system

Abstract

We provide the system. [Solution] A means of tracking the user's gaze, A means of analyzing the speaker's mouth movements, A method for estimating the content of speech from analyzed mouth movements, A system that includes means for displaying the estimated content of a statement.

Description

Technical Field

[0005] ,

[0001] The technology of the present disclosure relates to a system.

Background Art

[0002] Patent Document 1 discloses a method for controlling a persona chatbot, which is performed by at least one processor, including steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to an explanation of a 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

Summary of the Invention

Problems to be Solved by the Invention

[0004] Hearing-impaired people have difficulty in voice communication and often use alternative means such as sign language and written conversation. However, these means block both hands, so there is a problem that they cannot perform another task simultaneously. Also, when multiple speakers are conversing at the same time, it is difficult to follow the flow of the conversation. Therefore, hearing-impaired people may experience information loss and a sense of social isolation. There is a need for new communication means to solve these problems.

Means for Solving the Problems

[0005] The present invention provides a system that includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, and means for displaying the estimated content of the speech. It also includes means for selecting one speaker from among multiple speakers based on the user's gaze. This system includes means for arranging the content of the speech into natural sentences based on context, means for displaying the content of the speech in speech bubble format, and means for adjusting the display position of the content of the speech for each speaker the user is focusing on. As a result, people with hearing impairments can communicate without using their hands and can easily understand conversations with multiple speakers simultaneously.

[0006] "User" refers to a person with hearing impairment who uses this system.

[0007] "Eye-tracking" is a technology that detects the user's pupils and gaze direction and analyzes that data.

[0008] The "speaker" is the person with whom the user communicates, and whose mouth movements are analyzed by this system.

[0009] "Mouth movement analysis" is a process that captures the movement of a speaker's lips using cameras or sensors and estimates the content of their speech from that movement.

[0010] "Speech content estimation" is a process that predicts with high accuracy what a speaker is saying based on analyzed lip movements.

[0011] "Displaying the content of a statement" is the process of visually providing the user with the estimated content of a statement.

[0012] "Speech bubble format" refers to a method of displaying text in the shape of a speech bubble that surrounds the spoken content, as is often used in manga and anime.

[0013] "Eye-gaze-based speaker selection" is a method that uses the user's eye-gaze information to automatically identify the speaker they are focusing on from among multiple speakers.

[0014] "Context-based natural sentence formatting" is the process of modifying the estimated text of a statement, taking context into account, to make it easier to read as natural language.

[0015] "Adjusting the display position" is the process of optimizing the display position of spoken content according to the user's eye gaze and the speaker's position. [Brief explanation of the drawing]

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

Mode for Carrying Out the Invention

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

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

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

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

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

[0022] In the following embodiments, the signed communication interface (I / F) is an interface that includes a communication processor and an antenna, etc. The communication interface manages communication between multiple computers. Examples of communication standards applicable to the communication interface include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0023] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0024] [First Embodiment]

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

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

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

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

[0029] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0030] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0031] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

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

[0033] As shown in Figure 2, in the data processing device 12, a specific processing 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" related to the technology of this 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 according to the specific processing program 56 executed on the RAM 30.

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

[0035] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

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

[0037] This invention relates to a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content. More specifically, it relates to a system that enables hearing-impaired individuals to communicate with others without using their hands and to easily understand conversations with multiple speakers.

[0038] System Configuration

[0039] This system consists of the following main components:

[0040] 1. Device (smart glasses)

[0041] 2. Server

[0042] 3. User

[0043] terminal

[0044] The device takes the form of smart glasses. The device is equipped with the following functions:

[0045] Eye-tracking function: Tracks the user's gaze and collects eye-tracking information.

[0046] Camera: Captures video of the speaker the user is focusing on.

[0047] Display: Visually displays the estimated content of the statement.

[0048] server

[0049] The server receives data sent from the terminal and performs the following processing:

[0050] Mouth movement analysis: Analyze the speaker's mouth movements to estimate the content of their speech.

[0051] Natural language processing: Refining estimated speech content into natural-sounding sentences based on context.

[0052] Data transmission: The estimated content of the statement is sent to the terminal.

[0053] Program processing

[0054] Eye tracking and speaker identification

[0055] 1. The device uses eye-tracking technology to track the user's gaze in real time. This allows it to identify the speaker the user is focusing on.

[0056] 2. The device captures video of the identified speaker using its camera and sends that data to the server.

[0057] Estimation of the content of the statement

[0058] 3. The server receives the transmitted speaker's video data and analyzes the speaker's mouth movements using Gemini®-based lip-reading technology.

[0059] 4. The server accurately estimates the content of speech based on lip-sync data. The estimated text data is analyzed using a ChatGPT®-based natural language processing model to refine it into natural and accurate sentences based on context.

[0060] Display of the content of the statement

[0061] 5. The server sends the generated text data to the terminal.

[0062] 6. The device displays the received text data in a speech bubble format on the screen. The display position is adjusted based on the user's eye-tracking information.

[0063] Specific example

[0064] Example 1: One-on-one conversation

[0065] If user A is having a conversation with friend B

[0066] The device detects that user A's gaze is directed towards friend B.

[0067] The device captures friend B's face with its camera and sends the video to the server.

[0068] The server analyzes friend B's mouth movements and estimates that he said, "It's a nice day today."

[0069] The estimated text is processed using natural language processing on the server and then sent to user A's terminal.

[0070] The device displays the text "It's a nice day today" in a speech bubble.

[0071] Example 2: Conversation involving multiple people

[0072] If user A is having a conversation with friends B and C

[0073] The device detects that user A's gaze has shifted from B to C.

[0074] Based on each user's gaze data, the device sends the corresponding video to the server.

[0075] The server estimates the content of what B and C said and formats it as text data.

[0076] The device displays B's statement, "What are your plans for tomorrow?", and C's statement, "I want to go to the movies," in speech bubble format at their respective positions on the screen.

[0077] In this way, this system enables hearing-impaired individuals to communicate smoothly with others without using their hands. As a result, users will not miss important information and will be able to live a more natural life without feeling socially isolated.

[0078] The following describes the processing flow.

[0079] Step 1:

[0080] The device uses eye-tracking technology to track the user's gaze. It detects the position of the user's pupils and the direction of their gaze to determine which direction they are looking.

[0081] Step 2:

[0082] The device identifies the person (speaker) the user is focusing on based on eye-tracking information. If there are multiple speakers, the person being focused on the most is selected as the speaker.

[0083] Step 3:

[0084] The device captures the face of the identified speaker using its camera. The acquired video data is processed in real time, so it is captured continuously without interruption.

[0085] Step 4:

[0086] The terminal encodes the captured video data and sends it to the server over the network. A high-speed communication protocol is used to minimize latency during this process.

[0087] Step 5:

[0088] The server analyzes the received video data using Gemini-based lip-reading technology. It extracts the speaker's mouth movements and estimates the content of their speech with high accuracy based on those movements.

[0089] Step 6:

[0090] The server further analyzes the estimated speech content using a ChatGPT-based natural language processing model. This refines the estimated text into a natural and accurate sentence based on the context.

[0091] Step 7:

[0092] The server encodes the generated text data and sends it to the terminal over the network. It applies protocols to ensure data integrity during transmission.

[0093] Step 8:

[0094] The device decodes the received text data and displays it on the screen in a speech bubble format. The display position is adjusted based on the user's eye gaze and the speaker's position.

[0095] Step 9:

[0096] Users read speech bubbles displayed on the smart glasses' screen to visually understand what the speaker is saying. This allows them to grasp the conversation without using their hands.

[0097] (Example 1)

[0098] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0099] There is a need for a system that allows people with hearing impairments to communicate with others without using their hands. In particular, the challenge lies in smoothly understanding conversations and facilitating smooth communication, even when multiple speakers are involved. This requires advanced processing such as eye tracking, lip-syncing, speech content estimation, and natural language processing for text formatting, and there is a demand for a system that can achieve these capabilities.

[0100] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0101] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, means for displaying the estimated content of the speech, means for collecting video data from a camera, and means for processing the estimated content of the speech into natural-sounding text using natural language processing. This makes it possible for people with hearing impairments to obtain a lot of information without using their hands and to communicate smoothly with others.

[0102] A "user" is a person who uses a system to communicate.

[0103] A "means of tracking eye movements" refers to a device that has the function of detecting and tracking a user's gaze in real time.

[0104] The "speaker" is the person the user is focusing their gaze on and who is speaking.

[0105] A "means for analyzing mouth movements" is a device that analyzes the movements of a speaker's mouth and estimates the content of what they are saying from those movements.

[0106] A "means for estimating the content of speech" is a device that estimates a speaker's speech as text data based on analyzed lip-movement data.

[0107] "Means for displaying the content of a statement" refers to a device for visually conveying the estimated content of a statement to the user.

[0108] "Means for collecting video data using a camera" refers to a device for capturing video of a speaker and collecting that data.

[0109] "Natural language processing" is a technique for shaping estimated speech content into natural-sounding sentences based on context.

[0110] A "server" is a device that receives data transmitted from a terminal, performs analysis and estimation, and transmits data.

[0111] A "device" is a device worn by the user that tracks their gaze, captures video with a camera, and transmits data to a server.

[0112] This invention relates to a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the results. Specifically, it is a system that enables hearing-impaired individuals to communicate smoothly with others without using their hands.

[0113] System Configuration

[0114] This system is mainly composed of the following components:

[0115] 1. Device (smart glasses)

[0116] 2. Server

[0117] 3. User

[0118] terminal

[0119] The device takes the form of smart glasses and has the following functions:

[0120] Eye-tracking function: Tracks the user's gaze in real time and collects gaze information.

[0121] Camera: Captures video of the speaker the user is focusing on.

[0122] Display: Visually displays the estimated content of the statement.

[0123] server

[0124] The server receives data sent from the terminal and performs the following processing:

[0125] Mouth movement analysis: The speaker's mouth movements are analyzed using Gemini-based technology.

[0126] Natural Language Processing: The estimated utterance content is refined based on context using a ChatGPT-based natural language processing model.

[0127] Data transmission: The estimated content of the statement is sent to the terminal.

[0128] System operation

[0129] Eye tracking and speaker identification

[0130] The device uses eye-tracking technology to track the user's gaze. For example, when a user looks at someone they are talking to, that gaze information is instantly collected. This allows the device to identify the speaker the user is focusing on.

[0131] Data capture and transmission

[0132] The device captures video of the identified speaker using its built-in camera. The captured video data is sent to a server. This transmission uses high-speed communication technology (e.g., a 5G network) to process the data in real time.

[0133] Processing video data and analyzing mouth movements

[0134] The server receives video data transmitted from the terminal and analyzes it using Gemini-based lip-reading technology. The server analyzes the speaker's mouth movements in detail from the video data and extracts the content of the speech from those movements.

[0135] Estimation of spoken content and natural language processing

[0136] The server accurately estimates the content of the speech based on the analyzed mouth movement data. The estimated content is then formatted into natural-sounding text using a ChatGPT-based natural language processing model.

[0137] Sending and displaying estimated results

[0138] The server sends the formatted text to the terminal. The terminal displays the received text data in a speech bubble format on its screen. The display position is adjusted based on the user's eye gaze to ensure that the text is intuitively understandable.

[0139] Specific example

[0140] Example 1: One-on-one conversation

[0141] If user A is having a conversation with friend B:

[0142] The device detects that user A's gaze is directed towards friend B.

[0143] The device captures friend B's face with its camera and sends the video to the server.

[0144] The server analyzes friend B's mouth movements and estimates that he said, "It's a nice day today."

[0145] The estimated text is processed using natural language processing on the server and then sent to user A's terminal.

[0146] The device displays the text "It's a nice day today" in a speech bubble.

[0147] Example 2: Conversation involving multiple people

[0148] If user A is having a conversation with friends B and C:

[0149] The device detects that user A's gaze has shifted from B to C.

[0150] Based on each user's gaze data, the device sends the corresponding video to the server.

[0151] The server estimates the content of what B and C said and formats it as text data.

[0152] The device displays B's statement, "What are your plans for tomorrow?", and C's statement, "I want to go to the movies," in speech bubble format at their respective positions on the screen.

[0153] Example of a prompt

[0154] Please describe a system that tracks the user's gaze and analyzes the speaker's mouth movements to estimate what they are saying. Explain this using specific examples of one-on-one and group conversations.

[0155] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0156] Step 1:

[0157] The device uses eye-tracking technology to track the user's gaze in real time. The input is the user's gaze information, and the main output is coordinate data of the direction the gaze is directed. Based on this coordinate data, the device identifies the speaker the user is focusing on. A high-precision eye-tracking sensor built into the device captures even the slightest movement of the gaze and collects it as gaze information.

[0158] Step 2:

[0159] The device captures video of the speaker, identified based on eye-tracking information, using its built-in camera. The input is the speaker's location information identified by eye tracking, and the output is the captured video data. The built-in camera acquires high-resolution video and collects video data by focusing on the speaker's mouth. The collected video data is processed in real time and transmitted to the server.

[0160] Step 3:

[0161] The server receives video data transmitted from the terminal. The input is video data transmitted from the terminal, and the output is data in which the speaker's mouth movements have been analyzed. The server analyzes the video data using Gemini-based lip-reading technology and extracts specific movements that form words from the speaker's mouth movements. It performs noise reduction and correction on the data to achieve highly accurate mouth movement analysis.

[0162] Step 4:

[0163] The server estimates the speaker's speech based on analyzed lip-sync data. The input is lip-sync data, and the output is estimated text data. The server uses machine learning algorithms to estimate the actual spoken words with high accuracy. The estimated speech content is then formatted using contextual information and known language models.

[0164] Step 5:

[0165] The server uses a ChatGPT-based natural language processing model to format the estimated utterance into natural-sounding text. The input is estimated text data, and the output is context-formatted natural-sounding text. The server performs contextual analysis and formats the text as natural dialogue.

[0166] Step 6:

[0167] The server sends the formatted speech to the terminal. The input is formatted, natural-sounding text, and the output is text data sent to the terminal. The data is compressed and transmitted efficiently using high-speed communication technology (e.g., 5G network).

[0168] Step 7:

[0169] The terminal displays received text data in a speech bubble format on its screen. The input is text data sent from the server, and the output is text information visually displayed to the user. The display adjusts its position based on the user's eye gaze, positioning the text in an intuitively understandable location.

[0170] (Application Example 1)

[0171] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0172] When hearing-impaired individuals or workers have difficulty communicating effectively, it can be particularly challenging in work environments such as factories to accurately understand work instructions, potentially compromising safety and efficiency. This invention aims to solve these problems and provide a system that supports users in easily communicating with others.

[0173] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0174] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, means for displaying the estimated content of the speech, means for visually displaying work instructions, and means for facilitating smooth communication with robots and other workers. As a result, the user can visually confirm the content of the speech of the speaker they are looking at, and can also clearly and quickly grasp work instructions within the factory.

[0175] A "user" is an individual who uses the system to utilize eye-tracking and speech content estimation features.

[0176] "Methods for tracking eye movements" refer to devices and technologies that detect the direction of a user's gaze in real time and acquire that data.

[0177] The "speaker" is the person the user is looking at, and whose mouth movements are being analyzed.

[0178] "Methods for analyzing mouth movements" refer to signal processing and algorithms that detect mouth movements based on video data captured by a camera and estimate the content of speech based on those movements.

[0179] "Means for estimating the content of speech" refers to methods and techniques that analyze the speaker's lip-sync data to infer what the speaker is saying.

[0180] "Means for displaying estimated speech content" refers to displays or devices that visually present estimated text to the user.

[0181] "Means of visually displaying work instructions" refers to a system that displays and presents work instructions and announcements within a factory within the user's field of vision.

[0182] A "robot" is a mechanical device that operates autonomously or remotely and performs tasks within a factory.

[0183] "Means of facilitating smooth communication" refers to methods and technologies for seamlessly transferring information between users, robots, and other workers.

[0184] This invention is a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content. This system is particularly intended to help factory workers communicate smoothly with robots and other workers.

[0185] System Configuration

[0186] This system consists of the following main components:

[0187] 1. Device (smart glasses)

[0188] 2. Server

[0189] 3. User

[0190] terminal

[0191] The device takes the form of smart glasses. The device includes the following functions:

[0192] Eye-tracking function: Tracks the user's gaze and collects eye-tracking information.

[0193] Camera: Captures video of the speaker the user is focusing on.

[0194] Display: Visually displays estimated spoken content and work instructions.

[0195] server

[0196] The server receives data sent from the terminal and performs the following processing:

[0197] Mouth movement analysis: Analyze the speaker's mouth movements to estimate the content of their speech.

[0198] Natural language processing: Refining estimated speech content into natural-sounding sentences based on context.

[0199] Data transmission: Estimated spoken content and work instructions are sent to the terminal.

[0200] Program Description

[0201] The server processes data using Python, OpenCV, and dlib. Eye-tracking and camera data are acquired from sensors in smart glasses. The speaker's mouth movements are detected and analyzed using the dlib library. The estimated speech content is formatted using natural language processing techniques such as Google® Text-to-Speech (gTTS) and displayed on the screen.

[0202] Specific example

[0203] Example 1: Work instructions given in a one-on-one conversation

[0204] A scene where worker A receives instructions from robot operator B in a factory:

[0205] The device detects that A's gaze is directed towards B.

[0206] The device captures B's face with its camera and sends the video to the server.

[0207] The server analyzes B's mouth movements and infers that he said, "Please attach the parts."

[0208] The estimated text is processed using natural language processing and then sent to terminal A.

[0209] The device displays the text "Please install the parts" on its screen.

[0210] Example 2: Communication with multiple speakers

[0211] A scene in a factory where worker A interacts with multiple robot operators C and D:

[0212] The device detects that A's gaze has shifted from C to D.

[0213] The corresponding video is sent to the server, and the mouth movements of each person are analyzed.

[0214] The server infers the meaning of C's statement, "Proceed to the next step," and D's statement, "Check the parts."

[0215] The device displays each statement on its screen.

[0216] Examples of prompts to input into a generative AI model

[0217] TXT

[0218] Person A, working in a factory, is wearing smart glasses. Operator B says, "Next, please attach this part." The smart glasses track Person A's gaze and analyze B's mouth movements, displaying the spoken words as text. The smart glasses' display shows "Next, please attach this part," and Person A understands the instructions accurately.

[0219] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0220] Step 1:

[0221] The device tracks the user's gaze in real time using eye-tracking technology. The input is the user's gaze information, and the output is data on the direction the gaze is directed. This gaze information is collected using built-in sensors and a camera.

[0222] Step 2:

[0223] The device captures the face of the person it is looking at using its camera. The input is gaze information and camera footage, and the output is video data of the person the device is looking at.

[0224] Step 3:

[0225] The terminal sends the captured video data to the server. The input is video data, and the output is video data transmitted over the network.

[0226] Step 4:

[0227] The server analyzes the received video data and detects the speaker's mouth movements. The input is the transmitted video data, and the output is the analysis data regarding the mouth movements. The dlib library is used to detect the mouth movements.

[0228] Step 5:

[0229] The server estimates the content of a speaker's speech based on analysis data of their mouth movements. The input is mouth movement data, and the output is estimated text data. Natural language processing technology (generative AI model) is used for the estimation.

[0230] Step 6:

[0231] The server reshapes the estimated text data into natural-sounding sentences based on context. The input is estimated text data, and the output is contextually reshaped sentences.

[0232] Step 7:

[0233] The server sends formatted text data to the terminal. The input is formatted text data, and the output is text data transmitted over the network.

[0234] Step 8:

[0235] The device displays received text data on its screen. The input is text data sent from the server, and the output is text displayed on the smart glasses' screen. The display format is adjusted based on the user's eye-tracking information.

[0236] Step 9:

[0237] The user visually confirms the spoken content displayed on the smart glasses' screen. This action allows the user to understand the robot operator's instructions and work commands.

[0238] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0239] This invention is a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content, further combining it with an emotion engine that recognizes the user's emotions. This system allows hearing-impaired individuals to communicate with others without using their hands, and can improve the quality of communication by presenting the content of what is being said in a display format that corresponds to the user's emotions. Specifically, it has the following configuration and processing flow.

[0240] System Configuration

[0241] This system consists of the following main components:

[0242] 1. Device (smart glasses)

[0243] 2. Server

[0244] 3. Emotional Engine

[0245] terminal

[0246] The device takes the form of smart glasses and is equipped with the following functions:

[0247] Eye-tracking function: Tracks the user's gaze and collects eye-tracking information.

[0248] Camera: Captures video of the speaker the user is focusing on.

[0249] Emotion recognition function: Analyzes the user's emotions from their facial expressions, voice, etc.

[0250] Display: Visually displays the estimated content of the statement.

[0251] server

[0252] The server receives data sent from the terminal and performs the following processing:

[0253] Mouth movement analysis: Analyze the speaker's mouth movements to estimate the content of their speech.

[0254] Natural language processing: Refining estimated speech content into natural-sounding sentences based on context.

[0255] Integration with the emotion engine: Use data from the emotion engine to improve the accuracy of estimating the content of speech.

[0256] Emotional Engine

[0257] The emotion engine recognizes the user's emotions and has the following functions:

[0258] Sentiment analysis: Analyzes the user's facial expressions and voice to identify their emotional state.

[0259] Display format adjustment: The display format of the user's comments will change according to their emotions.

[0260] Program processing

[0261] Eye tracking and speaker identification

[0262] 1. The device uses eye-tracking technology to track the user's gaze in real time. This allows it to identify the speaker the user is focusing on.

[0263] 2. The device captures the face of the identified speaker using its camera and sends the video data to the server.

[0264] Estimation of the content of the statement and recognition of the sentiment

[0265] 3. The server analyzes the transmitted speaker's video data and estimates the content of the speech based on the speaker's mouth movements. The estimated text data is then analyzed using a natural language processing model and refined into natural and accurate sentences based on the context.

[0266] 4. The device analyzes the user's facial expressions and voice using emotion recognition technology to identify the user's emotional state. This data is then sent to the server.

[0267] Display of content and emotion

[0268] 5. The server uses the user's sentiment data received from the sentiment engine to adjust the display format of the utterances. It also improves the accuracy of estimating the content of utterances based on the sentiment data.

[0269] 6. The server sends the generated text data and display format information to the terminal.

[0270] 7. The device displays received text data in a speech bubble format on the screen. The display position and design are adjusted based on the user's eye gaze and emotional state.

[0271] Specific example

[0272] Example 1: One-on-one conversation and emotion recognition

[0273] If user A is having a conversation with friend B

[0274] The terminal detects that user A's line of sight is directed at friend B, captures B's face with the camera, and transmits the video to the server.

[0275] The server analyzes the movement of B's mouth, estimates the statement "It's a nice day today", and arranges it with natural language processing.

[0276] The terminal recognizes user A's smiling face and transmits the emotion data to the server.

[0277] The server confirms user A's happy state, arranges the text in a positive display format, and transmits it to the terminal.

[0278] The terminal displays the text "It's a nice day today" in a bright-colored balloon format.

[0279] Example 2: Conversation and emotion recognition among multiple people

[0280] When user A is talking to friends B and C

[0281] The terminal detects that user A's line of sight has moved from B to C, and transmits each video to the server.

[0282] The server analyzes the mouth movements of B and C and estimates the content of each statement.

[0283] The terminal recognizes user A's emotional state and transmits the emotion data to the server.

[0284] The server adjusts the display format based on user A's emotion and transmits it to the terminal.

[0285] The terminal displays the texts "What are your plans for tomorrow?" and "I want to go to the movies" in a balloon format with a design suitable for each position.

[0286] In this way, this system enables communication for hearing-impaired people without using their hands, while also providing displays that take into account the user's emotions. It is expected that this system can reduce information loss and eliminate the user's sense of social isolation.

[0287] The following describes the processing flow.

[0288] Step 1:

[0289] The terminal uses an eye-tracking function to track the user's gaze in real time. It detects the position of the user's pupils and the direction of the gaze, and identifies which direction the user is looking. This data is continuously collected and processed.

[0290] Step 2:

[0291] Based on the collected gaze information, the terminal identifies the object (speaker) that the user is looking at. If there are multiple speakers, the most gazed-at person is selected as the speaker, thereby identifying the appropriate speaker.

[0292] Step 3:

[0293] The terminal captures the face of the identified speaker with a camera and encodes the video data in real time. This video data is converted into an appropriate format and immediately transferred to the server.

[0294] Step 4:

[0295] The server analyzes the received video data. Using Gemini-based lip-reading technology, it extracts the movements of the speaker's mouth and estimates the speech content from those movements. This analysis is performed with high precision and the results are obtained quickly.

[0296] Step 5:

[0297] The server further analyzes the estimated utterance using a ChatGPT-based natural language processing model. This refines the estimated text into a natural and accurate sentence based on the context. This eliminates ambiguity regarding the utterance and provides information in a way that is easy for the user to understand.

[0298] Step 6:

[0299] The device analyzes the user's facial expressions and voice data using emotion recognition technology. This analysis identifies the user's emotional state (e.g., joy, sadness, surprise, etc.) and sends that data to the server.

[0300] Step 7:

[0301] The server uses an emotion engine to analyze the user's emotional data. Based on the analysis results, it adjusts the display format of the user's statements. For example, if the user is happy, it will display the text using positive colors and fonts.

[0302] Step 8:

[0303] The server encodes the final text data and display format information and sends it to the terminal. It uses protocols to ensure data integrity and delivers the data correctly to the terminal.

[0304] Step 9:

[0305] The device decodes the received text data and displays it on the screen in a speech bubble format. The display position and design are appropriately adjusted based on the user's eye gaze and emotional state.

[0306] Step 10:

[0307] Users read speech bubbles displayed on the smart glasses' screen to visually understand what the speaker is saying. This allows them to continue conversations hands-free and communicate comfortably.

[0308] Specific example: One-on-one conversation and emotion recognition

[0309] Steps 1 - 3: While user A is talking to friend B, the terminal identifies that A's line of sight is directed towards B and captures B's face with the camera. The video data is sent to the server.

[0310] Steps 4 - 5: The server analyzes the transmitted video, estimates that B said "It's a nice day today", and arranges it with natural language processing.

[0311] Steps 6 - 7: The terminal analyzes A's expression, recognizes that A is smiling, and sends the emotion data to the server. The server confirms A's state of happiness and arranges the text in a positive display format (e.g., bright colors).

[0312] Steps 8 - 10: The final text data is sent to the terminal and displayed in a pop-up format as "It's a nice day today" on the display. A reads this pop-up and understands B's statement.

[0313] Specific example: Conversation among multiple people and emotion recognition

[0314] Steps 1 - 3: While user A is talking to friends B and C, the terminal detects that A's line of sight moves from B to C, and captures each video. The video data is sent to the server.

[0315] Steps 4 - 5: The server analyzes the mouth movements of B and C, estimates the statement content of "What's your plan for tomorrow?" and "I want to go to the movies", and arranges it with natural language processing.

[0316] Steps 6 - 7: The terminal recognizes A's emotional state and sends the data to the server. The server adjusts the display format based on the emotion data (e.g., colors and positions according to the emotion).

[0317] Steps 8-10: The final text data is sent to the device, and "What are your plans for tomorrow?" and "I want to go to the movies" appear in speech bubbles in the appropriate positions. A reads these speech bubbles and understands what B and C said.

[0318] In this way, this system enables hearing-impaired individuals to communicate without using their hands and presents their statements in an appropriate, emotion-based display format.

[0319] (Example 2)

[0320] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0321] Conventional communication systems for the visually impaired focus on eye tracking and inference of spoken content, but they often fail to adjust the display format to take user emotions into consideration, resulting in a decline in the quality of communication. Furthermore, spoken content may not be presented in a natural, contextual manner, making it difficult to understand.

[0322] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0323] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the statement from the analyzed mouth movements, means for displaying the estimated content of the statement, means for recognizing the user's emotions, and means for adjusting the display format of the content of the statement according to the user's emotions. This makes it possible to display the content of the statement according to the user's emotional state, thereby improving the quality of communication.

[0324] "Means of tracking a user's gaze" refers to all devices, including sensors and software, that detect which direction a user is looking.

[0325] "Means for analyzing a speaker's mouth movements" refers to all devices, including image processing technologies and algorithms, that analyze the movements and shape of a speaker's mouth to estimate what they are saying.

[0326] "Means for estimating the content of speech from analyzed mouth movements" refers to all devices, including speech recognition technology and machine learning models, that use data on mouth movements to estimate what a speaker is saying.

[0327] "Means for displaying estimated speech content" refers to all devices, including displays and projectors, that visually provide users with text and information that has been analyzed and estimated.

[0328] "Means of recognizing user emotions" refers to all devices, including sensors and software, that analyze data such as the user's facial expressions and voice to identify the user's emotional state.

[0329] "Means of adjusting the display format of spoken content according to the user's emotions" refers to all devices, including software and algorithms, that appropriately adjust the visual attributes of the displayed text and information, such as color and font, according to the recognized emotional state.

[0330] "Means of selecting a single speaker from among multiple speakers based on the user's gaze" refers to any device that includes technologies and algorithms for automatically selecting a specific speaker that the user is focusing on, using the user's gaze information, when there are multiple speakers.

[0331] "Means of refining spoken content into natural-sounding sentences based on context" refers to all devices, including natural language processing technologies and generative models, that convert estimated spoken content into more natural and understandable sentences.

[0332] This invention combines a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content with an emotion engine that recognizes the user's emotions. This system allows hearing-impaired individuals to communicate with others without using their hands, and improves the quality of communication by presenting the content of what is being said in a display format that corresponds to the user's emotions.

[0333] System Configuration

[0334] This system consists of three main components: a terminal (e.g., smart glasses), a server, and an emotion engine.

[0335] Device (smart glasses)

[0336] The device has the following features:

[0337] Eye-tracking function: Tracks the user's gaze and collects gaze information. For example, using an "eye-tracking sensor".

[0338] Camera: Captures video of the speaker the user is focusing on. For example, use a "camera device".

[0339] Emotion recognition function: Analyzes the user's facial expressions to identify their emotional state. For example, using "emotion recognition software".

[0340] Display: A method of visually displaying the estimated content of the speech, which is built into the smart glasses.

[0341] server

[0342] The server receives data sent from the terminal and performs the following processing:

[0343] Mouth movement analysis: Analyzes the speaker's mouth movements to estimate the content of their speech. For example, using "speech recognition technology."

[0344] Natural Language Processing: This process involves refining estimated speech content into natural-sounding sentences based on context. For example, it uses "natural language processing techniques."

[0345] Integration with the emotion engine: Use data from the emotion engine to improve the accuracy of estimating the content of speech.

[0346] Emotional Engine

[0347] The emotion engine recognizes the user's emotions and has the following functions:

[0348] Sentiment analysis: Analyzes the user's facial expressions and voice to identify their emotional state.

[0349] Display format adjustment: The display format of the user's comments will change according to their emotions.

[0350] Program processing

[0351] The program for this system processes the data in the following steps: First, the smart glasses worn by the user track the user's gaze using eye-tracking sensors to identify what they are looking at. Next, a camera device captures the face of the identified speaker and sends the video to the server.

[0352] The server uses speech recognition technology to analyze the speaker's mouth movements based on the received video data and estimates what the speaker is saying. Since the estimated speech may sound unnatural in context, natural language processing technology is used to reshape it into more natural-sounding sentences.

[0353] Simultaneously, the smart glasses use emotion recognition software to analyze the user's emotions and send the results to a server. The server uses this emotion data to adjust the display format according to the user's emotions and sends the final text data to the smart glasses.

[0354] Smart glasses display received text data on their screen. In this way, users can obtain information based on their gaze and emotions without using their hands or interacting with others.

[0355] Specific example

[0356] Example 1: One-on-one conversation and emotion recognition

[0357] If user A is having a conversation with friend B:

[0358] The device detects that user A's gaze is directed towards friend B and captures B's face. It then sends the video to the server.

[0359] The server analyzes B's mouth movements and estimates that B said, "It's a nice day today." It then refines this into a more natural-sounding sentence.

[0360] The device recognizes A's smile as an emotion and sends the emotion data to the server.

[0361] The server confirms A's happiness status, formats the text in a positive format, and sends it to the terminal.

[0362] The device displays the text "It's a nice day today" in a brightly colored speech bubble.

[0363] Example 2: Conversation involving multiple people and emotion recognition

[0364] If user A is having a conversation with friends B and C:

[0365] The device detects when user A's gaze moves from B to C, captures the video footage from each location, and sends it to the server.

[0366] The server analyzes the mouth movements of B and C and estimates what each of them is saying.

[0367] The device recognizes A's emotional state and sends the emotional data to the server.

[0368] The server adjusts the display format based on A's emotions and sends it to the terminal.

[0369] The device displays the text "What are your plans for tomorrow?" and "I want to go to the movies" in speech bubble format, with designs appropriate to their respective positions.

[0370] Example of a prompt

[0371] Example 1: Prompts when handling one-on-one conversations

[0372] User A is having a conversation with Friend B. Taking into account User A's gaze and emotional state, transcribe Friend B's statement, "It's a nice day today," and generate the text to display on User A's smart glasses.

[0373] Example 2: Prompts when handling conversations with multiple people

[0374] User A is in a conversation with friend B and friend C. User A's gaze shifts from B to C. Transcribe friend B's statement, "What are your plans for tomorrow?" and friend C's statement, "I want to go to the movies," and generate text to display on the user's smart glasses. Display the text in an appropriate format based on User A's gaze and emotional state.

[0375] In this way, the system integrates user eye tracking, speaker lip-syncing, emotion recognition, and natural language processing and display adjustment of spoken content to provide a high-quality communication experience.

[0376] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0377] Step 1:

[0378] Eye tracking and speaker identification

[0379] Input: User eye-tracking data (real-time location information)

[0380] Processing: The device uses eye-tracking technology to track the user's gaze in real time. It identifies the area the user is fixated on based on their gaze.

[0381] Output: Information about the direction the user is looking.

[0382] Specific operation: In a scenario where the user is talking to friend B, the device uses an eye-tracking sensor to detect the user's gaze direction and identify the face in that direction.

[0383] Step 2:

[0384] Speaker video capture

[0385] Input: User's gaze information, location information of the speaker they are looking at.

[0386] Processing: The device captures video of the identified speaker using its camera and obtains the video data.

[0387] Output: Speaker's video data

[0388] Specific operation: The device uses its camera to capture video of friend B's face, which the user is looking at, and sends that data to the server.

[0389] Step 3:

[0390] Analysis of mouth movements and estimation of spoken content

[0391] Input: Speaker's video data

[0392] Processing: The server uses the received video data and applies speech recognition technology to analyze the speaker's mouth movements. Based on the analysis results, it estimates what the speaker is saying.

[0393] Output: Estimated content of the statement (text data)

[0394] Specific operation: The server analyzes the speaker's mouth movements and estimates the content of the statement, for example, "It's a nice day today." The estimated text data is then retrieved by a generating AI model.

[0395] Step 4:

[0396] Reshaping the content of the statement

[0397] Input: Estimated content of the statement (text data)

[0398] Processing: The server uses a generative AI model to format the spoken content into contextually natural-sounding sentences.

[0399] Output: Formatted speech (natural-sounding text)

[0400] Specific operation: The server uses natural language processing techniques to format text data that it has estimated to be "It's a nice day today" into a natural-sounding sentence. At this stage, it performs processing to adjust sentence endings and context.

[0401] Step 5:

[0402] Recognition of user emotions

[0403] Input: User's facial expression data, voice data

[0404] Processing: The device uses emotion recognition software to analyze the user's facial expressions and voice to identify their current emotional state.

[0405] Output: User emotion data (happiness, sadness, surprise, etc.)

[0406] Specific operation: The device analyzes the user's smile, recognizes that the user is in a happy state, and sends that emotional data to the server.

[0407] Step 6:

[0408] Adjusting the display format

[0409] Input: Formatted speech (natural-sounding text), user sentiment data

[0410] Processing: The server adjusts the display format of the utterances based on the sentiment data obtained. Specifically, it changes visual attributes such as color, font, and background.

[0411] Output: Adjusted display format information

[0412] Specific operation: The server formats the text in a brightly colored speech bubble format based on the user's happy emotional state.

[0413] Step 7:

[0414] Display of the content of the statement

[0415] Input: Adjusted display format information, formatted speech content

[0416] Processing: The terminal displays the content of the message on the screen based on the received text data and display format information.

[0417] Output: The content of the statement that is visually displayed to the user.

[0418] Specific action: The device displays the text "It's a nice day today" to the user in a brightly colored speech bubble.

[0419] Through the processing steps described above, the present invention realizes communication support based on the user's gaze and emotions.

[0420] (Application Example 2)

[0421] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0422] One of the communication challenges for people with hearing impairments is their limited means of quickly understanding what others are saying. Furthermore, to improve the quality of communication, not only visual information but also the recognition of emotions and appropriate feedback are crucial. Especially in physical stores, smooth communication between employees and customers is essential, making a system that allows employees and customers with hearing impairments to communicate smoothly necessary.

[0423] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0424] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, means for displaying the estimated content of the speech, means for recognizing the user's emotions, means for adjusting the display format based on the recognized user's emotions, and means for supporting communication between employees and customers in a physical store. This enables employees to communicate smoothly with customers who are hearing impaired.

[0425] "Means of tracking user gaze" refers to devices or technologies that detect the direction and focus of a user's gaze in real time and acquire data accordingly.

[0426] "Means for analyzing the speaker's mouth movements" refers to devices or technologies that analyze the movements and shape of a speaker's mouth and estimate the content of their speech from that analysis.

[0427] "Methods for estimating the content of a statement from analyzed mouth movements" refers to algorithms or devices that estimate the content of a statement based on data obtained from the speaker's mouth movements.

[0428] "Means for displaying estimated speech content" refers to technologies such as displays and projection devices for visually displaying estimated speech content.

[0429] "Means of recognizing user emotions" refer to sensors and algorithms that identify emotional states from the user's facial expressions, posture, voice, etc.

[0430] "Means for adjusting the display format based on the recognized emotions of the user" refers to devices or technologies for dynamically changing the display format of the content of a statement according to the emotional state of the user.

[0431] "Means of supporting communication between employees and customers in physical stores" refers to systems and applications that support smooth communication between employees and customers in physical stores.

[0432] This invention is a system that supports smooth communication between hearing-impaired customers and employees in physical stores. This system consists of the following main components:

[0433] hardware

[0434] 1. Smart Glasses

[0435] Eye-tracking device: Detects the user's (employee's) gaze direction and point of focus in real time.

[0436] Camera device: Captures video of the speaker's (customer's) face and sends the data to the server.

[0437] Display: Visually displays the estimated content of the statement.

[0438] 2. Server

[0439] Data Analysis Unit: Analyzes the speaker's mouth movements to estimate the content of their speech.

[0440] Natural Language Processing Unit: Reshapes estimated speech content into natural-sounding sentences based on context.

[0441] Emotion Recognition Unit: Analyzes the user's emotional state and adjusts the display format accordingly.

[0442] software

[0443] 1. Frontend Program (JavaScript (registered trademark))

[0444] The smart glasses collect data from various sensors and transmit it to the server in real time.

[0445] 2. Backend program (Python, TENSORFLOW®, OpenCV)

[0446] We perform video data analysis, natural language processing, and emotion recognition.

[0447] Data processing flow

[0448] The server receives user eye-tracking data and speaker video data, analyzes the speaker's mouth movements to estimate the content of the speech. Next, the estimated content is analyzed by a natural language processing unit and refined into natural-sounding sentences. The emotion recognition unit analyzes the user's emotional state and adjusts the display format of the speech content based on that data.

[0449] The smart glasses receive text data sent from a server and display it on the screen in a speech bubble format. The display position and design are adjusted based on the user's gaze information and emotional state.

[0450] Specific example

[0451] As an example, consider a situation where a customer asks about the price of a product.

[0452] 1. When a customer shows a product to an employee wearing smart glasses, the smart glasses' eye-tracking device detects the employee's gaze and captures the customer's face with a camera.

[0453] 2. The server analyzes the video data, estimates the spoken word (e.g., "How much is this product?"), and reshapes it into a natural-sounding sentence.

[0454] 3. The smart glasses' emotion recognition unit identifies the customer's interest from their facial expressions and displays text in a positive format.

[0455] 4. Employees can see clearly and visually appropriate text and respond appropriately to customers.

[0456] Example of inputting prompt text into a generative AI model:

[0457] Please describe a system that supports communication between staff and customers in physical stores. This system involves staff using smart glasses to estimate and visually display what customers are saying, utilizing eye-tracking and emotion recognition capabilities. Please provide a detailed explanation, including specific examples of how it works.

[0458] Thus, this invention is expected to effectively support communication between employees and customers with hearing impairments in physical stores and deepen understanding between them.

[0459] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0460] Step 1:

[0461] The device (smart glasses) uses an eye-tracking device to acquire the user's (employee's) gaze information in real time.

[0462] Input: User eye-tracking data

[0463] Processing: Detection of gaze direction and point of gaze

[0464] Output: Eye-tracking information

[0465] Step 2:

[0466] The device captures the speaker's (customer's) face with a camera based on eye-tracking information and acquires the video data.

[0467] Input: Eye-tracking information

[0468] Processing: Identify the speaker's face and capture the video.

[0469] Output: Video data

[0470] Step 3:

[0471] The device sends the acquired video data to the server.

[0472] Input: Video data

[0473] Processing: Sending video data to the server

[0474] Output: Video data sent to the server

[0475] Step 4:

[0476] The server analyzes the transmitted video data and estimates the content of what the speaker is saying based on their mouth movements.

[0477] Input: Video data

[0478] Processing: Analysis of the speaker's mouth movements and estimation of the content of their speech.

[0479] Output: Estimated content of the statement

[0480] Step 5:

[0481] The server analyzes the estimated speech content using a natural language processing unit and formats it into natural-sounding sentences.

[0482] Input: Estimated content of the statement

[0483] Processing: Refine the text based on context using a natural language processing model.

[0484] Output: A well-formatted text

[0485] Step 6:

[0486] The terminal analyzes the user's (employee's) facial expressions and voice using an emotion recognition unit to identify the user's emotional state. This data is then sent to the server.

[0487] Input: Facial expression data, audio data

[0488] Processing: Identify emotional states using an emotion recognition algorithm.

[0489] Output: Sentiment data

[0490] Step 7:

[0491] The server uses sentiment data to adjust the display format and generate speech content with improved estimation accuracy.

[0492] Input: Sentimental data, formatted text

[0493] Processing: Adjusting display format, regenerating message content

[0494] Output: Adjusted display format and content of the message

[0495] Step 8:

[0496] The server sends the generated text data and display format information to the terminal.

[0497] Input: Adjusted display format and content of the statement

[0498] Processing: Sending text data and display format information to the terminal.

[0499] Output: Text data and display format information sent to the terminal

[0500] Step 9:

[0501] The device displays received text data in a speech bubble format on its screen. The display position and design are adjusted based on the user's eye gaze and emotional state.

[0502] Input: Received text data and display format information

[0503] Processing: Display in speech bubble format, adjust display position and design.

[0504] Output: The content of the statement displayed on the screen

[0505] The above outlines the specific processing steps of the system of the present invention. This enables smooth communication between hearing-impaired customers and employees in physical stores.

[0506] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating 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.

[0507] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0508] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0509] [Second Embodiment]

[0510] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0511] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0512] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0513] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0514] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0515] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0516] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0517] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0518] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0519] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0520] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0521] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. 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".

[0522] This invention relates to a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content. More specifically, it relates to a system that enables hearing-impaired individuals to communicate with others without using their hands and to easily understand conversations with multiple speakers.

[0523] System Configuration

[0524] This system consists of the following main components:

[0525] 1. Device (smart glasses)

[0526] 2. Server

[0527] 3. User

[0528] terminal

[0529] The device takes the form of smart glasses. The device is equipped with the following functions:

[0530] Eye-tracking function: Tracks the user's gaze and collects eye-tracking information.

[0531] Camera: Captures video of the speaker the user is focusing on.

[0532] Display: Visually displays the estimated content of the statement.

[0533] server

[0534] The server receives data sent from the terminal and performs the following processing:

[0535] Mouth movement analysis: Analyze the speaker's mouth movements to estimate the content of their speech.

[0536] Natural language processing: Refining estimated speech content into natural-sounding sentences based on context.

[0537] Data transmission: The estimated content of the statement is sent to the terminal.

[0538] Program processing

[0539] Eye tracking and speaker identification

[0540] 1. The device uses eye-tracking technology to track the user's gaze in real time. This allows it to identify the speaker the user is focusing on.

[0541] 2. The device captures video of the identified speaker using its camera and sends that data to the server.

[0542] Estimation of the content of the statement

[0543] 3. The server receives the transmitted speaker's video data and analyzes the speaker's mouth movements using Gemini-based lip-reading technology.

[0544] 4. The server accurately estimates the content of speech based on lip-sync data. The estimated text data is analyzed using a ChatGPT-based natural language processing model to refine it into natural and accurate sentences based on the context.

[0545] Display of the content of the statement

[0546] 5. The server sends the generated text data to the terminal.

[0547] 6. The device displays the received text data on the screen in a speech bubble format. The display position is adjusted based on the user's eye-tracking information.

[0548] Specific example

[0549] Example 1: One-on-one conversation

[0550] If user A is having a conversation with friend B

[0551] The device detects that user A's gaze is directed towards friend B.

[0552] The device captures friend B's face with its camera and sends the video to the server.

[0553] The server analyzes friend B's mouth movements and estimates that he said, "It's a nice day today."

[0554] The estimated text is processed using natural language processing on the server and then sent to user A's terminal.

[0555] The device displays the text "It's a nice day today" in a speech bubble.

[0556] Example 2: Conversation involving multiple people

[0557] If user A is having a conversation with friends B and C

[0558] The device detects that user A's gaze has shifted from B to C.

[0559] Based on each user's gaze data, the device sends the corresponding video to the server.

[0560] The server estimates the content of what B and C said and formats it as text data.

[0561] The device displays B's statement, "What are your plans for tomorrow?", and C's statement, "I want to go to the movies," in speech bubble format at their respective positions on the screen.

[0562] In this way, this system enables hearing-impaired individuals to communicate smoothly with others without using their hands. As a result, users will not miss important information and will be able to live a more natural life without feeling socially isolated.

[0563] The following describes the processing flow.

[0564] Step 1:

[0565] The device uses eye-tracking technology to track the user's gaze. It detects the position of the user's pupils and the direction of their gaze to determine which direction they are looking.

[0566] Step 2:

[0567] The device identifies the person (speaker) the user is focusing on based on eye-tracking information. If there are multiple speakers, the person being focused on the most is selected as the speaker.

[0568] Step 3:

[0569] The device captures the face of the identified speaker using its camera. The acquired video data is processed in real time, so it is captured continuously without interruption.

[0570] Step 4:

[0571] The terminal encodes the captured video data and sends it to the server over the network. A high-speed communication protocol is used to minimize latency during this process.

[0572] Step 5:

[0573] The server analyzes the received video data using Gemini-based lip-reading technology. It extracts the speaker's mouth movements and estimates the content of their speech with high accuracy based on those movements.

[0574] Step 6:

[0575] The server further analyzes the estimated speech content using a ChatGPT-based natural language processing model. This refines the estimated text into a natural and accurate sentence based on the context.

[0576] Step 7:

[0577] The server encodes the generated text data and sends it to the terminal over the network. It applies protocols to ensure data integrity during transmission.

[0578] Step 8:

[0579] The device decodes the received text data and displays it on the screen in a speech bubble format. The display position is adjusted based on the user's eye gaze and the speaker's position.

[0580] Step 9:

[0581] Users read speech bubbles displayed on the smart glasses' screen to visually understand what the speaker is saying. This allows them to grasp the conversation without using their hands.

[0582] (Example 1)

[0583] Next, we will describe Example 1. 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."

[0584] There is a need for a system that allows people with hearing impairments to communicate with others without using their hands. In particular, the challenge lies in smoothly understanding conversations and facilitating smooth communication, even when multiple speakers are involved. This requires advanced processing such as eye tracking, lip-syncing, speech content estimation, and natural language processing for text formatting, and there is a demand for a system that can achieve these capabilities.

[0585] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0586] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, means for displaying the estimated content of the speech, means for collecting video data from a camera, and means for processing the estimated content of the speech into natural-sounding text using natural language processing. This makes it possible for people with hearing impairments to obtain a lot of information without using their hands and to communicate smoothly with others.

[0587] A "user" is a person who uses a system to communicate.

[0588] A "means of tracking eye movements" refers to a device that has the function of detecting and tracking a user's gaze in real time.

[0589] The "speaker" is the person the user is focusing their gaze on and who is speaking.

[0590] A "means for analyzing mouth movements" is a device that analyzes the movements of a speaker's mouth and estimates the content of what they are saying from those movements.

[0591] A "means for estimating the content of speech" is a device that estimates a speaker's speech as text data based on analyzed lip-movement data.

[0592] "Means for displaying the content of a statement" refers to a device for visually conveying the estimated content of a statement to the user.

[0593] "Means for collecting video data using a camera" refers to a device for capturing video of a speaker and collecting that data.

[0594] "Natural language processing" is a technique for shaping estimated speech content into natural-sounding sentences based on context.

[0595] A "server" is a device that receives data transmitted from a terminal, performs analysis and estimation, and transmits data.

[0596] A "device" is a device worn by the user that tracks their gaze, captures video with a camera, and transmits data to a server.

[0597] This invention relates to a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the results. Specifically, it is a system that enables hearing-impaired individuals to communicate smoothly with others without using their hands.

[0598] System Configuration

[0599] This system is mainly composed of the following components:

[0600] 1. Device (smart glasses)

[0601] 2. Server

[0602] 3. User

[0603] terminal

[0604] The device takes the form of smart glasses and has the following functions:

[0605] Eye-tracking function: Tracks the user's gaze in real time and collects gaze information.

[0606] Camera: Captures video of the speaker the user is focusing on.

[0607] Display: Visually displays the estimated content of the statement.

[0608] server

[0609] The server receives data sent from the terminal and performs the following processing:

[0610] Mouth movement analysis: The speaker's mouth movements are analyzed using Gemini-based technology.

[0611] Natural Language Processing: The estimated utterance content is refined based on context using a ChatGPT-based natural language processing model.

[0612] Data transmission: The estimated content of the statement is sent to the terminal.

[0613] System operation

[0614] Eye tracking and speaker identification

[0615] The device uses eye-tracking technology to track the user's gaze. For example, when a user looks at someone they are talking to, that gaze information is instantly collected. This allows the device to identify the speaker the user is focusing on.

[0616] Data capture and transmission

[0617] The device captures video of the identified speaker using its built-in camera. The captured video data is sent to a server. This transmission uses high-speed communication technology (e.g., a 5G network) to process the data in real time.

[0618] Processing of video data and analysis of mouth movements

[0619] The server receives video data transmitted from the terminal and analyzes it using Gemini-based lip-reading technology. The server analyzes the speaker's mouth movements in detail from the video data and extracts the content of the speech from those movements.

[0620] Estimation of spoken content and natural language processing

[0621] The server accurately estimates the content of the speech based on the analyzed mouth movement data. The estimated content is then formatted into natural-sounding sentences using a ChatGPT-based natural language processing model.

[0622] Sending and displaying estimated results

[0623] The server sends the formatted text to the terminal. The terminal displays the received text data in a speech bubble format on its screen. The display position is adjusted based on the user's eye gaze to ensure that the text is intuitively understandable.

[0624] Specific example

[0625] Example 1: One-on-one conversation

[0626] If user A is having a conversation with friend B:

[0627] The device detects that user A's gaze is directed towards friend B.

[0628] The device captures friend B's face with its camera and sends the video to the server.

[0629] The server analyzes friend B's mouth movements and estimates that he said, "It's a nice day today."

[0630] The estimated text is processed using natural language processing on the server and then sent to user A's terminal.

[0631] The device displays the text "It's a nice day today" in a speech bubble.

[0632] Example 2: Conversation involving multiple people

[0633] If user A is having a conversation with friends B and C:

[0634] The device detects that user A's gaze has shifted from B to C.

[0635] Based on each user's gaze data, the device sends the corresponding video to the server.

[0636] The server estimates the content of what B and C said and formats it as text data.

[0637] The device displays B's statement, "What are your plans for tomorrow?", and C's statement, "I want to go to the movies," in speech bubble format at their respective positions on the screen.

[0638] Example of a prompt

[0639] Please describe a system that tracks the user's gaze and analyzes the speaker's mouth movements to estimate what they are saying. Explain this using specific examples of one-on-one and group conversations.

[0640] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0641] Step 1:

[0642] The device uses eye-tracking technology to track the user's gaze in real time. The input is the user's gaze information, and the main output is coordinate data of the direction the gaze is directed. Based on this coordinate data, the device identifies the speaker the user is focusing on. A high-precision eye-tracking sensor built into the device captures even the slightest movement of the gaze and collects it as gaze information.

[0643] Step 2:

[0644] The device captures video of the speaker, identified based on eye-tracking information, using its built-in camera. The input is the speaker's location information identified by eye tracking, and the output is the captured video data. The built-in camera acquires high-resolution video and collects video data by focusing on the speaker's mouth. The collected video data is processed in real time and transmitted to the server.

[0645] Step 3:

[0646] The server receives video data transmitted from the terminal. The input is video data transmitted from the terminal, and the output is data in which the speaker's mouth movements have been analyzed. The server analyzes the video data using Gemini-based lip-reading technology and extracts specific movements that form words from the speaker's mouth movements. It performs noise reduction and correction on the data to achieve highly accurate mouth movement analysis.

[0647] Step 4:

[0648] The server estimates the speaker's speech based on analyzed lip-sync data. The input is lip-sync data, and the output is estimated text data. The server uses machine learning algorithms to estimate the actual spoken words with high accuracy. The estimated speech content is then formatted using contextual information and known language models.

[0649] Step 5:

[0650] The server uses a ChatGPT-based natural language processing model to format the estimated utterance into natural-sounding text. The input is estimated text data, and the output is context-formatted natural-sounding text. The server performs contextual analysis and formats the text as natural dialogue.

[0651] Step 6:

[0652] The server sends the formatted speech to the terminal. The input is formatted, natural-sounding text, and the output is text data sent to the terminal. The data is compressed and transmitted efficiently using high-speed communication technology (e.g., 5G network).

[0653] Step 7:

[0654] The terminal displays received text data in a speech bubble format on its screen. The input is text data sent from the server, and the output is text information visually displayed to the user. The display adjusts its position based on the user's eye gaze, positioning the text in an intuitively understandable location.

[0655] (Application Example 1)

[0656] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0657] When hearing-impaired individuals or workers have difficulty communicating effectively, it can be particularly challenging in work environments such as factories to accurately understand work instructions, potentially compromising safety and efficiency. This invention aims to solve these problems and provide a system that supports users in easily communicating with others.

[0658] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0659] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, means for displaying the estimated content of the speech, means for visually displaying work instructions, and means for facilitating smooth communication with robots and other workers. As a result, the user can visually confirm the content of the speech of the speaker they are looking at, and can also clearly and quickly grasp work instructions within the factory.

[0660] A "user" is an individual who uses the system to utilize eye-tracking and speech content estimation features.

[0661] "Methods for tracking eye movements" refer to devices and technologies that detect the direction of a user's gaze in real time and acquire that data.

[0662] The "speaker" is the person the user is looking at, and whose mouth movements are being analyzed.

[0663] "Methods for analyzing mouth movements" refer to signal processing and algorithms that detect mouth movements based on video data captured by a camera and estimate the content of speech based on those movements.

[0664] "Means for estimating the content of speech" refers to methods and techniques that analyze the speaker's lip-sync data to infer what the speaker is saying.

[0665] "Means for displaying estimated speech content" refers to displays or devices that visually present estimated text to the user.

[0666] "Means of visually displaying work instructions" refers to a system that displays and presents work instructions and announcements within a factory within the user's field of vision.

[0667] A "robot" is a mechanical device that operates autonomously or remotely and performs tasks within a factory.

[0668] "Means of facilitating smooth communication" refers to methods and technologies for seamlessly transferring information between users, robots, and other workers.

[0669] This invention is a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content. This system is particularly intended to support factory workers in facilitating smooth communication with robots and other workers.

[0670] System Configuration

[0671] This system consists of the following main components:

[0672] 1. Device (smart glasses)

[0673] 2. Server

[0674] 3. User

[0675] terminal

[0676] The device takes the form of smart glasses. The device includes the following functions:

[0677] Eye-tracking function: Tracks the user's gaze and collects eye-tracking information.

[0678] Camera: Captures video of the speaker the user is focusing on.

[0679] Display: Visually displays estimated spoken content and work instructions.

[0680] server

[0681] The server receives data sent from the terminal and performs the following processing:

[0682] Mouth movement analysis: Analyze the speaker's mouth movements to estimate the content of their speech.

[0683] Natural language processing: Refining estimated speech content into natural-sounding sentences based on context.

[0684] Data transmission: Estimated spoken content and work instructions are sent to the terminal.

[0685] Program Description

[0686] The server processes data using Python, OpenCV, and dlib. Eye-tracking and camera data are acquired from sensors in smart glasses. The speaker's mouth movements are detected and analyzed using the dlib library. The estimated speech is formatted using natural language processing techniques such as Google Text-to-Speech (gTTS) and displayed on the screen.

[0687] Specific example

[0688] Example 1: Work instructions given in a one-on-one conversation

[0689] A scene where worker A receives instructions from robot operator B in a factory:

[0690] The device detects that A's gaze is directed towards B.

[0691] The device captures B's face with its camera and sends the video to the server.

[0692] The server analyzes B's mouth movements and infers that he said, "Please attach the parts."

[0693] The estimated text is processed using natural language processing and then sent to terminal A.

[0694] The device displays the text "Please install the parts" on its screen.

[0695] Example 2: Communication with multiple speakers

[0696] A scene in a factory where worker A interacts with multiple robot operators C and D:

[0697] The device detects that A's gaze has shifted from C to D.

[0698] The corresponding video is sent to the server, and the mouth movements of each person are analyzed.

[0699] The server infers the meaning of C's statement, "Proceed to the next step," and D's statement, "Check the parts."

[0700] The device displays each statement on its screen.

[0701] Examples of prompts to input into a generative AI model

[0702] TXT

[0703] Person A, working in a factory, is wearing smart glasses. Operator B says, "Next, please attach this part." The smart glasses track Person A's gaze and analyze B's mouth movements, displaying the spoken words as text. The smart glasses' display shows "Next, please attach this part," and Person A understands the instructions accurately.

[0704] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0705] Step 1:

[0706] The device tracks the user's gaze in real time using eye-tracking technology. The input is the user's gaze information, and the output is data on the direction the gaze is directed. This gaze information is collected using built-in sensors and a camera.

[0707] Step 2:

[0708] The device captures the face of the person it is looking at using its camera. The input is gaze information and camera footage, and the output is video data of the person the device is looking at.

[0709] Step 3:

[0710] The terminal sends the captured video data to the server. The input is video data, and the output is video data transmitted over the network.

[0711] Step 4:

[0712] The server analyzes the received video data and detects the speaker's mouth movements. The input is the transmitted video data, and the output is the analysis data regarding the mouth movements. The dlib library is used to detect the mouth movements.

[0713] Step 5:

[0714] The server estimates the content of a speaker's speech based on analysis data of their mouth movements. The input is mouth movement data, and the output is estimated text data. Natural language processing technology (generative AI model) is used for the estimation.

[0715] Step 6:

[0716] The server reshapes the estimated text data into natural-sounding sentences based on context. The input is estimated text data, and the output is contextually reshaped sentences.

[0717] Step 7:

[0718] The server sends formatted text data to the terminal. The input is formatted text data, and the output is text data transmitted over the network.

[0719] Step 8:

[0720] The device displays received text data on its screen. The input is text data sent from the server, and the output is text displayed on the smart glasses' screen. The display format is adjusted based on the user's eye-tracking information.

[0721] Step 9:

[0722] The user visually confirms the spoken content displayed on the smart glasses' screen. This action allows the user to understand the robot operator's instructions and work commands.

[0723] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0724] This invention is a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content, further combining it with an emotion engine that recognizes the user's emotions. This system allows hearing-impaired individuals to communicate with others without using their hands, and can improve the quality of communication by presenting the content of what is being said in a display format that corresponds to the user's emotions. Specifically, it has the following configuration and processing flow.

[0725] System Configuration

[0726] This system consists of the following main components:

[0727] 1. Device (smart glasses)

[0728] 2. Server

[0729] 3. Emotional Engine

[0730] terminal

[0731] The device takes the form of smart glasses and is equipped with the following functions:

[0732] Eye-tracking function: Tracks the user's gaze and collects eye-tracking information.

[0733] Camera: Captures video of the speaker the user is focusing on.

[0734] Emotion recognition function: Analyzes the user's emotions from their facial expressions, voice, etc.

[0735] Display: Visually displays the estimated content of the statement.

[0736] server

[0737] The server receives data sent from the terminal and performs the following processing:

[0738] Mouth movement analysis: Analyze the speaker's mouth movements to estimate the content of their speech.

[0739] Natural language processing: Refining estimated speech content into natural-sounding sentences based on context.

[0740] Integration with the emotion engine: Use data from the emotion engine to improve the accuracy of estimating the content of speech.

[0741] Emotional Engine

[0742] The emotion engine recognizes the user's emotions and has the following functions:

[0743] Sentiment analysis: Analyzes the user's facial expressions and voice to identify their emotional state.

[0744] Display format adjustment: The display format of the user's comments will change according to their emotions.

[0745] Program processing

[0746] Eye tracking and speaker identification

[0747] 1. The device uses eye-tracking technology to track the user's gaze in real time. This allows it to identify the speaker the user is focusing on.

[0748] 2. The device captures the face of the identified speaker using its camera and sends the video data to the server.

[0749] Estimation of the content of the statement and recognition of the sentiment

[0750] 3. The server analyzes the transmitted speaker's video data and estimates the content of the speech based on the speaker's mouth movements. The estimated text data is then analyzed using a natural language processing model and refined into natural and accurate sentences based on the context.

[0751] 4. The device analyzes the user's facial expressions and voice using emotion recognition technology to identify the user's emotional state. This data is then sent to the server.

[0752] Display of content and emotion

[0753] 5. The server uses the user's sentiment data received from the sentiment engine to adjust the display format of the utterances. It also improves the accuracy of estimating the content of utterances based on the sentiment data.

[0754] 6. The server sends the generated text data and display format information to the terminal.

[0755] 7. The device displays received text data in a speech bubble format on the screen. The display position and design are adjusted based on the user's eye gaze and emotional state.

[0756] Specific example

[0757] Example 1: One-on-one conversation and emotion recognition

[0758] If user A is having a conversation with friend B

[0759] The device detects that user A's gaze is directed towards friend B, captures B's face with its camera, and sends the video to the server.

[0760] The server analyzes B's mouth movements, estimates that B is saying "It's a nice day today," and then refines it using natural language processing.

[0761] The device recognizes A's smile as an emotion and sends the emotion data to the server.

[0762] The server confirms A's happiness status, formats the text in a positive format, and sends it to the terminal.

[0763] The device displays the text "It's a nice day today" in a brightly colored speech bubble.

[0764] Example 2: Conversation involving multiple people and emotion recognition

[0765] If user A is having a conversation with friends B and C

[0766] The device detects when user A's gaze moves from B to C and sends the respective video feeds to the server.

[0767] The server analyzes the mouth movements of B and C and estimates what each of them is saying.

[0768] The device recognizes A's emotional state and sends the emotional data to the server.

[0769] The server adjusts the display format based on A's emotions and sends it to the terminal.

[0770] The device displays the text "What are your plans for tomorrow?" and "I want to go to the movies" in speech bubble format, with designs appropriate to their respective positions.

[0771] In this way, this system allows hearing-impaired individuals to communicate without using their hands, while also displaying information in a way that takes the user's emotions into consideration. This system is expected to reduce information loss and alleviate users' feelings of social isolation.

[0772] The following describes the processing flow.

[0773] Step 1:

[0774] The device uses eye-tracking technology to track the user's gaze in real time. It detects the position of the user's pupils and the direction of their gaze to determine which direction they are looking. This data is continuously collected and processed.

[0775] Step 2:

[0776] The device identifies the person (speaker) the user is focusing on based on the collected eye-tracking data. If there are multiple speakers, it selects the person being focused on most as the speaker, thereby identifying the appropriate speaker.

[0777] Step 3:

[0778] The device captures the face of the identified speaker with its camera and encodes the video data in real time. This video data is converted to the appropriate format and immediately transferred to the server.

[0779] Step 4:

[0780] The server analyzes the received video data. Using Gemini-based lip-reading technology, it extracts the speaker's mouth movements and estimates the content of their speech from those movements. This analysis is highly accurate and provides results quickly.

[0781] Step 5:

[0782] The server further analyzes the estimated utterance using a ChatGPT-based natural language processing model. This refines the estimated text into a natural and accurate sentence based on the context. This eliminates ambiguity regarding the utterance and provides information in a way that is easy for the user to understand.

[0783] Step 6:

[0784] The device analyzes the user's facial expressions and voice data using emotion recognition technology. This analysis identifies the user's emotional state (e.g., joy, sadness, surprise, etc.) and sends that data to the server.

[0785] Step 7:

[0786] The server uses an emotion engine to analyze the user's emotional data. Based on the analysis results, it adjusts the display format of the user's statements. For example, if the user is happy, it will display the text using positive colors and fonts.

[0787] Step 8:

[0788] The server encodes the final text data and display format information and sends it to the terminal. It uses protocols to ensure data integrity and delivers the data correctly to the terminal.

[0789] Step 9:

[0790] The device decodes the received text data and displays it on the screen in a speech bubble format. The display position and design are appropriately adjusted based on the user's eye gaze and emotional state.

[0791] Step 10:

[0792] Users read speech bubbles displayed on the smart glasses' screen to visually understand what the speaker is saying. This allows them to continue conversations hands-free and communicate comfortably.

[0793] Specific example: One-on-one conversation and emotion recognition

[0794] Steps 1-3: While user A is talking to friend B, the device detects that A's gaze is directed towards B and captures B's face with its camera. The video data is sent to the server.

[0795] Steps 4-5: The server analyzes the transmitted video, estimates that B said "It's a nice day today," and then processes it using natural language processing.

[0796] Steps 6-7: The device analyzes A's facial expression, recognizes that A is smiling, and sends the emotion data to the server. The server confirms A's happiness level and formats the text in a positive format (e.g., bright colors).

[0797] Steps 8-10: The final text data is sent to the terminal and displayed on the screen in the form of a speech bubble saying "It's a nice day today." A reads this speech bubble and understands what B said.

[0798] Specific example: Conversation involving multiple people and emotion recognition

[0799] Steps 1-3: While user A is talking to friends B and C, the device detects when A's gaze shifts from B to C and captures video of each. The video data is sent to the server.

[0800] Steps 4-5: The server analyzes the mouth movements of B and C, estimates the content of their statements, "What are your plans for tomorrow?" and "I want to go to the movies," and processes them using natural language processing.

[0801] Steps 6-7: The device recognizes A's emotional state and sends that data to the server. The server adjusts the display format based on the emotional data (e.g., color and position according to the emotion).

[0802] Steps 8-10: The final text data is sent to the device, and "What are your plans for tomorrow?" and "I want to go to the movies" appear in speech bubbles in the appropriate positions. A reads these speech bubbles and understands what B and C said.

[0803] In this way, this system enables hearing-impaired individuals to communicate without using their hands and presents their statements in an appropriate, emotion-based display format.

[0804] (Example 2)

[0805] Next, we will describe Example 2. 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".

[0806] Conventional communication systems for the visually impaired focus on eye tracking and inference of spoken content, but they often fail to adjust the display format to take user emotions into consideration, resulting in a decline in the quality of communication. Furthermore, spoken content may not be presented in a natural, contextual manner, making it difficult to understand.

[0807] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0808] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the statement from the analyzed mouth movements, means for displaying the estimated content of the statement, means for recognizing the user's emotions, and means for adjusting the display format of the content of the statement according to the user's emotions. This makes it possible to display the content of the statement according to the user's emotional state, thereby improving the quality of communication.

[0809] "Means of tracking a user's gaze" refers to all devices, including sensors and software, that detect which direction a user is looking.

[0810] "Means for analyzing a speaker's mouth movements" refers to all devices, including image processing technologies and algorithms, that analyze the movements and shape of a speaker's mouth to estimate what they are saying.

[0811] "Means for estimating the content of speech from analyzed mouth movements" refers to all devices, including speech recognition technology and machine learning models, that use data on mouth movements to estimate what a speaker is saying.

[0812] "Means for displaying estimated speech content" refers to all devices, including displays and projectors, that visually provide users with text and information that has been analyzed and estimated.

[0813] "Means of recognizing user emotions" refers to all devices, including sensors and software, that analyze data such as the user's facial expressions and voice to identify the user's emotional state.

[0814] "Means of adjusting the display format of spoken content according to the user's emotions" refers to all devices, including software and algorithms, that appropriately adjust the visual attributes of the displayed text and information, such as color and font, according to the recognized emotional state.

[0815] "Means of selecting a single speaker from among multiple speakers based on the user's gaze" refers to any device that includes technologies and algorithms for automatically selecting a specific speaker that the user is focusing on, using the user's gaze information, when there are multiple speakers.

[0816] "Means of refining spoken content into natural-sounding sentences based on context" refers to all devices, including natural language processing technologies and generative models, that convert estimated spoken content into more natural and understandable sentences.

[0817] This invention combines a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content with an emotion engine that recognizes the user's emotions. This system allows hearing-impaired individuals to communicate with others without using their hands, and improves the quality of communication by presenting the content of what is being said in a display format that corresponds to the user's emotions.

[0818] System Configuration

[0819] This system consists of three main components: a terminal (e.g., smart glasses), a server, and an emotion engine.

[0820] Device (smart glasses)

[0821] The device has the following features:

[0822] Eye-tracking function: Tracks the user's gaze and collects gaze information. For example, using an "eye-tracking sensor".

[0823] Camera: Captures video of the speaker the user is focusing on. For example, use a "camera device".

[0824] Emotion recognition function: Analyzes the user's facial expressions to identify their emotional state. For example, using "emotion recognition software".

[0825] Display: A method of visually displaying the estimated content of the speech, which is built into the smart glasses.

[0826] server

[0827] The server receives data sent from the terminal and performs the following processing:

[0828] Mouth movement analysis: Analyzes the speaker's mouth movements to estimate the content of their speech. For example, using "speech recognition technology."

[0829] Natural Language Processing: This process reshapes estimated speech content into natural-sounding sentences based on context. For example, it uses "natural language processing techniques."

[0830] Integration with the emotion engine: Use data from the emotion engine to improve the accuracy of estimating the content of speech.

[0831] Emotional Engine

[0832] The emotion engine recognizes the user's emotions and has the following functions:

[0833] Sentiment analysis: Analyzes the user's facial expressions and voice to identify their emotional state.

[0834] Display format adjustment: The display format of the user's comments will change according to their emotions.

[0835] Program processing

[0836] The program for this system processes the data in the following steps: First, the smart glasses worn by the user track the user's gaze using eye-tracking sensors to identify what they are looking at. Next, a camera device captures the face of the identified speaker and sends the video to the server.

[0837] The server uses speech recognition technology to analyze the speaker's mouth movements based on the received video data and estimates the content of their speech. Since the estimated content may sound unnatural in context, natural language processing technology is used to reshape it into more natural-sounding sentences.

[0838] Simultaneously, the smart glasses use emotion recognition software to analyze the user's emotions and send the results to a server. The server uses this emotion data to adjust the display format according to the user's emotions and then sends the final text data to the smart glasses.

[0839] Smart glasses display received text data on their screen. In this way, users can obtain information based on their gaze and emotions without using their hands or interacting with others.

[0840] Specific example

[0841] Example 1: One-on-one conversation and emotion recognition

[0842] If user A is having a conversation with friend B:

[0843] The device detects that user A's gaze is directed towards friend B and captures B's face. It then sends the video to the server.

[0844] The server analyzes B's mouth movements and estimates that B said, "It's a nice day today." It then refines this into a more natural-sounding sentence.

[0845] The device recognizes A's smile as an emotion and sends the emotion data to the server.

[0846] The server confirms A's happiness status, formats the text in a positive format, and sends it to the terminal.

[0847] The device displays the text "It's a nice day today" in a brightly colored speech bubble.

[0848] Example 2: Conversation involving multiple people and emotion recognition

[0849] If user A is having a conversation with friends B and C:

[0850] The device detects when user A's gaze moves from B to C, captures the video footage from each location, and sends it to the server.

[0851] The server analyzes the mouth movements of B and C and estimates what each of them is saying.

[0852] The device recognizes A's emotional state and sends the emotional data to the server.

[0853] The server adjusts the display format based on A's emotions and sends it to the terminal.

[0854] The device displays the text "What are your plans for tomorrow?" and "I want to go to the movies" in speech bubble format, with designs appropriate to their respective positions.

[0855] Example of a prompt

[0856] Example 1: Prompts when handling one-on-one conversations

[0857] User A is having a conversation with Friend B. Taking into account User A's gaze and emotional state, transcribe Friend B's statement, "It's a nice day today," and generate the text to display on User A's smart glasses.

[0858] Example 2: Prompts when handling conversations with multiple people

[0859] User A is in a conversation with friend B and friend C. User A's gaze shifts from B to C. Transcribe friend B's statement, "What are your plans for tomorrow?" and friend C's statement, "I want to go to the movies," and generate text to display on the user's smart glasses. Display the text in an appropriate format based on User A's gaze and emotional state.

[0860] In this way, the system integrates user eye tracking, speaker lip-syncing, emotion recognition, and natural language processing and display adjustment of spoken content to provide a high-quality communication experience.

[0861] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0862] Step 1:

[0863] Eye tracking and speaker identification

[0864] Input: User eye-tracking data (real-time location information)

[0865] Processing: The device uses eye-tracking technology to track the user's gaze in real time. It identifies the area the user is fixated on based on their gaze.

[0866] Output: Information about the direction the user is looking.

[0867] Specific operation: In a scenario where the user is talking to friend B, the device uses an eye-tracking sensor to detect the user's gaze direction and identify the face in that direction.

[0868] Step 2:

[0869] Speaker video capture

[0870] Input: User's gaze information, location information of the speaker they are looking at.

[0871] Processing: The device captures video of the identified speaker using its camera and obtains the video data.

[0872] Output: Speaker's video data

[0873] Specific operation: The device uses its camera to capture video of friend B's face, which the user is looking at, and sends that data to the server.

[0874] Step 3:

[0875] Analysis of mouth movements and estimation of spoken content

[0876] Input: Speaker's video data

[0877] Processing: The server uses the received video data and applies speech recognition technology to analyze the speaker's mouth movements. Based on the analysis results, it estimates what the speaker is saying.

[0878] Output: Estimated content of the statement (text data)

[0879] Specific operation: The server analyzes the speaker's mouth movements and estimates the content of the statement, for example, "It's a nice day today." The estimated text data is then retrieved by a generating AI model.

[0880] Step 4:

[0881] Reshaping the content of the statement

[0882] Input: Estimated content of the statement (text data)

[0883] Processing: The server uses a generative AI model to format the spoken content into contextually natural-sounding sentences.

[0884] Output: Formatted speech (natural-sounding text)

[0885] Specific operation: The server uses natural language processing techniques to format text data that it has estimated to be "It's a nice day today" into a natural-sounding sentence. At this stage, it performs processing to adjust sentence endings and context.

[0886] Step 5:

[0887] Recognition of user emotions

[0888] Input: User's facial expression data, voice data

[0889] Processing: The device uses emotion recognition software to analyze the user's facial expressions and voice to identify their current emotional state.

[0890] Output: User emotion data (happiness, sadness, surprise, etc.)

[0891] Specific operation: The device analyzes the user's smile, recognizes that the user is in a happy state, and sends that emotional data to the server.

[0892] Step 6:

[0893] Adjusting the display format

[0894] Input: Formatted speech (natural-sounding text), user sentiment data

[0895] Processing: The server adjusts the display format of the utterances based on the sentiment data obtained. Specifically, it changes visual attributes such as color, font, and background.

[0896] Output: Adjusted display format information

[0897] Specific operation: The server formats the text in a brightly colored speech bubble format based on the user's happy emotional state.

[0898] Step 7:

[0899] Display of the content of the statement

[0900] Input: Adjusted display format information, formatted speech content

[0901] Processing: The terminal displays the content of the message on the screen based on the received text data and display format information.

[0902] Output: The content of the statement that is visually displayed to the user.

[0903] Specific action: The device displays the text "It's a nice day today" to the user in a brightly colored speech bubble.

[0904] Through the processing steps described above, the present invention realizes communication support based on the user's gaze and emotions.

[0905] (Application Example 2)

[0906] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0907] One of the communication challenges for people with hearing impairments is their limited means of quickly understanding what others are saying. Furthermore, to improve the quality of communication, not only visual information but also the recognition of emotions and appropriate feedback are crucial. Especially in physical stores, smooth communication between employees and customers is essential, making a system that allows employees and customers with hearing impairments to communicate smoothly necessary.

[0908] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0909] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, means for displaying the estimated content of the speech, means for recognizing the user's emotions, means for adjusting the display format based on the recognized user's emotions, and means for supporting communication between employees and customers in a physical store. This enables employees to communicate smoothly with customers who are hearing impaired.

[0910] "Means of tracking a user's gaze" refers to devices or technologies that detect the direction and focus of a user's gaze in real time and acquire data accordingly.

[0911] "Means for analyzing the speaker's mouth movements" refers to devices or technologies that analyze the movements and shape of a speaker's mouth and estimate the content of their speech from that analysis.

[0912] "Methods for estimating the content of a statement from analyzed mouth movements" refers to algorithms or devices that estimate the content of a statement based on data obtained from the speaker's mouth movements.

[0913] "Means for displaying estimated speech content" refers to technologies such as displays and projection devices for visually displaying estimated speech content.

[0914] "Means of recognizing user emotions" refer to sensors and algorithms that identify emotional states from the user's facial expressions, posture, voice, etc.

[0915] "Means for adjusting the display format based on the recognized user's emotions" refers to devices or technologies for dynamically changing the display format of a user's statements according to their emotional state.

[0916] "Means of supporting communication between employees and customers in physical stores" refers to systems and applications that support smooth communication between employees and customers in physical stores.

[0917] This invention is a system that supports smooth communication between hearing-impaired customers and employees in physical stores. This system consists of the following main components:

[0918] hardware

[0919] 1. Smart Glasses

[0920] Eye-tracking device: Detects the user's (employee's) gaze direction and point of focus in real time.

[0921] Camera device: Captures video of the speaker's (customer's) face and sends the data to the server.

[0922] Display: Visually displays the estimated content of the statement.

[0923] 2. Server

[0924] Data Analysis Unit: Analyzes the speaker's mouth movements to estimate the content of their speech.

[0925] Natural Language Processing Unit: Reshapes estimated speech content into natural-sounding sentences based on context.

[0926] Emotion Recognition Unit: Analyzes the user's emotional state and adjusts the display format accordingly.

[0927] software

[0928] 1. Frontend Program (JavaScript)

[0929] The smart glasses collect data from various sensors and transmit it to the server in real time.

[0930] 2. Backend program (Python, TensorFlow, OpenCV)

[0931] We perform video data analysis, natural language processing, and emotion recognition.

[0932] Data processing flow

[0933] The server receives user eye-tracking data and speaker video data, analyzes the speaker's mouth movements to estimate the content of the speech. Next, the estimated content is analyzed by a natural language processing unit and refined into natural-sounding sentences. The emotion recognition unit analyzes the user's emotional state and adjusts the display format of the speech content based on that data.

[0934] The smart glasses receive text data sent from a server and display it on the screen in a speech bubble format. The display position and design are adjusted based on the user's gaze information and emotional state.

[0935] Specific example

[0936] As an example, consider a situation where a customer asks about the price of a product.

[0937] 1. When a customer shows a product to an employee wearing smart glasses, the smart glasses' eye-tracking device detects the employee's gaze and captures the customer's face with a camera.

[0938] 2. The server analyzes the video data, estimates the spoken word (e.g., "How much is this product?"), and reshapes it into a natural-sounding sentence.

[0939] 3. The smart glasses' emotion recognition unit identifies the customer's interest from their facial expressions and displays text in a positive format.

[0940] 4. Employees can see clearly and visually appropriate text and respond appropriately to customers.

[0941] Example of inputting prompt text into a generative AI model:

[0942] Please describe a system that supports communication between staff and customers in physical stores. This system involves staff using smart glasses to estimate and visually display what customers are saying, utilizing eye-tracking and emotion recognition capabilities. Please provide a detailed explanation, including specific examples of how it works.

[0943] Thus, this invention is expected to effectively support communication between employees and customers with hearing impairments in physical stores and deepen understanding between them.

[0944] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0945] Step 1:

[0946] The device (smart glasses) uses an eye-tracking device to acquire the user's (employee's) gaze information in real time.

[0947] Input: User eye-tracking data

[0948] Processing: Detection of gaze direction and point of gaze

[0949] Output: Eye-tracking information

[0950] Step 2:

[0951] The device captures the speaker's (customer's) face with a camera based on eye-tracking information and acquires the video data.

[0952] Input: Eye-tracking information

[0953] Processing: Identify the speaker's face and capture the video.

[0954] Output: Video data

[0955] Step 3:

[0956] The device sends the acquired video data to the server.

[0957] Input: Video data

[0958] Processing: Sending video data to the server

[0959] Output: Video data sent to the server

[0960] Step 4:

[0961] The server analyzes the transmitted video data and estimates the content of what the speaker is saying based on their mouth movements.

[0962] Input: Video data

[0963] Processing: Analysis of the speaker's mouth movements and estimation of the content of their speech.

[0964] Output: Estimated content of the statement

[0965] Step 5:

[0966] The server analyzes the estimated speech content using a natural language processing unit and formats it into natural-sounding sentences.

[0967] Input: Estimated content of the statement

[0968] Processing: Refine the text based on context using a natural language processing model.

[0969] Output: A well-formatted text

[0970] Step 6:

[0971] The terminal analyzes the user's (employee's) facial expressions and voice using an emotion recognition unit to identify the user's emotional state. This data is then sent to the server.

[0972] Input: Facial expression data, audio data

[0973] Processing: Identify emotional states using an emotion recognition algorithm.

[0974] Output: Sentiment data

[0975] Step 7:

[0976] The server uses sentiment data to adjust the display format and generate speech content with improved estimation accuracy.

[0977] Input: Sentimental data, formatted text

[0978] Processing: Adjusting display format, regenerating message content

[0979] Output: Adjusted display format and content of the message

[0980] Step 8:

[0981] The server sends the generated text data and display format information to the terminal.

[0982] Input: Adjusted display format and content of the statement

[0983] Processing: Sending text data and display format information to the terminal.

[0984] Output: Text data and display format information sent to the terminal

[0985] Step 9:

[0986] The device displays received text data in a speech bubble format on its screen. The display position and design are adjusted based on the user's eye gaze and emotional state.

[0987] Input: Received text data and display format information

[0988] Processing: Display in speech bubble format, adjust display position and design.

[0989] Output: The content of the statement displayed on the screen

[0990] The above outlines the specific processing steps of the system of the present invention. This enables smooth communication between hearing-impaired customers and employees in physical stores.

[0991] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0992] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0993] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0994] [Third Embodiment]

[0995] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0996] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0997] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0998] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0999] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1000] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1001] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1002] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1003] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[1004] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1005] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1006] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[1007] This invention relates to a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content. More specifically, it relates to a system that enables hearing-impaired individuals to communicate with others without using their hands and to easily understand conversations with multiple speakers.

[1008] System Configuration

[1009] This system consists of the following main components:

[1010] 1. Device (smart glasses)

[1011] 2. Server

[1012] 3. User

[1013] terminal

[1014] The device takes the form of smart glasses. The device is equipped with the following functions:

[1015] Eye-tracking function: Tracks the user's gaze and collects eye-tracking information.

[1016] Camera: Captures video of the speaker the user is focusing on.

[1017] Display: Visually displays the estimated content of the statement.

[1018] server

[1019] The server receives data sent from the terminal and performs the following processing:

[1020] Mouth movement analysis: Analyze the speaker's mouth movements to estimate the content of their speech.

[1021] Natural language processing: Refining estimated speech content into natural-sounding sentences based on context.

[1022] Data transmission: The estimated content of the statement is sent to the terminal.

[1023] Program processing

[1024] Eye tracking and speaker identification

[1025] 1. The device uses eye-tracking technology to track the user's gaze in real time. This allows it to identify the speaker the user is focusing on.

[1026] 2. The device captures video of the identified speaker using its camera and sends that data to the server.

[1027] Estimation of the content of the statement

[1028] 3. The server receives the transmitted speaker's video data and analyzes the speaker's mouth movements using Gemini-based lip-reading technology.

[1029] 4. The server accurately estimates the content of speech based on lip-sync data. The estimated text data is analyzed using a ChatGPT-based natural language processing model to refine it into natural and accurate sentences based on the context.

[1030] Display of the content of the statement

[1031] 5. The server sends the generated text data to the terminal.

[1032] 6. The device displays the received text data on the screen in a speech bubble format. The display position is adjusted based on the user's eye-tracking information.

[1033] Specific example

[1034] Example 1: One-on-one conversation

[1035] If user A is having a conversation with friend B

[1036] The device detects that user A's gaze is directed towards friend B.

[1037] The device captures friend B's face with its camera and sends the video to the server.

[1038] The server analyzes friend B's mouth movements and estimates that he said, "It's a nice day today."

[1039] The estimated text is processed using natural language processing on the server and then sent to user A's terminal.

[1040] The device displays the text "It's a nice day today" in a speech bubble.

[1041] Example 2: Conversation involving multiple people

[1042] If user A is having a conversation with friends B and C

[1043] The device detects that user A's gaze has shifted from B to C.

[1044] Based on each user's gaze data, the device sends the corresponding video to the server.

[1045] The server estimates the content of what B and C said and formats it as text data.

[1046] The device displays B's statement, "What are your plans for tomorrow?", and C's statement, "I want to go to the movies," in speech bubble format at their respective positions on the screen.

[1047] In this way, this system enables hearing-impaired individuals to communicate smoothly with others without using their hands. As a result, users will not miss important information and will be able to live a more natural life without feeling socially isolated.

[1048] The following describes the processing flow.

[1049] Step 1:

[1050] The device uses eye-tracking technology to track the user's gaze. It detects the position of the user's pupils and the direction of their gaze to determine which direction they are looking.

[1051] Step 2:

[1052] The device identifies the person (speaker) the user is focusing on based on eye-tracking information. If there are multiple speakers, the person being focused on the most is selected as the speaker.

[1053] Step 3:

[1054] The device captures the face of the identified speaker using its camera. The acquired video data is processed in real time, so it is captured continuously without interruption.

[1055] Step 4:

[1056] The terminal encodes the captured video data and sends it to the server over the network. A high-speed communication protocol is used to minimize latency during this process.

[1057] Step 5:

[1058] The server analyzes the received video data using Gemini-based lip-reading technology. It extracts the speaker's mouth movements and estimates the content of their speech with high accuracy based on those movements.

[1059] Step 6:

[1060] The server further analyzes the estimated speech content using a ChatGPT-based natural language processing model. This refines the estimated text into a natural and accurate sentence based on the context.

[1061] Step 7:

[1062] The server encodes the generated text data and sends it to the terminal over the network. It applies protocols to ensure data integrity during transmission.

[1063] Step 8:

[1064] The device decodes the received text data and displays it on the screen in a speech bubble format. The display position is adjusted based on the user's eye gaze and the speaker's position.

[1065] Step 9:

[1066] Users read speech bubbles displayed on the smart glasses' screen to visually understand what the speaker is saying. This allows them to grasp the conversation without using their hands.

[1067] (Example 1)

[1068] Next, we will describe Example 1. 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."

[1069] There is a need for a system that allows people with hearing impairments to communicate with others without using their hands. In particular, the challenge lies in smoothly understanding conversations and facilitating smooth communication, even when multiple speakers are involved. This requires advanced processing such as eye tracking, lip-syncing, speech content estimation, and natural language processing for text formatting, and there is a demand for a system that can achieve these capabilities.

[1070] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1071] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, means for displaying the estimated content of the speech, means for collecting video data from a camera, and means for processing the estimated content of the speech into natural-sounding text using natural language processing. This makes it possible for people with hearing impairments to obtain a lot of information without using their hands and to communicate smoothly with others.

[1072] A "user" is a person who uses a system to communicate.

[1073] A "means of tracking eye movements" refers to a device that has the function of detecting and tracking a user's gaze in real time.

[1074] The "speaker" is the person the user is focusing their gaze on and who is speaking.

[1075] A "means for analyzing mouth movements" is a device that analyzes the movements of a speaker's mouth and estimates the content of what they are saying from those movements.

[1076] A "means for estimating the content of speech" is a device that estimates a speaker's speech as text data based on analyzed lip-movement data.

[1077] "Means for displaying the content of a statement" refers to a device for visually conveying the estimated content of a statement to the user.

[1078] "Means for collecting video data using a camera" refers to a device for capturing video of a speaker and collecting that data.

[1079] "Natural language processing" is a technique for shaping estimated speech content into natural-sounding sentences based on context.

[1080] A "server" is a device that receives data transmitted from a terminal, performs analysis and estimation, and transmits data.

[1081] A "device" is a device worn by the user that tracks their gaze, captures video with a camera, and transmits data to a server.

[1082] This invention relates to a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the results. Specifically, it is a system that enables hearing-impaired individuals to communicate smoothly with others without using their hands.

[1083] System Configuration

[1084] This system is mainly composed of the following components:

[1085] 1. Device (smart glasses)

[1086] 2. Server

[1087] 3. User

[1088] terminal

[1089] The device takes the form of smart glasses and has the following functions:

[1090] Eye-tracking function: Tracks the user's gaze in real time and collects gaze information.

[1091] Camera: Captures video of the speaker the user is focusing on.

[1092] Display: Visually displays the estimated content of the statement.

[1093] server

[1094] The server receives data sent from the terminal and performs the following processing:

[1095] Mouth movement analysis: The speaker's mouth movements are analyzed using Gemini-based technology.

[1096] Natural Language Processing: The estimated utterance content is refined based on context using a ChatGPT-based natural language processing model.

[1097] Data transmission: The estimated content of the statement is sent to the terminal.

[1098] System operation

[1099] Eye tracking and speaker identification

[1100] The device uses eye-tracking technology to track the user's gaze. For example, when a user looks at someone they are talking to, that gaze information is instantly collected. This allows the device to identify the speaker the user is focusing on.

[1101] Data capture and transmission

[1102] The device captures video of the identified speaker using its built-in camera. The captured video data is sent to a server. This transmission uses high-speed communication technology (e.g., a 5G network) to process the data in real time.

[1103] Processing of video data and analysis of mouth movements

[1104] The server receives video data transmitted from the terminal and analyzes it using Gemini-based lip-reading technology. The server analyzes the speaker's mouth movements in detail from the video data and extracts the content of the speech from those movements.

[1105] Estimation of spoken content and natural language processing

[1106] The server accurately estimates the content of the speech based on the analyzed mouth movement data. The estimated content is then formatted into natural-sounding sentences using a ChatGPT-based natural language processing model.

[1107] Sending and displaying estimated results

[1108] The server sends the formatted text to the terminal. The terminal displays the received text data in a speech bubble format on its screen. The display position is adjusted based on the user's eye gaze to ensure that the text is intuitively understandable.

[1109] Specific example

[1110] Example 1: One-on-one conversation

[1111] If user A is having a conversation with friend B:

[1112] The device detects that user A's gaze is directed towards friend B.

[1113] The device captures friend B's face with its camera and sends the video to the server.

[1114] The server analyzes friend B's mouth movements and estimates that he said, "It's a nice day today."

[1115] The estimated text is processed using natural language processing on the server and then sent to user A's terminal.

[1116] The device displays the text "It's a nice day today" in a speech bubble.

[1117] Example 2: Conversation involving multiple people

[1118] If user A is having a conversation with friends B and C:

[1119] The device detects that user A's gaze has shifted from B to C.

[1120] Based on each user's gaze data, the device sends the corresponding video to the server.

[1121] The server estimates the content of what B and C said and formats it as text data.

[1122] The device displays B's statement, "What are your plans for tomorrow?", and C's statement, "I want to go to the movies," in speech bubble format at their respective positions on the screen.

[1123] Example of a prompt

[1124] Please describe a system that tracks the user's gaze and analyzes the speaker's mouth movements to estimate what they are saying. Explain this using specific examples of one-on-one and group conversations.

[1125] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1126] Step 1:

[1127] The device uses eye-tracking technology to track the user's gaze in real time. The input is the user's gaze information, and the main output is coordinate data of the direction the gaze is directed. Based on this coordinate data, the device identifies the speaker the user is focusing on. A high-precision eye-tracking sensor built into the device captures even the slightest movement of the gaze and collects it as gaze information.

[1128] Step 2:

[1129] The device captures video of the speaker, identified based on eye-tracking information, using its built-in camera. The input is the speaker's location information identified by eye tracking, and the output is the captured video data. The built-in camera acquires high-resolution video and collects video data by focusing on the speaker's mouth. The collected video data is processed in real time and transmitted to the server.

[1130] Step 3:

[1131] The server receives video data transmitted from the terminal. The input is video data transmitted from the terminal, and the output is data in which the speaker's mouth movements have been analyzed. The server analyzes the video data using Gemini-based lip-reading technology and extracts specific movements that form words from the speaker's mouth movements. It performs noise reduction and correction on the data to achieve highly accurate mouth movement analysis.

[1132] Step 4:

[1133] The server estimates the speaker's speech based on analyzed lip-sync data. The input is lip-sync data, and the output is estimated text data. The server uses machine learning algorithms to estimate the actual spoken words with high accuracy. The estimated speech content is then formatted using contextual information and known language models.

[1134] Step 5:

[1135] The server uses a ChatGPT-based natural language processing model to format the estimated utterance into natural-sounding text. The input is estimated text data, and the output is context-formatted natural-sounding text. The server performs contextual analysis and formats the text as natural dialogue.

[1136] Step 6:

[1137] The server sends the formatted speech to the terminal. The input is formatted, natural-sounding text, and the output is text data sent to the terminal. The data is compressed and transmitted efficiently using high-speed communication technology (e.g., 5G network).

[1138] Step 7:

[1139] The terminal displays received text data in a speech bubble format on its screen. The input is text data sent from the server, and the output is text information visually displayed to the user. The display adjusts its position based on the user's eye gaze, positioning the text in an intuitively understandable location.

[1140] (Application Example 1)

[1141] Next, we will explain Application Example 1. In the following explanation, 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."

[1142] When hearing-impaired individuals or workers have difficulty communicating effectively, it can be particularly challenging in work environments such as factories to accurately understand work instructions, potentially compromising safety and efficiency. This invention aims to solve these problems and provide a system that supports users in easily communicating with others.

[1143] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1144] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, means for displaying the estimated content of the speech, means for visually displaying work instructions, and means for facilitating smooth communication with robots and other workers. As a result, the user can visually confirm the content of the speech of the speaker they are looking at, and can also clearly and quickly grasp work instructions within the factory.

[1145] A "user" is an individual who uses the system to utilize eye-tracking and speech content estimation features.

[1146] "Methods for tracking eye movements" refer to devices and technologies that detect the direction of a user's gaze in real time and acquire that data.

[1147] The "speaker" is the person the user is looking at, and whose mouth movements are being analyzed.

[1148] "Methods for analyzing mouth movements" refer to signal processing and algorithms that detect mouth movements based on video data captured by a camera and estimate the content of speech based on those movements.

[1149] "Means for estimating the content of speech" refers to methods and techniques that analyze the speaker's lip-sync data to infer what the speaker is saying.

[1150] "Means for displaying estimated speech content" refers to displays or devices that visually present estimated text to the user.

[1151] "Means of visually displaying work instructions" refers to a system that displays and presents work instructions and announcements within a factory within the user's field of vision.

[1152] A "robot" is a mechanical device that operates autonomously or remotely and performs tasks within a factory.

[1153] "Means of facilitating smooth communication" refers to methods and technologies for seamlessly transferring information between users, robots, and other workers.

[1154] This invention is a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content. This system is particularly intended to support factory workers in facilitating smooth communication with robots and other workers.

[1155] System Configuration

[1156] This system consists of the following main components:

[1157] 1. Device (smart glasses)

[1158] 2. Server

[1159] 3. User

[1160] terminal

[1161] The device takes the form of smart glasses. The device includes the following functions:

[1162] Eye-tracking function: Tracks the user's gaze and collects eye-tracking information.

[1163] Camera: Captures video of the speaker the user is focusing on.

[1164] Display: Visually displays estimated spoken content and work instructions.

[1165] server

[1166] The server receives data sent from the terminal and performs the following processing:

[1167] Mouth movement analysis: Analyze the speaker's mouth movements to estimate the content of their speech.

[1168] Natural language processing: Refining estimated speech content into natural-sounding sentences based on context.

[1169] Data transmission: Estimated spoken content and work instructions are sent to the terminal.

[1170] Program Description

[1171] The server processes data using Python, OpenCV, and dlib. Eye-tracking and camera data are acquired from sensors in smart glasses. The speaker's mouth movements are detected and analyzed using the dlib library. The estimated speech is formatted using natural language processing techniques such as Google Text-to-Speech (gTTS) and displayed on the screen.

[1172] Specific example

[1173] Example 1: Work instructions given in a one-on-one conversation

[1174] A scene where worker A receives instructions from robot operator B in a factory:

[1175] The device detects that A's gaze is directed towards B.

[1176] The device captures B's face with its camera and sends the video to the server.

[1177] The server analyzes B's mouth movements and infers that he said, "Please attach the parts."

[1178] The estimated text is processed using natural language processing and then sent to terminal A.

[1179] The device displays the text "Please install the parts" on its screen.

[1180] Example 2: Communication with multiple speakers

[1181] A scene in a factory where worker A interacts with multiple robot operators C and D:

[1182] The device detects that A's gaze has shifted from C to D.

[1183] The corresponding video is sent to the server, and the mouth movements of each person are analyzed.

[1184] The server infers the meaning of C's statement, "Proceed to the next step," and D's statement, "Check the parts."

[1185] The device displays each statement on its screen.

[1186] Examples of prompts to input into a generative AI model

[1187] TXT

[1188] Person A, working in a factory, is wearing smart glasses. Operator B says, "Next, please attach this part." The smart glasses track Person A's gaze and analyze B's mouth movements, displaying the spoken words as text. The smart glasses' display shows "Next, please attach this part," and Person A understands the instructions accurately.

[1189] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1190] Step 1:

[1191] The device tracks the user's gaze in real time using eye-tracking technology. The input is the user's gaze information, and the output is data on the direction the gaze is directed. This gaze information is collected using built-in sensors and a camera.

[1192] Step 2:

[1193] The device captures the face of the person it is looking at using its camera. The input is gaze information and camera footage, and the output is video data of the person the device is looking at.

[1194] Step 3:

[1195] The terminal sends the captured video data to the server. The input is video data, and the output is video data transmitted over the network.

[1196] Step 4:

[1197] The server analyzes the received video data and detects the speaker's mouth movements. The input is the transmitted video data, and the output is the analysis data regarding the mouth movements. The dlib library is used to detect the mouth movements.

[1198] Step 5:

[1199] The server estimates the content of a speaker's speech based on analysis data of their mouth movements. The input is mouth movement data, and the output is estimated text data. Natural language processing technology (generative AI model) is used for the estimation.

[1200] Step 6:

[1201] The server reshapes the estimated text data into natural-sounding sentences based on context. The input is estimated text data, and the output is contextually reshaped sentences.

[1202] Step 7:

[1203] The server sends formatted text data to the terminal. The input is formatted text data, and the output is text data transmitted over the network.

[1204] Step 8:

[1205] The device displays received text data on its screen. The input is text data sent from the server, and the output is text displayed on the smart glasses' screen. The display format is adjusted based on the user's eye-tracking information.

[1206] Step 9:

[1207] The user visually confirms the spoken content displayed on the smart glasses' screen. This action allows the user to understand the robot operator's instructions and work commands.

[1208] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1209] This invention is a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content, further combining it with an emotion engine that recognizes the user's emotions. This system allows hearing-impaired individuals to communicate with others without using their hands, and can improve the quality of communication by presenting the content of what is being said in a display format that corresponds to the user's emotions. Specifically, it has the following configuration and processing flow.

[1210] System Configuration

[1211] This system consists of the following main components:

[1212] 1. Device (smart glasses)

[1213] 2. Server

[1214] 3. Emotional Engine

[1215] terminal

[1216] The device takes the form of smart glasses and is equipped with the following functions:

[1217] Eye-tracking function: Tracks the user's gaze and collects eye-tracking information.

[1218] Camera: Captures video of the speaker the user is focusing on.

[1219] Emotion recognition function: Analyzes the user's emotions from their facial expressions, voice, etc.

[1220] Display: Visually displays the estimated content of the statement.

[1221] server

[1222] The server receives data sent from the terminal and performs the following processing:

[1223] Mouth movement analysis: Analyze the speaker's mouth movements to estimate the content of their speech.

[1224] Natural language processing: Refining estimated speech content into natural-sounding sentences based on context.

[1225] Integration with the emotion engine: Use data from the emotion engine to improve the accuracy of estimating the content of speech.

[1226] Emotional Engine

[1227] The emotion engine recognizes the user's emotions and has the following functions:

[1228] Sentiment analysis: Analyzes the user's facial expressions and voice to identify their emotional state.

[1229] Display format adjustment: The display format of the user's comments will change according to their emotions.

[1230] Program processing

[1231] Eye tracking and speaker identification

[1232] 1. The device uses eye-tracking technology to track the user's gaze in real time. This allows it to identify the speaker the user is focusing on.

[1233] 2. The device captures the face of the identified speaker using its camera and sends the video data to the server.

[1234] Estimation of the content of the statement and recognition of the sentiment

[1235] 3. The server analyzes the transmitted speaker's video data and estimates the content of the speech based on the speaker's mouth movements. The estimated text data is then analyzed using a natural language processing model and refined into natural and accurate sentences based on the context.

[1236] 4. The device analyzes the user's facial expressions and voice using emotion recognition technology to identify the user's emotional state. This data is then sent to the server.

[1237] Display of content and emotion

[1238] 5. The server uses the user's sentiment data received from the sentiment engine to adjust the display format of the utterances. It also improves the accuracy of estimating the content of utterances based on the sentiment data.

[1239] 6. The server sends the generated text data and display format information to the terminal.

[1240] 7. The device displays received text data in a speech bubble format on the screen. The display position and design are adjusted based on the user's eye gaze and emotional state.

[1241] Specific example

[1242] Example 1: One-on-one conversation and emotion recognition

[1243] If user A is having a conversation with friend B

[1244] The device detects that user A's gaze is directed towards friend B, captures B's face with its camera, and sends the video to the server.

[1245] The server analyzes B's mouth movements, estimates that B is saying "It's a nice day today," and then refines it using natural language processing.

[1246] The device recognizes A's smile as an emotion and sends the emotion data to the server.

[1247] The server confirms A's happiness status, formats the text in a positive format, and sends it to the terminal.

[1248] The device displays the text "It's a nice day today" in a brightly colored speech bubble.

[1249] Example 2: Conversation involving multiple people and emotion recognition

[1250] If user A is having a conversation with friends B and C

[1251] The device detects when user A's gaze moves from B to C and sends the respective video feeds to the server.

[1252] The server analyzes the mouth movements of B and C and estimates what each of them is saying.

[1253] The device recognizes A's emotional state and sends the emotional data to the server.

[1254] The server adjusts the display format based on A's emotions and sends it to the terminal.

[1255] The device displays the text "What are your plans for tomorrow?" and "I want to go to the movies" in speech bubble format, with designs appropriate to their respective positions.

[1256] In this way, this system allows hearing-impaired individuals to communicate without using their hands, while also displaying information in a way that takes the user's emotions into consideration. This system is expected to reduce information loss and alleviate users' feelings of social isolation.

[1257] The following describes the processing flow.

[1258] Step 1:

[1259] The device uses eye-tracking technology to track the user's gaze in real time. It detects the position of the user's pupils and the direction of their gaze to determine which direction they are looking. This data is continuously collected and processed.

[1260] Step 2:

[1261] The device identifies the subject (speaker) the user is focusing on based on the collected eye-tracking data. If there are multiple speakers, it selects the person being focused on most as the speaker, thereby identifying the appropriate speaker.

[1262] Step 3:

[1263] The device captures the face of the identified speaker with its camera and encodes the video data in real time. This video data is converted to the appropriate format and immediately transferred to the server.

[1264] Step 4:

[1265] The server analyzes the received video data. Using Gemini-based lip-reading technology, it extracts the speaker's mouth movements and estimates the content of their speech from those movements. This analysis is highly accurate and provides results quickly.

[1266] Step 5:

[1267] The server further analyzes the estimated utterance using a ChatGPT-based natural language processing model. This refines the estimated text into a natural and accurate sentence based on the context. This eliminates ambiguity regarding the utterance and provides information in a way that is easy for the user to understand.

[1268] Step 6:

[1269] The device analyzes the user's facial expressions and voice data using emotion recognition technology. This analysis identifies the user's emotional state (e.g., joy, sadness, surprise, etc.) and sends that data to the server.

[1270] Step 7:

[1271] The server uses an emotion engine to analyze the user's emotional data. Based on the analysis results, it adjusts the display format of the user's statements. For example, if the user is happy, it will display the text using positive colors and fonts.

[1272] Step 8:

[1273] The server encodes the final text data and display format information and sends it to the terminal. It uses protocols to ensure data integrity and delivers the data correctly to the terminal.

[1274] Step 9:

[1275] The device decodes the received text data and displays it on the screen in a speech bubble format. The display position and design are appropriately adjusted based on the user's eye gaze and emotional state.

[1276] Step 10:

[1277] Users read speech bubbles displayed on the smart glasses' screen to visually understand what the speaker is saying. This allows them to continue conversations hands-free and communicate comfortably.

[1278] Specific example: One-on-one conversation and emotion recognition

[1279] Steps 1-3: While user A is talking to friend B, the device detects that A's gaze is directed towards B and captures B's face with its camera. The video data is sent to the server.

[1280] Steps 4-5: The server analyzes the transmitted video, estimates that B said "It's a nice day today," and then processes it using natural language processing.

[1281] Steps 6-7: The device analyzes A's facial expression, recognizes that A is smiling, and sends the emotion data to the server. The server confirms A's happiness level and formats the text in a positive format (e.g., bright colors).

[1282] Steps 8-10: The final text data is sent to the terminal and displayed on the screen in the form of a speech bubble saying "It's a nice day today." A reads this speech bubble and understands what B said.

[1283] Specific example: Conversation involving multiple people and emotion recognition

[1284] Steps 1-3: While user A is talking to friends B and C, the device detects when A's gaze shifts from B to C and captures video of each. The video data is sent to the server.

[1285] Steps 4-5: The server analyzes the mouth movements of B and C, estimates the content of their statements, "What are your plans for tomorrow?" and "I want to go to the movies," and processes them using natural language processing.

[1286] Steps 6-7: The device recognizes A's emotional state and sends that data to the server. The server adjusts the display format based on the emotional data (e.g., color and position according to the emotion).

[1287] Steps 8-10: The final text data is sent to the device, and "What are your plans for tomorrow?" and "I want to go to the movies" appear in speech bubbles in the appropriate positions. A reads these speech bubbles and understands what B and C said.

[1288] In this way, this system enables hearing-impaired individuals to communicate without using their hands and presents their statements in an appropriate, emotion-based display format.

[1289] (Example 2)

[1290] Next, we will describe Example 2. 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."

[1291] Conventional communication systems for the visually impaired focus on eye tracking and inference of spoken content, but they often fail to adjust the display format to take user emotions into consideration, resulting in a decline in the quality of communication. Furthermore, spoken content may not be presented in a natural, contextual manner, making it difficult to understand.

[1292] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1293] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the statement from the analyzed mouth movements, means for displaying the estimated content of the statement, means for recognizing the user's emotions, and means for adjusting the display format of the content of the statement according to the user's emotions. This makes it possible to display the content of the statement according to the user's emotional state, thereby improving the quality of communication.

[1294] "Means of tracking a user's gaze" refers to all devices, including sensors and software, that detect which direction a user is looking.

[1295] "Means for analyzing a speaker's mouth movements" refers to all devices, including image processing technologies and algorithms, that analyze the movements and shape of a speaker's mouth to estimate what they are saying.

[1296] "Means for estimating the content of speech from analyzed mouth movements" refers to all devices, including speech recognition technology and machine learning models, that use data on mouth movements to estimate what a speaker is saying.

[1297] "Means for displaying estimated speech content" refers to all devices, including displays and projectors, that visually provide users with text and information that has been analyzed and estimated.

[1298] "Means of recognizing user emotions" refers to all devices, including sensors and software, that analyze data such as the user's facial expressions and voice to identify the user's emotional state.

[1299] "Means of adjusting the display format of spoken content according to the user's emotions" refers to all devices, including software and algorithms, that appropriately adjust the visual attributes of the displayed text and information, such as color and font, according to the recognized emotional state.

[1300] "Means of selecting a single speaker from among multiple speakers based on the user's gaze" refers to any device that includes technologies and algorithms for automatically selecting a specific speaker that the user is focusing on, using the user's gaze information, when there are multiple speakers.

[1301] "Means of refining spoken content into natural-sounding sentences based on context" refers to all devices, including natural language processing technologies and generative models, that convert estimated spoken content into more natural and understandable sentences.

[1302] This invention combines a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content with an emotion engine that recognizes the user's emotions. This system allows hearing-impaired individuals to communicate with others without using their hands, and improves the quality of communication by presenting the content of what is being said in a display format that corresponds to the user's emotions.

[1303] System Configuration

[1304] This system consists of three main components: a terminal (e.g., smart glasses), a server, and an emotion engine.

[1305] Device (smart glasses)

[1306] The device has the following features:

[1307] Eye-tracking function: Tracks the user's gaze and collects gaze information. For example, using an "eye-tracking sensor".

[1308] Camera: Captures video of the speaker the user is focusing on. For example, use a "camera device".

[1309] Emotion recognition function: Analyzes the user's facial expressions to identify their emotional state. For example, using "emotion recognition software".

[1310] Display: A method of visually displaying the estimated content of the speech, which is built into the smart glasses.

[1311] server

[1312] The server receives data sent from the terminal and performs the following processing:

[1313] Mouth movement analysis: Analyzes the speaker's mouth movements to estimate the content of their speech. For example, using "speech recognition technology."

[1314] Natural Language Processing: This process reshapes estimated speech content into natural-sounding sentences based on context. For example, it uses "natural language processing techniques."

[1315] Integration with the emotion engine: Use data from the emotion engine to improve the accuracy of estimating the content of speech.

[1316] Emotional Engine

[1317] The emotion engine recognizes the user's emotions and has the following functions:

[1318] Sentiment analysis: Analyzes the user's facial expressions and voice to identify their emotional state.

[1319] Display format adjustment: The display format of the user's comments will change according to their emotions.

[1320] Program processing

[1321] The program for this system processes the data in the following steps: First, the smart glasses worn by the user track the user's gaze using eye-tracking sensors to identify what they are looking at. Next, a camera device captures the face of the identified speaker and sends the video to the server.

[1322] The server uses speech recognition technology to analyze the speaker's mouth movements based on the received video data and estimates the content of their speech. Since the estimated content may sound unnatural in context, natural language processing technology is used to reshape it into more natural-sounding sentences.

[1323] Simultaneously, the smart glasses use emotion recognition software to analyze the user's emotions and send the results to a server. The server uses this emotion data to adjust the display format according to the user's emotions and then sends the final text data to the smart glasses.

[1324] Smart glasses display received text data on their screen. In this way, users can obtain information based on their gaze and emotions without using their hands or interacting with others.

[1325] Specific example

[1326] Example 1: One-on-one conversation and emotion recognition

[1327] If user A is having a conversation with friend B:

[1328] The device detects that user A's gaze is directed towards friend B and captures B's face. It then sends the video to the server.

[1329] The server analyzes B's mouth movements and estimates that B said, "It's a nice day today." It then refines this into a more natural-sounding sentence.

[1330] The device recognizes A's smile as an emotion and sends the emotion data to the server.

[1331] The server confirms A's happiness status, formats the text in a positive format, and sends it to the terminal.

[1332] The device displays the text "It's a nice day today" in a brightly colored speech bubble.

[1333] Example 2: Conversation involving multiple people and emotion recognition

[1334] If user A is having a conversation with friends B and C:

[1335] The device detects when user A's gaze moves from B to C, captures the video footage from each location, and sends it to the server.

[1336] The server analyzes the mouth movements of B and C and estimates what each of them is saying.

[1337] The device recognizes A's emotional state and sends the emotional data to the server.

[1338] The server adjusts the display format based on A's emotions and sends it to the terminal.

[1339] The device displays the text "What are your plans for tomorrow?" and "I want to go to the movies" in speech bubble format, with designs appropriate to their respective positions.

[1340] Example of a prompt

[1341] Example 1: Prompts when handling one-on-one conversations

[1342] User A is having a conversation with Friend B. Taking into account User A's gaze and emotional state, transcribe Friend B's statement, "It's a nice day today," and generate the text to display on User A's smart glasses.

[1343] Example 2: Prompts when handling conversations with multiple people

[1344] User A is in a conversation with friend B and friend C. User A's gaze shifts from B to C. Transcribe friend B's statement, "What are your plans for tomorrow?" and friend C's statement, "I want to go to the movies," and generate text to display on the user's smart glasses. Display the text in an appropriate format based on User A's gaze and emotional state.

[1345] In this way, the system integrates user eye tracking, speaker lip-syncing, emotion recognition, and natural language processing and display adjustment of spoken content to provide a high-quality communication experience.

[1346] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1347] Step 1:

[1348] Eye tracking and speaker identification

[1349] Input: User eye-tracking data (real-time location information)

[1350] Processing: The device uses eye-tracking technology to track the user's gaze in real time. It identifies the area the user is fixated on based on their gaze.

[1351] Output: Information about the direction the user is looking.

[1352] Specific operation: In a scenario where the user is talking to friend B, the device uses an eye-tracking sensor to detect the user's gaze direction and identify the face in that direction.

[1353] Step 2:

[1354] Speaker video capture

[1355] Input: User's gaze information, location information of the speaker they are looking at.

[1356] Processing: The device captures video of the identified speaker using its camera and obtains the video data.

[1357] Output: Speaker's video data

[1358] Specific operation: The device uses its camera to capture video of friend B's face, which the user is looking at, and sends that data to the server.

[1359] Step 3:

[1360] Analysis of mouth movements and estimation of spoken content

[1361] Input: Speaker's video data

[1362] Processing: The server uses the received video data and applies speech recognition technology to analyze the speaker's mouth movements. Based on the analysis results, it estimates what the speaker is saying.

[1363] Output: Estimated content of the statement (text data)

[1364] Specific operation: The server analyzes the speaker's mouth movements and estimates the content of the statement, for example, "It's a nice day today." The estimated text data is then retrieved by a generating AI model.

[1365] Step 4:

[1366] Reshaping the content of the statement

[1367] Input: Estimated content of the statement (text data)

[1368] Processing: The server uses a generative AI model to format the spoken content into contextually natural-sounding sentences.

[1369] Output: Formatted speech (natural-sounding text)

[1370] Specific operation: The server uses natural language processing techniques to format text data that it has estimated to be "It's a nice day today" into a natural-sounding sentence. At this stage, it performs processing to adjust sentence endings and context.

[1371] Step 5:

[1372] Recognition of user emotions

[1373] Input: User's facial expression data, voice data

[1374] Processing: The device uses emotion recognition software to analyze the user's facial expressions and voice to identify their current emotional state.

[1375] Output: User emotion data (happiness, sadness, surprise, etc.)

[1376] Specific operation: The device analyzes the user's smile, recognizes that the user is in a happy state, and sends that emotional data to the server.

[1377] Step 6:

[1378] Adjusting the display format

[1379] Input: Formatted speech (natural-sounding text), user sentiment data

[1380] Processing: The server adjusts the display format of the utterances based on the sentiment data obtained. Specifically, it changes visual attributes such as color, font, and background.

[1381] Output: Adjusted display format information

[1382] Specific operation: The server formats the text in a brightly colored speech bubble format based on the user's happy emotional state.

[1383] Step 7:

[1384] Display of the content of the statement

[1385] Input: Adjusted display format information, formatted speech content

[1386] Processing: The terminal displays the content of the message on the screen based on the received text data and display format information.

[1387] Output: The content of the statement that is visually displayed to the user.

[1388] Specific action: The device displays the text "It's a nice day today" to the user in a brightly colored speech bubble.

[1389] Through the processing steps described above, the present invention realizes communication support based on the user's gaze and emotions.

[1390] (Application Example 2)

[1391] Next, we will explain application example 2. In the following explanation, 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."

[1392] One of the communication challenges for people with hearing impairments is their limited means of quickly understanding what others are saying. Furthermore, to improve the quality of communication, not only visual information but also the recognition of emotions and appropriate feedback are crucial. Especially in physical stores, smooth communication between employees and customers is essential, making a system that allows employees and customers with hearing impairments to communicate smoothly necessary.

[1393] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1394] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, means for displaying the estimated content of the speech, means for recognizing the user's emotions, means for adjusting the display format based on the recognized user's emotions, and means for supporting communication between employees and customers in a physical store. This enables employees to communicate smoothly with customers who are hearing impaired.

[1395] "Means of tracking user gaze" refers to devices or technologies that detect the direction and focus of a user's gaze in real time and acquire data accordingly.

[1396] "Means for analyzing the speaker's mouth movements" refers to devices or technologies that analyze the movements and shape of a speaker's mouth and estimate the content of their speech from that analysis.

[1397] "Methods for estimating the content of a statement from analyzed mouth movements" refers to algorithms or devices that estimate the content of a statement based on data obtained from the speaker's mouth movements.

[1398] "Means for displaying estimated speech content" refers to technologies such as displays and projection devices for visually displaying estimated speech content.

[1399] "Means of recognizing user emotions" refer to sensors and algorithms that identify emotional states from the user's facial expressions, posture, voice, etc.

[1400] "Means for adjusting the display format based on the recognized user's emotions" refers to devices or technologies for dynamically changing the display format of a user's statements according to their emotional state.

[1401] "Means of supporting communication between employees and customers in physical stores" refers to systems and applications that support smooth communication between employees and customers in physical stores.

[1402] This invention is a system that supports smooth communication between hearing-impaired customers and employees in physical stores. This system consists of the following main components:

[1403] hardware

[1404] 1. Smart Glasses

[1405] Eye-tracking device: Detects the user's (employee's) gaze direction and point of focus in real time.

[1406] Camera device: Captures video of the speaker's (customer's) face and sends the data to the server.

[1407] Display: Visually displays the estimated content of the statement.

[1408] 2. Server

[1409] Data Analysis Unit: Analyzes the speaker's mouth movements to estimate the content of their speech.

[1410] Natural Language Processing Unit: Reshapes estimated speech content into natural-sounding sentences based on context.

[1411] Emotion Recognition Unit: Analyzes the user's emotional state and adjusts the display format accordingly.

[1412] software

[1413] 1. Frontend Program (JavaScript)

[1414] The smart glasses collect data from various sensors and transmit it to the server in real time.

[1415] 2. Backend program (Python, TensorFlow, OpenCV)

[1416] We perform video data analysis, natural language processing, and emotion recognition.

[1417] Data processing flow

[1418] The server receives user eye-tracking data and speaker video data, analyzes the speaker's mouth movements to estimate the content of the speech. Next, the estimated content is analyzed by a natural language processing unit and refined into natural-sounding sentences. The emotion recognition unit analyzes the user's emotional state and adjusts the display format of the speech content based on that data.

[1419] The smart glasses receive text data sent from a server and display it on the screen in a speech bubble format. The display position and design are adjusted based on the user's gaze information and emotional state.

[1420] Specific example

[1421] As an example, consider a situation where a customer asks about the price of a product.

[1422] 1. When a customer shows a product to an employee wearing smart glasses, the smart glasses' eye-tracking device detects the employee's gaze and captures the customer's face with a camera.

[1423] 2. The server analyzes the video data, estimates the spoken word (e.g., "How much is this product?"), and reshapes it into a natural-sounding sentence.

[1424] 3. The smart glasses' emotion recognition unit identifies the customer's interest from their facial expressions and displays text in a positive format.

[1425] 4. Employees can see clearly and visually appropriate text and respond appropriately to customers.

[1426] Example of inputting prompt text into a generative AI model:

[1427] Please describe a system that supports communication between staff and customers in physical stores. This system involves staff using smart glasses to estimate and visually display what customers are saying, utilizing eye-tracking and emotion recognition capabilities. Please provide a detailed explanation, including specific examples of how it works.

[1428] Thus, this invention is expected to effectively support communication between employees and customers with hearing impairments in physical stores and deepen understanding between them.

[1429] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1430] Step 1:

[1431] The device (smart glasses) uses an eye-tracking device to acquire the user's (employee's) gaze information in real time.

[1432] Input: User eye-tracking data

[1433] Processing: Detection of gaze direction and point of gaze

[1434] Output: Eye-tracking information

[1435] Step 2:

[1436] The device captures the speaker's (customer's) face with a camera based on eye-tracking information and acquires the video data.

[1437] Input: Eye-tracking information

[1438] Processing: Identify the speaker's face and capture the video.

[1439] Output: Video data

[1440] Step 3:

[1441] The device sends the acquired video data to the server.

[1442] Input: Video data

[1443] Processing: Sending video data to the server

[1444] Output: Video data sent to the server

[1445] Step 4:

[1446] The server analyzes the transmitted video data and estimates the content of what the speaker is saying based on their mouth movements.

[1447] Input: Video data

[1448] Processing: Analysis of the speaker's mouth movements and estimation of the content of their speech.

[1449] Output: Estimated content of the statement

[1450] Step 5:

[1451] The server analyzes the estimated speech content using a natural language processing unit and formats it into natural-sounding sentences.

[1452] Input: Estimated content of the statement

[1453] Processing: Refine the text based on context using a natural language processing model.

[1454] Output: A well-formatted text

[1455] Step 6:

[1456] The terminal analyzes the user's (employee's) facial expressions and voice using an emotion recognition unit to identify the user's emotional state. This data is then sent to the server.

[1457] Input: Facial expression data, audio data

[1458] Processing: Identify emotional states using an emotion recognition algorithm.

[1459] Output: Sentiment data

[1460] Step 7:

[1461] The server uses sentiment data to adjust the display format and generate speech content with improved estimation accuracy.

[1462] Input: Sentimental data, formatted text

[1463] Processing: Adjusting display format, regenerating message content

[1464] Output: Adjusted display format and content of the message

[1465] Step 8:

[1466] The server sends the generated text data and display format information to the terminal.

[1467] Input: Adjusted display format and content of the statement

[1468] Processing: Sending text data and display format information to the terminal.

[1469] Output: Text data and display format information sent to the terminal

[1470] Step 9:

[1471] The device displays received text data in a speech bubble format on its screen. The display position and design are adjusted based on the user's eye gaze and emotional state.

[1472] Input: Received text data and display format information

[1473] Processing: Display in speech bubble format, adjust display position and design.

[1474] Output: The content of the statement displayed on the screen

[1475] The above outlines the specific processing steps of the system of the present invention. This enables smooth communication between hearing-impaired customers and employees in physical stores.

[1476] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1477] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1478] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[1479] [Fourth Embodiment]

[1480] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[1481] As shown in Figure 7, the 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.

[1482] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1483] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[1484] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1485] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1486] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1487] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive 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 robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[1488] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1489] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[1490] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1491] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1492] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1493] This invention relates to a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content. More specifically, it relates to a system that enables hearing-impaired individuals to communicate with others without using their hands and to easily understand conversations with multiple speakers.

[1494] System Configuration

[1495] This system consists of the following main components:

[1496] 1. Device (smart glasses)

[1497] 2. Server

[1498] 3. User

[1499] terminal

[1500] The device takes the form of smart glasses. The device is equipped with the following functions:

[1501] Eye-tracking function: Tracks the user's gaze and collects eye-tracking information.

[1502] Camera: Captures video of the speaker the user is focusing on.

[1503] Display: Visually displays the estimated content of the statement.

[1504] server

[1505] The server receives data sent from the terminal and performs the following processing:

[1506] Mouth movement analysis: Analyze the speaker's mouth movements to estimate the content of their speech.

[1507] Natural language processing: Refining estimated speech content into natural-sounding sentences based on context.

[1508] Data transmission: The estimated content of the statement is sent to the terminal.

[1509] Program processing

[1510] Eye tracking and speaker identification

[1511] 1. The device uses eye-tracking technology to track the user's gaze in real time. This allows it to identify the speaker the user is focusing on.

[1512] 2. The device captures video of the identified speaker using its camera and sends that data to the server.

[1513] Estimation of the content of the statement

[1514] 3. The server receives the transmitted speaker's video data and analyzes the speaker's mouth movements using Gemini-based lip-reading technology.

[1515] 4. The server accurately estimates the content of speech based on lip-sync data. The estimated text data is analyzed using a ChatGPT-based natural language processing model to refine it into natural and accurate sentences based on the context.

[1516] Display of the content of the statement

[1517] 5. The server sends the generated text data to the terminal.

[1518] 6. The device displays the received text data on the screen in a speech bubble format. The display position is adjusted based on the user's eye-tracking information.

[1519] Specific example

[1520] Example 1: One-on-one conversation

[1521] If user A is having a conversation with friend B

[1522] The device detects that user A's gaze is directed towards friend B.

[1523] The device captures friend B's face with its camera and sends the video to the server.

[1524] The server analyzes friend B's mouth movements and estimates that he said, "It's a nice day today."

[1525] The estimated text is processed using natural language processing on the server and then sent to user A's terminal.

[1526] The device displays the text "It's a nice day today" in a speech bubble.

[1527] Example 2: Conversation involving multiple people

[1528] If user A is having a conversation with friends B and C

[1529] The device detects that user A's gaze has shifted from B to C.

[1530] Based on each user's gaze data, the device sends the corresponding video to the server.

[1531] The server estimates the content of what B and C said and formats it as text data.

[1532] The device displays B's statement, "What are your plans for tomorrow?", and C's statement, "I want to go to the movies," in speech bubble format at their respective positions on the screen.

[1533] In this way, this system enables hearing-impaired individuals to communicate smoothly with others without using their hands. As a result, users will not miss important information and will be able to live a more natural life without feeling socially isolated.

[1534] The following describes the processing flow.

[1535] Step 1:

[1536] The device uses eye-tracking technology to track the user's gaze. It detects the position of the user's pupils and the direction of their gaze to determine which direction they are looking.

[1537] Step 2:

[1538] The device identifies the person (speaker) the user is focusing on based on eye-tracking information. If there are multiple speakers, the person being focused on the most is selected as the speaker.

[1539] Step 3:

[1540] The device captures the face of the identified speaker using its camera. The acquired video data is processed in real time, so it is captured continuously without interruption.

[1541] Step 4:

[1542] The terminal encodes the captured video data and sends it to the server over the network. A high-speed communication protocol is used to minimize latency during this process.

[1543] Step 5:

[1544] The server analyzes the received video data using Gemini-based lip-reading technology. It extracts the speaker's mouth movements and estimates the content of their speech with high accuracy based on those movements.

[1545] Step 6:

[1546] The server further analyzes the estimated speech content using a ChatGPT-based natural language processing model. This refines the estimated text into a natural and accurate sentence based on the context.

[1547] Step 7:

[1548] The server encodes the generated text data and sends it to the terminal over the network. It applies protocols to ensure data integrity during transmission.

[1549] Step 8:

[1550] The device decodes the received text data and displays it on the screen in a speech bubble format. The display position is adjusted based on the user's eye gaze and the speaker's position.

[1551] Step 9:

[1552] Users read speech bubbles displayed on the smart glasses' screen to visually understand what the speaker is saying. This allows them to grasp the conversation without using their hands.

[1553] (Example 1)

[1554] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1555] There is a need for a system that allows people with hearing impairments to communicate with others without using their hands. In particular, the challenge lies in smoothly understanding conversations and facilitating smooth communication, even when multiple speakers are involved. This requires advanced processing such as eye tracking, lip-syncing, speech content estimation, and natural language processing for text formatting, and there is a demand for a system that can achieve these capabilities.

[1556] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1557] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, means for displaying the estimated content of the speech, means for collecting video data from a camera, and means for processing the estimated content of the speech into natural-sounding text using natural language processing. This makes it possible for people with hearing impairments to obtain a lot of information without using their hands and to communicate smoothly with others.

[1558] A "user" is a person who uses a system to communicate.

[1559] A "means of tracking eye movements" refers to a device that has the function of detecting and tracking a user's gaze in real time.

[1560] The "speaker" is the person the user is focusing their gaze on and who is speaking.

[1561] A "means for analyzing mouth movements" is a device that analyzes the movements of a speaker's mouth and estimates the content of what they are saying from those movements.

[1562] A "means for estimating the content of speech" is a device that estimates a speaker's speech as text data based on analyzed lip-movement data.

[1563] "Means for displaying the content of a statement" refers to a device for visually conveying the estimated content of a statement to the user.

[1564] "Means for collecting video data using a camera" refers to a device for capturing video of a speaker and collecting that data.

[1565] "Natural language processing" is a technique for shaping estimated speech content into natural-sounding sentences based on context.

[1566] A "server" is a device that receives data transmitted from a terminal, performs analysis and estimation, and transmits data.

[1567] A "device" is a device worn by the user that tracks their gaze, captures video with a camera, and transmits data to a server.

[1568] This invention relates to a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the results. Specifically, it is a system that enables hearing-impaired individuals to communicate smoothly with others without using their hands.

[1569] System Configuration

[1570] This system is mainly composed of the following components:

[1571] 1. Device (smart glasses)

[1572] 2. Server

[1573] 3. User

[1574] terminal

[1575] The device takes the form of smart glasses and has the following functions:

[1576] Eye-tracking function: Tracks the user's gaze in real time and collects gaze information.

[1577] Camera: Captures video of the speaker the user is focusing on.

[1578] Display: Visually displays the estimated content of the statement.

[1579] server

[1580] The server receives data sent from the terminal and performs the following processing:

[1581] Mouth movement analysis: The speaker's mouth movements are analyzed using Gemini-based technology.

[1582] Natural Language Processing: The estimated utterance content is refined based on context using a ChatGPT-based natural language processing model.

[1583] Data transmission: The estimated content of the statement is sent to the terminal.

[1584] System operation

[1585] Eye tracking and speaker identification

[1586] The device uses eye-tracking technology to track the user's gaze. For example, when a user looks at someone they are talking to, that gaze information is instantly collected. This allows the device to identify the speaker the user is focusing on.

[1587] Data capture and transmission

[1588] The device captures video of the identified speaker using its built-in camera. The captured video data is sent to a server. This transmission uses high-speed communication technology (e.g., a 5G network) to process the data in real time.

[1589] Processing of video data and analysis of mouth movements

[1590] The server receives video data transmitted from the terminal and analyzes it using Gemini-based lip-reading technology. The server analyzes the speaker's mouth movements in detail from the video data and extracts the content of the speech from those movements.

[1591] Estimation of spoken content and natural language processing

[1592] The server accurately estimates the content of the speech based on the analyzed mouth movement data. The estimated content is then formatted into natural-sounding sentences using a ChatGPT-based natural language processing model.

[1593] Sending and displaying estimated results

[1594] The server sends the formatted text to the terminal. The terminal displays the received text data in a speech bubble format on its screen. The display position is adjusted based on the user's eye gaze to ensure that the text is intuitively understandable.

[1595] Specific example

[1596] Example 1: One-on-one conversation

[1597] If user A is having a conversation with friend B:

[1598] The device detects that user A's gaze is directed towards friend B.

[1599] The device captures friend B's face with its camera and sends the video to the server.

[1600] The server analyzes friend B's mouth movements and estimates that he said, "It's a nice day today."

[1601] The estimated text is processed using natural language processing on the server and then sent to user A's terminal.

[1602] The device displays the text "It's a nice day today" in a speech bubble.

[1603] Example 2: Conversation involving multiple people

[1604] If user A is having a conversation with friends B and C:

[1605] The device detects that user A's gaze has shifted from B to C.

[1606] Based on each user's gaze data, the device sends the corresponding video to the server.

[1607] The server estimates the content of what B and C said and formats it as text data.

[1608] The device displays B's statement, "What are your plans for tomorrow?", and C's statement, "I want to go to the movies," in speech bubble format at their respective positions on the screen.

[1609] Example of a prompt

[1610] Please describe a system that tracks the user's gaze and analyzes the speaker's mouth movements to estimate what they are saying. Explain this using specific examples of one-on-one and group conversations.

[1611] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1612] Step 1:

[1613] The device uses eye-tracking technology to track the user's gaze in real time. The input is the user's gaze information, and the main output is coordinate data of the direction the gaze is directed. Based on this coordinate data, the device identifies the speaker the user is focusing on. A high-precision eye-tracking sensor built into the device captures even the slightest movement of the gaze and collects it as gaze information.

[1614] Step 2:

[1615] The device captures video of the speaker, identified based on eye-tracking information, using its built-in camera. The input is the speaker's location information identified by eye tracking, and the output is the captured video data. The built-in camera acquires high-resolution video and collects video data by focusing on the speaker's mouth. The collected video data is processed in real time and transmitted to the server.

[1616] Step 3:

[1617] The server receives video data transmitted from the terminal. The input is video data transmitted from the terminal, and the output is data in which the speaker's mouth movements have been analyzed. The server analyzes the video data using Gemini-based lip-reading technology and extracts specific movements that form words from the speaker's mouth movements. It performs noise reduction and correction on the data to achieve highly accurate mouth movement analysis.

[1618] Step 4:

[1619] The server estimates the speaker's speech based on analyzed lip-sync data. The input is lip-sync data, and the output is estimated text data. The server uses machine learning algorithms to estimate the actual spoken words with high accuracy. The estimated speech content is then formatted using contextual information and known language models.

[1620] Step 5:

[1621] The server uses a ChatGPT-based natural language processing model to format the estimated utterance into natural-sounding text. The input is estimated text data, and the output is context-formatted natural-sounding text. The server performs contextual analysis and formats the text as natural dialogue.

[1622] Step 6:

[1623] The server sends the formatted speech to the terminal. The input is formatted, natural-sounding text, and the output is text data sent to the terminal. The data is compressed and transmitted efficiently using high-speed communication technology (e.g., 5G network).

[1624] Step 7:

[1625] The terminal displays received text data in a speech bubble format on its screen. The input is text data sent from the server, and the output is text information visually displayed to the user. The display adjusts its position based on the user's eye gaze, positioning the text in an intuitively understandable location.

[1626] (Application Example 1)

[1627] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1628] When hearing-impaired individuals or workers have difficulty communicating effectively, it can be particularly challenging in work environments such as factories to accurately understand work instructions, potentially compromising safety and efficiency. This invention aims to solve these problems and provide a system that supports users in easily communicating with others.

[1629] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1630] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, means for displaying the estimated content of the speech, means for visually displaying work instructions, and means for facilitating smooth communication with robots and other workers. As a result, the user can visually confirm the content of the speech of the speaker they are looking at, and can also clearly and quickly grasp work instructions within the factory.

[1631] A "user" is an individual who uses the system to utilize eye-tracking and speech content estimation features.

[1632] "Methods for tracking eye movements" refer to devices and technologies that detect the direction of a user's gaze in real time and acquire that data.

[1633] The "speaker" is the person the user is looking at, and whose mouth movements are being analyzed.

[1634] "Methods for analyzing mouth movements" refer to signal processing and algorithms that detect mouth movements based on video data captured by a camera and estimate the content of speech based on those movements.

[1635] "Means for estimating the content of speech" refers to methods and techniques that analyze the speaker's lip-sync data to infer what the speaker is saying.

[1636] "Means for displaying estimated speech content" refers to displays or devices that visually present estimated text to the user.

[1637] "Means of visually displaying work instructions" refers to a system that displays and presents work instructions and announcements within a factory within the user's field of vision.

[1638] A "robot" is a mechanical device that operates autonomously or remotely and performs tasks within a factory.

[1639] "Means of facilitating smooth communication" refers to methods and technologies for seamlessly transferring information between users, robots, and other workers.

[1640] This invention is a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content. This system is particularly intended to support factory workers in facilitating smooth communication with robots and other workers.

[1641] System Configuration

[1642] This system consists of the following main components:

[1643] 1. Device (smart glasses)

[1644] 2. Server

[1645] 3. User

[1646] terminal

[1647] The device takes the form of smart glasses. The device includes the following functions:

[1648] Eye-tracking function: Tracks the user's gaze and collects eye-tracking information.

[1649] Camera: Captures video of the speaker the user is focusing on.

[1650] Display: Visually displays estimated spoken content and work instructions.

[1651] server

[1652] The server receives data sent from the terminal and performs the following processing:

[1653] Mouth movement analysis: Analyze the speaker's mouth movements to estimate the content of their speech.

[1654] Natural language processing: Refining estimated speech content into natural-sounding sentences based on context.

[1655] Data transmission: Estimated spoken content and work instructions are sent to the terminal.

[1656] Program Description

[1657] The server processes data using Python, OpenCV, and dlib. Eye-tracking and camera data are acquired from sensors in smart glasses. The speaker's mouth movements are detected and analyzed using the dlib library. The estimated speech is formatted using natural language processing techniques such as Google Text-to-Speech (gTTS) and displayed on the screen.

[1658] Specific example

[1659] Example 1: Work instructions given in a one-on-one conversation

[1660] A scene where worker A receives instructions from robot operator B in a factory:

[1661] The device detects that A's gaze is directed towards B.

[1662] The device captures B's face with its camera and sends the video to the server.

[1663] The server analyzes B's mouth movements and infers that he said, "Please attach the parts."

[1664] The estimated text is processed using natural language processing and then sent to terminal A.

[1665] The device displays the text "Please install the parts" on its screen.

[1666] Example 2: Communication with multiple speakers

[1667] A scene in a factory where worker A interacts with multiple robot operators C and D:

[1668] The device detects that A's gaze has shifted from C to D.

[1669] The corresponding video is sent to the server, and the mouth movements of each person are analyzed.

[1670] The server infers the meaning of C's statement, "Proceed to the next step," and D's statement, "Check the parts."

[1671] The device displays each statement on its screen.

[1672] Examples of prompts to input into a generative AI model

[1673] TXT

[1674] Person A, working in a factory, is wearing smart glasses. Operator B says, "Next, please attach this part." The smart glasses track Person A's gaze and analyze B's mouth movements, displaying the spoken words as text. The smart glasses' display shows "Next, please attach this part," and Person A understands the instructions accurately.

[1675] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1676] Step 1:

[1677] The device tracks the user's gaze in real time using eye-tracking technology. The input is the user's gaze information, and the output is data on the direction the gaze is directed. This gaze information is collected using built-in sensors and a camera.

[1678] Step 2:

[1679] The device captures the face of the person it is looking at using its camera. The input is gaze information and camera footage, and the output is video data of the person the device is looking at.

[1680] Step 3:

[1681] The terminal sends the captured video data to the server. The input is video data, and the output is video data transmitted over the network.

[1682] Step 4:

[1683] The server analyzes the received video data and detects the speaker's mouth movements. The input is the transmitted video data, and the output is the analysis data regarding the mouth movements. The dlib library is used to detect the mouth movements.

[1684] Step 5:

[1685] The server estimates the content of a speaker's speech based on analysis data of their mouth movements. The input is mouth movement data, and the output is estimated text data. Natural language processing technology (generative AI model) is used for the estimation.

[1686] Step 6:

[1687] The server reshapes the estimated text data into natural-sounding sentences based on context. The input is estimated text data, and the output is contextually reshaped sentences.

[1688] Step 7:

[1689] The server sends formatted text data to the terminal. The input is formatted text data, and the output is text data transmitted over the network.

[1690] Step 8:

[1691] The device displays received text data on its screen. The input is text data sent from the server, and the output is text displayed on the smart glasses' screen. The display format is adjusted based on the user's eye-tracking information.

[1692] Step 9:

[1693] The user visually confirms the spoken content displayed on the smart glasses' screen. This action allows the user to understand the robot operator's instructions and work commands.

[1694] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1695] This invention is a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content, further combining it with an emotion engine that recognizes the user's emotions. This system allows hearing-impaired individuals to communicate with others without using their hands, and can improve the quality of communication by presenting the content of what is being said in a display format that corresponds to the user's emotions. Specifically, it has the following configuration and processing flow.

[1696] System Configuration

[1697] This system consists of the following main components:

[1698] 1. Device (smart glasses)

[1699] 2. Server

[1700] 3. Emotional Engine

[1701] terminal

[1702] The device takes the form of smart glasses and is equipped with the following functions:

[1703] Eye-tracking function: Tracks the user's gaze and collects eye-tracking information.

[1704] Camera: Captures video of the speaker the user is focusing on.

[1705] Emotion recognition function: Analyzes the user's emotions from their facial expressions, voice, etc.

[1706] Display: Visually displays the estimated content of the statement.

[1707] server

[1708] The server receives data sent from the terminal and performs the following processing:

[1709] Mouth movement analysis: Analyze the speaker's mouth movements to estimate the content of their speech.

[1710] Natural language processing: Refining estimated speech content into natural-sounding sentences based on context.

[1711] Integration with the emotion engine: Use data from the emotion engine to improve the accuracy of estimating the content of speech.

[1712] Emotional Engine

[1713] The emotion engine recognizes the user's emotions and has the following functions:

[1714] Sentiment analysis: Analyzes the user's facial expressions and voice to identify their emotional state.

[1715] Display format adjustment: The display format of the user's comments will change according to their emotions.

[1716] Program processing

[1717] Eye tracking and speaker identification

[1718] 1. The device uses eye-tracking technology to track the user's gaze in real time. This allows it to identify the speaker the user is focusing on.

[1719] 2. The device captures the face of the identified speaker using its camera and sends the video data to the server.

[1720] Estimation of the content of the statement and recognition of the sentiment

[1721] 3. The server analyzes the transmitted speaker's video data and estimates the content of the speech based on the speaker's mouth movements. The estimated text data is then analyzed using a natural language processing model and refined into natural and accurate sentences based on the context.

[1722] 4. The device analyzes the user's facial expressions and voice using emotion recognition technology to identify the user's emotional state. This data is then sent to the server.

[1723] Display of content and emotion

[1724] 5. The server uses the user's sentiment data received from the sentiment engine to adjust the display format of the utterances. It also improves the accuracy of estimating the content of utterances based on the sentiment data.

[1725] 6. The server sends the generated text data and display format information to the terminal.

[1726] 7. The device displays received text data in a speech bubble format on the screen. The display position and design are adjusted based on the user's eye gaze and emotional state.

[1727] Specific example

[1728] Example 1: One-on-one conversation and emotion recognition

[1729] If user A is having a conversation with friend B

[1730] The device detects that user A's gaze is directed towards friend B, captures B's face with its camera, and sends the video to the server.

[1731] The server analyzes B's mouth movements, estimates that B is saying "It's a nice day today," and then refines it using natural language processing.

[1732] The device recognizes A's smile as an emotion and sends the emotion data to the server.

[1733] The server confirms A's happiness status, formats the text in a positive format, and sends it to the terminal.

[1734] The device displays the text "It's a nice day today" in a brightly colored speech bubble.

[1735] Example 2: Conversation involving multiple people and emotion recognition

[1736] If user A is having a conversation with friends B and C

[1737] The device detects when user A's gaze moves from B to C and sends the respective video feeds to the server.

[1738] The server analyzes the mouth movements of B and C and estimates what each of them is saying.

[1739] The device recognizes A's emotional state and sends the emotional data to the server.

[1740] The server adjusts the display format based on A's emotions and sends it to the terminal.

[1741] The device displays the text "What are your plans for tomorrow?" and "I want to go to the movies" in speech bubble format, with designs appropriate to their respective positions.

[1742] In this way, this system allows hearing-impaired individuals to communicate without using their hands, while also displaying information in a way that takes the user's emotions into consideration. This system is expected to reduce information loss and alleviate users' feelings of social isolation.

[1743] The following describes the processing flow.

[1744] Step 1:

[1745] The device uses eye-tracking technology to track the user's gaze in real time. It detects the position of the user's pupils and the direction of their gaze to determine which direction they are looking. This data is continuously collected and processed.

[1746] Step 2:

[1747] The device identifies the subject (speaker) the user is focusing on based on the collected eye-tracking data. If there are multiple speakers, it selects the person being focused on most as the speaker, thereby identifying the appropriate speaker.

[1748] Step 3:

[1749] The device captures the face of the identified speaker with its camera and encodes the video data in real time. This video data is converted to the appropriate format and immediately transferred to the server.

[1750] Step 4:

[1751] The server analyzes the received video data. Using Gemini-based lip-reading technology, it extracts the speaker's mouth movements and estimates the content of their speech from those movements. This analysis is highly accurate and provides results quickly.

[1752] Step 5:

[1753] The server further analyzes the estimated utterance using a ChatGPT-based natural language processing model. This refines the estimated text into a natural and accurate sentence based on the context. This eliminates ambiguity regarding the utterance and provides information in a way that is easy for the user to understand.

[1754] Step 6:

[1755] The device analyzes the user's facial expressions and voice data using emotion recognition technology. This analysis identifies the user's emotional state (e.g., joy, sadness, surprise, etc.) and sends that data to the server.

[1756] Step 7:

[1757] The server uses an emotion engine to analyze the user's emotional data. Based on the analysis results, it adjusts the display format of the user's statements. For example, if the user is happy, it will display the text using positive colors and fonts.

[1758] Step 8:

[1759] The server encodes the final text data and display format information and sends it to the terminal. It uses protocols to ensure data integrity and delivers the data correctly to the terminal.

[1760] Step 9:

[1761] The device decodes the received text data and displays it on the screen in a speech bubble format. The display position and design are appropriately adjusted based on the user's eye gaze and emotional state.

[1762] Step 10:

[1763] Users read speech bubbles displayed on the smart glasses' screen to visually understand what the speaker is saying. This allows them to continue conversations hands-free and communicate comfortably.

[1764] Specific example: One-on-one conversation and emotion recognition

[1765] Steps 1-3: While user A is talking to friend B, the device detects that A's gaze is directed towards B and captures B's face with its camera. The video data is sent to the server.

[1766] Steps 4-5: The server analyzes the transmitted video, estimates that B said "It's a nice day today," and then processes it using natural language processing.

[1767] Steps 6-7: The device analyzes A's facial expression, recognizes that A is smiling, and sends the emotion data to the server. The server confirms A's happiness level and formats the text in a positive format (e.g., bright colors).

[1768] Steps 8-10: The final text data is sent to the terminal and displayed on the screen in the form of a speech bubble saying "It's a nice day today." A reads this speech bubble and understands what B said.

[1769] Specific example: Conversation involving multiple people and emotion recognition

[1770] Steps 1-3: While user A is talking to friends B and C, the device detects when A's gaze shifts from B to C and captures video of each. The video data is sent to the server.

[1771] Steps 4-5: The server analyzes the mouth movements of B and C, estimates the content of their statements, "What are your plans for tomorrow?" and "I want to go to the movies," and processes them using natural language processing.

[1772] Steps 6-7: The device recognizes A's emotional state and sends that data to the server. The server adjusts the display format based on the emotional data (e.g., color and position according to the emotion).

[1773] Steps 8-10: The final text data is sent to the device, and "What are your plans for tomorrow?" and "I want to go to the movies" appear in speech bubbles in the appropriate positions. A reads these speech bubbles and understands what B and C said.

[1774] In this way, this system enables hearing-impaired individuals to communicate without using their hands and presents their statements in an appropriate, emotion-based display format.

[1775] (Example 2)

[1776] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1777] Conventional communication systems for the visually impaired focus on eye tracking and inference of spoken content, but they often fail to adjust the display format to take user emotions into consideration, resulting in a decline in the quality of communication. Furthermore, spoken content may not be presented in a natural, contextual manner, making it difficult to understand.

[1778] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1779] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the statement from the analyzed mouth movements, means for displaying the estimated content of the statement, means for recognizing the user's emotions, and means for adjusting the display format of the content of the statement according to the user's emotions. This makes it possible to display the content of the statement according to the user's emotional state, thereby improving the quality of communication.

[1780] "Means of tracking a user's gaze" refers to all devices, including sensors and software, that detect which direction a user is looking.

[1781] "Means for analyzing a speaker's mouth movements" refers to all devices, including image processing technologies and algorithms, that analyze the movements and shape of a speaker's mouth to estimate what they are saying.

[1782] "Means for estimating the content of speech from analyzed mouth movements" refers to all devices, including speech recognition technology and machine learning models, that use data on mouth movements to estimate what a speaker is saying.

[1783] "Means for displaying estimated speech content" refers to all devices, including displays and projectors, that visually provide users with text and information that has been analyzed and estimated.

[1784] "Means of recognizing user emotions" refers to all devices, including sensors and software, that analyze data such as the user's facial expressions and voice to identify the user's emotional state.

[1785] "Means of adjusting the display format of spoken content according to the user's emotions" refers to all devices, including software and algorithms, that appropriately adjust the visual attributes of the displayed text and information, such as color and font, according to the recognized emotional state.

[1786] "Means of selecting a single speaker from among multiple speakers based on the user's gaze" refers to any device that includes technologies and algorithms for automatically selecting a specific speaker that the user is focusing on, using the user's gaze information, when there are multiple speakers.

[1787] "Means of refining spoken content into natural-sounding sentences based on context" refers to all devices, including natural language processing technologies and generative models, that convert estimated spoken content into more natural and understandable sentences.

[1788] This invention combines a system that tracks the user's gaze, analyzes the speaker's mouth movements to estimate the content of what is being said, and displays the estimated content with an emotion engine that recognizes the user's emotions. This system allows hearing-impaired individuals to communicate with others without using their hands, and improves the quality of communication by presenting the content of what is being said in a display format that corresponds to the user's emotions.

[1789] System Configuration

[1790] This system consists of three main components: a terminal (e.g., smart glasses), a server, and an emotion engine.

[1791] Device (smart glasses)

[1792] The device has the following features:

[1793] Eye-tracking function: Tracks the user's gaze and collects gaze information. For example, using an "eye-tracking sensor".

[1794] Camera: Captures video of the speaker the user is focusing on. For example, use a "camera device".

[1795] Emotion recognition function: Analyzes the user's facial expressions to identify their emotional state. For example, using "emotion recognition software".

[1796] Display: A method of visually displaying the estimated content of the speech, which is built into the smart glasses.

[1797] server

[1798] The server receives data sent from the terminal and performs the following processing:

[1799] Mouth movement analysis: Analyzes the speaker's mouth movements to estimate the content of their speech. For example, using "speech recognition technology."

[1800] Natural Language Processing: This process reshapes estimated speech content into natural-sounding sentences based on context. For example, it uses "natural language processing techniques."

[1801] Integration with the emotion engine: Use data from the emotion engine to improve the accuracy of estimating the content of speech.

[1802] Emotional Engine

[1803] The emotion engine recognizes the user's emotions and has the following functions:

[1804] Sentiment analysis: Analyzes the user's facial expressions and voice to identify their emotional state.

[1805] Display format adjustment: The display format of the user's comments will change according to their emotions.

[1806] Program processing

[1807] The program for this system processes the data in the following steps: First, the smart glasses worn by the user track the user's gaze using eye-tracking sensors to identify what they are looking at. Next, a camera device captures the face of the identified speaker and sends the video to the server.

[1808] The server uses speech recognition technology to analyze the speaker's mouth movements based on the received video data and estimates the content of their speech. Since the estimated content may sound unnatural in context, natural language processing technology is used to reshape it into more natural-sounding sentences.

[1809] Simultaneously, the smart glasses use emotion recognition software to analyze the user's emotions and send the results to a server. The server uses this emotion data to adjust the display format according to the user's emotions and then sends the final text data to the smart glasses.

[1810] Smart glasses display received text data on their screen. In this way, users can obtain information based on their gaze and emotions without using their hands or interacting with others.

[1811] Specific example

[1812] Example 1: One-on-one conversation and emotion recognition

[1813] If user A is having a conversation with friend B:

[1814] The device detects that user A's gaze is directed towards friend B and captures B's face. It then sends the video to the server.

[1815] The server analyzes B's mouth movements and estimates that B said, "It's a nice day today." It then refines this into a more natural-sounding sentence.

[1816] The device recognizes A's smile as an emotion and sends the emotion data to the server.

[1817] The server confirms A's happiness status, formats the text in a positive format, and sends it to the terminal.

[1818] The device displays the text "It's a nice day today" in a brightly colored speech bubble.

[1819] Example 2: Conversation involving multiple people and emotion recognition

[1820] If user A is having a conversation with friends B and C:

[1821] The device detects when user A's gaze moves from B to C, captures the video footage from each location, and sends it to the server.

[1822] The server analyzes the mouth movements of B and C and estimates what each of them is saying.

[1823] The device recognizes A's emotional state and sends the emotional data to the server.

[1824] The server adjusts the display format based on A's emotions and sends it to the terminal.

[1825] The device displays the text "What are your plans for tomorrow?" and "I want to go to the movies" in speech bubble format, with designs appropriate to their respective positions.

[1826] Example of a prompt

[1827] Example 1: Prompts when handling one-on-one conversations

[1828] User A is having a conversation with Friend B. Taking into account User A's gaze and emotional state, transcribe Friend B's statement, "It's a nice day today," and generate the text to display on User A's smart glasses.

[1829] Example 2: Prompts when handling conversations with multiple people

[1830] User A is in a conversation with friend B and friend C. User A's gaze shifts from B to C. Transcribe friend B's statement, "What are your plans for tomorrow?" and friend C's statement, "I want to go to the movies," and generate text to display on the user's smart glasses. Display the text in an appropriate format based on User A's gaze and emotional state.

[1831] In this way, the system integrates user eye tracking, speaker lip-syncing, emotion recognition, and natural language processing and display adjustment of spoken content to provide a high-quality communication experience.

[1832] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1833] Step 1:

[1834] Eye tracking and speaker identification

[1835] Input: User eye-tracking data (real-time location information)

[1836] Processing: The device uses eye-tracking technology to track the user's gaze in real time. It identifies the area the user is fixated on based on their gaze.

[1837] Output: Information about the direction the user is looking.

[1838] Specific operation: In a scenario where the user is talking to friend B, the device uses an eye-tracking sensor to detect the user's gaze direction and identify the face in that direction.

[1839] Step 2:

[1840] Speaker video capture

[1841] Input: User's gaze information, location information of the speaker they are looking at.

[1842] Processing: The device captures video of the identified speaker using its camera and obtains the video data.

[1843] Output: Speaker's video data

[1844] Specific operation: The device uses its camera to capture video of friend B's face, which the user is looking at, and sends that data to the server.

[1845] Step 3:

[1846] Analysis of mouth movements and estimation of spoken content

[1847] Input: Speaker's video data

[1848] Processing: The server uses the received video data and applies speech recognition technology to analyze the speaker's mouth movements. Based on the analysis results, it estimates what the speaker is saying.

[1849] Output: Estimated content of the statement (text data)

[1850] Specific operation: The server analyzes the speaker's mouth movements and estimates the content of the statement, for example, "It's a nice day today." The estimated text data is then retrieved by a generating AI model.

[1851] Step 4:

[1852] Reshaping the content of the statement

[1853] Input: Estimated content of the statement (text data)

[1854] Processing: The server uses a generative AI model to format the spoken content into contextually natural-sounding sentences.

[1855] Output: Formatted speech (natural-sounding text)

[1856] Specific operation: The server uses natural language processing techniques to format text data that it has estimated to be "It's a nice day today" into a natural-sounding sentence. At this stage, it performs processing to adjust sentence endings and context.

[1857] Step 5:

[1858] Recognition of user emotions

[1859] Input: User's facial expression data, voice data

[1860] Processing: The device uses emotion recognition software to analyze the user's facial expressions and voice to identify their current emotional state.

[1861] Output: User emotion data (happiness, sadness, surprise, etc.)

[1862] Specific operation: The device analyzes the user's smile, recognizes that the user is in a happy state, and sends that emotional data to the server.

[1863] Step 6:

[1864] Adjusting the display format

[1865] Input: Formatted speech (natural-sounding text), user sentiment data

[1866] Processing: The server adjusts the display format of the utterances based on the sentiment data obtained. Specifically, it changes visual attributes such as color, font, and background.

[1867] Output: Adjusted display format information

[1868] Specific operation: The server formats the text in a brightly colored speech bubble format based on the user's happy emotional state.

[1869] Step 7:

[1870] Display of the content of the statement

[1871] Input: Adjusted display format information, formatted speech content

[1872] Processing: The terminal displays the content of the message on the screen based on the received text data and display format information.

[1873] Output: The content of the statement that is visually displayed to the user.

[1874] Specific action: The device displays the text "It's a nice day today" to the user in a brightly colored speech bubble.

[1875] Through the processing steps described above, the present invention realizes communication support based on the user's gaze and emotions.

[1876] (Application Example 2)

[1877] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1878] One of the communication challenges for people with hearing impairments is their limited means of quickly understanding what others are saying. Furthermore, to improve the quality of communication, not only visual information but also the recognition of emotions and appropriate feedback are crucial. Especially in physical stores, smooth communication between employees and customers is essential, making a system that allows employees and customers with hearing impairments to communicate smoothly necessary.

[1879] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1880] In this invention, the server includes means for tracking the user's gaze, means for analyzing the speaker's mouth movements, means for estimating the content of the speech from the analyzed mouth movements, means for displaying the estimated content of the speech, means for recognizing the user's emotions, means for adjusting the display format based on the recognized user's emotions, and means for supporting communication between employees and customers in a physical store. This enables employees to communicate smoothly with customers who are hearing impaired.

[1881] "Means of tracking user gaze" refers to devices or technologies that detect the direction and focus of a user's gaze in real time and acquire data accordingly.

[1882] "Means for analyzing the speaker's mouth movements" refers to devices or technologies that analyze the movements and shape of a speaker's mouth and estimate the content of their speech from that analysis.

[1883] "Methods for estimating the content of a statement from analyzed mouth movements" refers to algorithms or devices that estimate the content of a statement based on data obtained from the speaker's mouth movements.

[1884] "Means for displaying estimated speech content" refers to technologies such as displays and projection devices for visually displaying estimated speech content.

[1885] "Means of recognizing user emotions" refer to sensors and algorithms that identify emotional states from the user's facial expressions, posture, voice, etc.

[1886] "Means for adjusting the display format based on the recognized user's emotions" refers to devices or technologies for dynamically changing the display format of a user's statements according to their emotional state.

[1887] "Means of supporting communication between employees and customers in physical stores" refers to systems and applications that support smooth communication between employees and customers in physical stores.

[1888] This invention is a system that supports smooth communication between hearing-impaired customers and employees in physical stores. This system consists of the following main components:

[1889] hardware

[1890] 1. Smart Glasses

[1891] Eye-tracking device: Detects the user's (employee's) gaze direction and point of focus in real time.

[1892] Camera device: Captures video of the speaker's (customer's) face and sends the data to the server.

[1893] Display: Visually displays the estimated content of the statement.

[1894] 2. Server

[1895] Data Analysis Unit: Analyzes the speaker's mouth movements to estimate the content of their speech.

[1896] Natural Language Processing Unit: Reshapes estimated speech content into natural-sounding sentences based on context.

[1897] Emotion Recognition Unit: Analyzes the user's emotional state and adjusts the display format accordingly.

[1898] software

[1899] 1. Frontend Program (JavaScript)

[1900] The smart glasses collect data from various sensors and transmit it to the server in real time.

[1901] 2. Backend program (Python, TensorFlow, OpenCV)

[1902] We perform video data analysis, natural language processing, and emotion recognition.

[1903] Data processing flow

[1904] The server receives user eye-tracking data and speaker video data, analyzes the speaker's mouth movements to estimate the content of the speech. Next, the estimated content is analyzed by a natural language processing unit and refined into natural-sounding sentences. The emotion recognition unit analyzes the user's emotional state and adjusts the display format of the speech content based on that data.

[1905] The smart glasses receive text data sent from a server and display it on the screen in a speech bubble format. The display position and design are adjusted based on the user's gaze information and emotional state.

[1906] Specific example

[1907] As an example, consider a situation where a customer asks about the price of a product.

[1908] 1. When a customer shows a product to an employee wearing smart glasses, the smart glasses' eye-tracking device detects the employee's gaze and captures the customer's face with a camera.

[1909] 2. The server analyzes the video data, estimates the spoken word (e.g., "How much is this product?"), and reshapes it into a natural-sounding sentence.

[1910] 3. The smart glasses' emotion recognition unit identifies the customer's interest from their facial expressions and displays text in a positive format.

[1911] 4. Employees can see clearly and visually appropriate text and respond appropriately to customers.

[1912] Example of inputting prompt text into a generative AI model:

[1913] Please describe a system that supports communication between staff and customers in physical stores. This system involves staff using smart glasses to estimate and visually display what customers are saying, utilizing eye-tracking and emotion recognition capabilities. Please provide a detailed explanation, including specific examples of how it works.

[1914] Thus, this invention is expected to effectively support communication between employees and customers with hearing impairments in physical stores and deepen understanding between them.

[1915] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1916] Step 1:

[1917] The device (smart glasses) uses an eye-tracking device to acquire the user's (employee's) gaze information in real time.

[1918] Input: User eye-tracking data

[1919] Processing: Detection of gaze direction and point of gaze

[1920] Output: Eye-tracking information

[1921] Step 2:

[1922] The device captures the speaker's (customer's) face with a camera based on eye-tracking information and acquires the video data.

[1923] Input: Eye-tracking information

[1924] Processing: Identify the speaker's face and capture the video.

[1925] Output: Video data

[1926] Step 3:

[1927] The device sends the acquired video data to the server.

[1928] Input: Video data

[1929] Processing: Sending video data to the server

[1930] Output: Video data sent to the server

[1931] Step 4:

[1932] The server analyzes the transmitted video data and estimates the content of what the speaker is saying based on their mouth movements.

[1933] Input: Video data

[1934] Processing: Analysis of the speaker's mouth movements and estimation of the content of their speech.

[1935] Output: Estimated content of the statement

[1936] Step 5:

[1937] The server analyzes the estimated speech content using a natural language processing unit and formats it into natural-sounding sentences.

[1938] Input: Estimated content of the statement

[1939] Processing: Refine the text based on context using a natural language processing model.

[1940] Output: A well-formatted text

[1941] Step 6:

[1942] The terminal analyzes the user's (employee's) facial expressions and voice using an emotion recognition unit to identify the user's emotional state. This data is then sent to the server.

[1943] Input: Facial expression data, audio data

[1944] Processing: Identify emotional states using an emotion recognition algorithm.

[1945] Output: Sentiment data

[1946] Step 7:

[1947] The server uses sentiment data to adjust the display format and generate speech content with improved estimation accuracy.

[1948] Input: Sentimental data, formatted text

[1949] Processing: Adjusting display format, regenerating message content

[1950] Output: Adjusted display format and content of the message

[1951] Step 8:

[1952] The server sends the generated text data and display format information to the terminal.

[1953] Input: Adjusted display format and content of the statement

[1954] Processing: Sending text data and display format information to the terminal.

[1955] Output: Text data and display format information sent to the terminal

[1956] Step 9:

[1957] The device displays received text data in a speech bubble format on its screen. The display position and design are adjusted based on the user's eye gaze and emotional state.

[1958] Input: Received text data and display format information

[1959] Processing: Display in speech bubble format, adjust display position and design.

[1960] Output: The content of the statement displayed on the screen

[1961] The above outlines the specific processing steps of the system of the present invention. This enables smooth communication between hearing-impaired customers and employees in physical stores.

[1962] 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 controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1963] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1964] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[1965] Furthermore, the emotion identification model 59, acting 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 a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1966] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[1967] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[1968] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[1969] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[1970] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is 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 the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[1971] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[1972] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[1973] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[1974] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[1975] 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.

[1976] Furthermore, it is not necessary to store the entirety 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 the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[1977] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[1978] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of 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). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[1979] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[1980] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[1981] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[1982] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

[1983] The following is further disclosed regarding the embodiments described above.

[1984] (Claim 1)

[1985] A means of tracking the user's gaze,

[1986] A means of analyzing the speaker's mouth movements,

[1987] A method for estimating the content of speech from analyzed mouth movements,

[1988] A system that includes means for displaying the estimated content of a statement.

[1989] (Claim 2)

[1990] The system according to claim 1, comprising means for selecting one speaker from among multiple speakers based on the user's gaze.

[1991] (Claim 3)

[1992] The system according to claim 1, comprising means for refining spoken content into natural-sounding sentences based on context.

[1993] (Claim 4)

[1994] The system according to claim 1, comprising means for displaying the content of a statement in speech bubble format.

[1995] (Claim 5)

[1996] The system according to claim 1, further comprising means for adjusting the display position of the spoken content for each speaker the user is focusing on.

[1997] "Example 1"

[1998] (Claim 1)

[1999] A means of tracking the user's gaze,

[2000] A means of analyzing the speaker's mouth movements,

[2001] A method for estimating the content of speech from analyzed mouth movements,

[2002] A means of displaying the estimated content of the statement,

[2003] A means of collecting video data from a camera,

[2004] A system that includes means for processing speech content estimated using natural language processing into natural-sounding sentences.

[2005] (Claim 2)

[2006] The system according to claim 1, comprising means for identifying a speaker based on the user's gaze and transmitting video data of the identified speaker to a server.

[2007] (Claim 3)

[2008] The system according to claim 1, comprising means for analyzing video data received by a server and estimating the content of a statement with high accuracy.

[2009] "Application Example 1"

[2010] (Claim 1)

[2011] A means of tracking the user's gaze,

[2012] A means of analyzing the speaker's mouth movements,

[2013] A method for estimating the content of speech from analyzed mouth movements,

[2014] A means of displaying the estimated content of the statement,

[2015] A means of visually displaying work instructions,

[2016] A means of facilitating communication between robots and other workers.

[2017] A system that includes this.

[2018] (Claim 2)

[2019] The system according to claim 1, comprising means for selecting one speaker from among multiple speakers based on the user's gaze.

[2020] (Claim 3)

[2021] The system according to claim 1, which includes means for refining spoken content into natural-sounding sentences based on context.

[2022] "Example 2 of combining an emotion engine"

[2023] (Claim 1)

[2024] A means of tracking the user's gaze,

[2025] A means of analyzing the speaker's mouth movements,

[2026] A method for estimating the content of speech from analyzed mouth movements,

[2027] A means of displaying the estimated content of the statement,

[2028] Means of recognizing user emotions,

[2029] A system that includes means for adjusting the display format of user comments according to the user's emotions.

[2030] (Claim 2)

[2031] The system according to claim 1, comprising means for selecting one speaker from among multiple speakers based on the user's gaze.

[2032] (Claim 3)

[2033] The system according to claim 1, comprising means for refining spoken content into natural-sounding sentences based on context.

[2034] "Application example 2 when combining with an emotional engine"

[2035] (Claim 1)

[2036] A means of tracking the user's gaze,

[2037] A means of analyzing the speaker's mouth movements,

[2038] A method for estimating the content of speech from analyzed mouth movements,

[2039] A means of displaying the estimated content of the statement,

[2040] Means of recognizing user emotions,

[2041] A means of adjusting the display format based on the recognized emotions of the user,

[2042] A system that includes means to support communication between employees and customers in physical stores.

[2043] (Claim 2)

[2044] The system according to claim 1, comprising means for selecting one speaker from among multiple speakers based on the user's gaze.

[2045] (Claim 3)

[2046] The system according to claim 1, comprising means for refining spoken content into natural-sounding sentences based on context. [Explanation of Symbols]

[2047] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots< / url:> < / url:> < / url:> < / url:>

Claims

1. A means of tracking the user's gaze, A means of analyzing the speaker's mouth movements, A method for estimating the content of speech from analyzed mouth movements, A system that includes means for displaying the estimated content of a statement.

2. The system according to claim 1, comprising means for selecting one speaker from among multiple speakers based on the user's gaze.

3. The system according to claim 1, comprising means for refining spoken content into natural-sounding sentences based on context.

4. The system according to claim 1, comprising means for displaying the content of a statement in speech bubble format.

5. The system according to claim 1, further comprising means for adjusting the display position of the spoken content for each speaker the user is focusing on.

Images