[1960] means for receiving voice input from a user via said digital portrait device and generating and displaying a response using said AI character;

[1961] A means for detecting abnormal behavior using a score prediction model and providing real-time audio warnings;

[1962] A system including:

[1963] (Claim 2)

[1964] The system of claim 1, wherein the communication data includes chat history, call logs, photos, and videos.

[1965] (Claim 3)

[1966] 2. The system of claim 1, wherein the AI ​​models include a natural language processing (NLP) model, a speech synthesis model, an image / video analysis model, and a score prediction model.

[1967] "Example 2: Combining Emotion Engines"

[1968] (Claim 1)

[1969] a means of collecting the deceased person's communications data;

[1970] means for analyzing and preprocessing the communication data;

[1971] A means for training an AI model based on the preprocessed data;

[1972] A means for integrating the learned AI model to generate an AI character that resembles the deceased person;

[1973] means for providing a display device for displaying the AI ​​character;

[1974] means for receiving a voice input from a user by said display device and generating and displaying a response using said AI character;

[1975] means for recognizing a user's emotion and generating a response according to the emotion;

[1976] A system including:

[1977] (Claim 2)

[1978] The system of claim 1, wherein the communication data includes chat history, call logs, photos, and videos.

[1979] (Claim 3)

[1980] The system of claim 1, wherein the AI ​​models include a natural language processing model, a speech synthesis model, and an image / video analysis model.

[1981] "Application example 2 when combining emotion engines"

[1982] (Claim 1)

[1983] a means of collecting the deceased person's communications data;

[1984] means for analyzing and preprocessing the communication data;

[1985] A means for training an AI model based on the preprocessed data;

[1986] A means for integrating the learned AI model to generate an AI character that resembles the deceased person;

[1987] A means for providing a digital memorial portrait device that displays the AI ​​character;

[1988] means for receiving voice input from a user via said digital portrait device and generating and displaying a response using said AI character;

[1989] means for analyzing a user's emotions using an emotion recognition engine;

[1990] means for generating a response based on the user's emotion;

[1991] a means for using the digital portrait device as a guide in a virtual store;

[1992] A system including:

[1993] (Claim 2)

[1994] 10. The system of claim 1, wherein the communication data includes chat history, call logs, photos, and videos.

[1995] (Claim 3)

[1996] The system of claim 1, wherein the AI ​​models include a natural language processing (NLP) model, a speech synthesis model, and an image / video analysis model. [Explanation of symbols]

[1997] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. a means of collecting the deceased person's communications data; means for analyzing and preprocessing the communication data; A means for training an AI model based on the preprocessed data; A means for integrating the learned AI model to generate an AI character that resembles the deceased person; A means for providing a digital memorial portrait device that displays the AI ​​character; means for receiving voice input from a user via said digital portrait device and generating and displaying a response using said AI character; A system including:

2. The system of claim 1 , wherein the communication data includes chat history, call logs, photos, and videos.

3. The system according to claim 1 , wherein the AI ​​models include a natural language processing model, a speech synthesis model, and an image / video analysis model.

Images