system

Abstract

We provide the system. [Solution] To assist the user's speech, an imaging means for detecting mouth movements is provided, An analysis means for comparing the mouth movements obtained by the imaging means with phoneme data, A speech generation means that converts phoneme data identified by an analysis means into speech data, An output means that outputs audio data generated by the audio generation means, A system that includes this.

Description

Technical Field

[0005] ,

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

Background Art

[0002] Patent Document 1 discloses a persona chatbot control method 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. <00​​​​​​​​​​​​​​​​​​​​​​​​​​​​This invention provides a system that generates speech data quickly and accurately by using an imaging means to detect the movement of the user's mouth and comparing the real-time mouth movement data with phoneme data using an analysis means. The analysis means generates speech data that closely resembles natural speech using pre-registered reference data and a machine learning model. The generated speech data is presented externally as speech through an output means, enabling the user to communicate smoothly through speech. This effectively solves the communication challenges faced by people who cannot speak.

[0006] A "user" refers to an individual or group that uses this system and is the person who receives assistance with voice-based communication.

[0007] "Imaging means" refers to a device or function for capturing the user's mouth movements in real time, and specifically includes a camera.

[0008] "Analysis means" refers to a device or algorithm for comparing mouth movements obtained by imaging means with phoneme data and identifying the corresponding phonemes.

[0009] "Speech generation means" refers to a device or function for generating speech data using phoneme data identified by analysis means.

[0010] "Audio data" refers to digital or analog audio information in a format that can be heard, generated by an audio generation means.

[0011] "Output means" refers to a device or function for presenting audio data generated by the audio generation means to an external source, and specifically includes speakers, displays, and the like.

[0012] A "machine learning model" is a pre-trained algorithm or program designed to assist in the analysis of a user's mouth movements.

[0013] "Reference data" refers to a pre-registered dataset of mouth movements used as a comparison target in the analysis. [Brief explanation of the drawing]

[0014] [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] This is a sequence diagram showing the processing flow of the data processing system in Example 2, which incorporates an emotion engine. [Figure 14]It is a sequence diagram showing the processing flow of a data processing system in Application Example 2 when a sentiment engine is combined.

Embodiments for Carrying Out the Invention

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

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

[0017] In the following embodiments, a 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.

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

[0019] In the following embodiments, a 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, and the like.

[0020] 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).

[0021] 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."

[0022] [First Embodiment]

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

[0024] 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.

[0025] 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).

[0026] 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.

[0027] 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.

[0028] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form 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.

[0029] 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.

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

[0031] 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.

[0032] 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.

[0033] 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.

[0034] 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".

[0035] The system according to the present invention provides a means for generating speech using mouth movements for users who have difficulty communicating using voice. This system mainly consists of an "imaging means," an "analysis means," a "speech generation means," and an "output means."

[0036] The user uses a terminal equipped with the system to prepare for recognition of their mouth movements. The terminal's imaging device captures the user's mouth in real time and acquires the video data. This ensures that the user's mouth movements are captured in a timely manner and prepared for the next processing step.

[0037] Next, the analysis mechanism operates on the terminal. The video data of mouth movements acquired by the imaging mechanism is compared by the analysis mechanism with reference data registered in advance. This comparison uses a machine learning model to determine which phoneme corresponds to the user's mouth movements. If analysis is difficult on the terminal side, it cooperates with a server and utilizes the server's analysis capabilities.

[0038] The phonemes identified by the analysis means are converted into speech data by the speech generation means. Based on the analysis results, the speech generation means concatenates the phonemes to generate the speech that the user intends to express. During this process, adjustments are also made to make the speech sound more natural.

[0039] Ultimately, the generated audio data is output through an output device in a format that can be heard by the user and those around them. For example, when a user moves their mouth to say "hello," the device analyzes that movement, generates the voice of "hello," and outputs it through the speaker. This system enables smooth voice communication.

[0040] This system can be used to facilitate voice communication in various everyday situations and to expand users' communication abilities.

[0041] The following describes the processing flow.

[0042] Step 1:

[0043] The user launches the "LipTalk" app on their device and prepares to point the camera at their mouth. The device then activates the camera and prepares to capture the user's mouth.

[0044] Step 2:

[0045] The device's imaging capabilities capture the user's mouth movements and acquire them as video frames. This generates real-time data of the user's mouth movements.

[0046] Step 3:

[0047] The terminal passes the captured video data to the analysis device. The analysis device applies a machine learning model to compare it with pre-registered reference data and identifies phonemes from the user's mouth movements.

[0048] Step 4:

[0049] If the analysis is complex on the terminal, the terminal sends the data to the server. The server uses its high-performance analysis capabilities to perform optimal phoneme identification. The results are then returned to the terminal.

[0050] Step 5:

[0051] The terminal's voice generation mechanism concatenates phonemes based on the analysis results to generate audible audio data. At this stage, adjustments are made to ensure smooth audio output.

[0052] Step 6:

[0053] The terminal's output mechanism plays the generated audio data through its speaker. The user can receive the output as audio along with visual feedback.

[0054] Step 7:

[0055] Users can utilize a function that allows them to review the content of the outputted audio and the accuracy of the analysis, and to adjust or correct the system for future use, as needed.

[0056] (Example 1)

[0057] 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."

[0058] In conventional voice communication systems, voice generation relies on the accuracy of speech recognition, which sometimes prevents users from accurately conveying what they want to express. Furthermore, for users who have difficulty communicating by voice, the real-time and natural-sounding voice generation was lacking. This, in turn, hindered the smoothness of user communication.

[0059] 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.

[0060] In this invention, the server includes imaging means for detecting mouth movements to assist the user's speech, analysis means for analyzing the acquired data using a machine learning model, and speech generation means for generating natural-sounding speech based on the analysis results. This makes it possible to quickly generate accurate and natural-sounding speech for the content that the user wants to express in voice-based communication.

[0061] The "imaging means" is a device for detecting the movement of the user's mouth and has the function of acquiring video data in real time.

[0062] An "analysis means" is a device that analyzes data obtained by an imaging means and performs phoneme-corresponding analysis using a machine learning model.

[0063] A "speech generation means" is a device that generates natural and smooth speech based on phoneme data identified by an analysis means.

[0064] "Output means" refers to a device that converts the audio data generated by the audio generation means into a format that can be heard by the user and those around them, and outputs it.

[0065] A "neural network" is a type of machine learning model that uses multiple layers to learn complex patterns.

[0066] "Cloud computing resources" refer to computing resources that utilize servers and storage distributed across multiple data centers to perform advanced computational processing.

[0067] A "generative AI model" is an artificial intelligence model that learns from a large amount of data in advance and then makes predictions and generates new data.

[0068] This invention is a system that generates speech based on mouth movements for users who have difficulty communicating using voice. This system mainly consists of a terminal and a server.

[0069] The user uses a device to detect mouth movements. The device is equipped with an imaging device that captures the user's mouth movements in real time. A general-purpose camera or a dedicated sensor can be used for this imaging.

[0070] The acquired video data is transmitted to the terminal's analysis system. The analysis system uses a neural network to analyze the image data and converts mouth movements into phonemes. If analysis is difficult to perform within the terminal, a server assists with the analysis. The server uses computing resources in the cloud to perform the analysis quickly and efficiently.

[0071] The data, converted into phonemes, is then transformed into natural-sounding speech by a speech generation system. Known speech synthesis software is used for speech generation, and adjustments are made between phonemes to provide high-quality, natural-sounding speech.

[0072] Finally, the generated audio is output externally through the device's output mechanism, allowing the user and those nearby to hear it. Output can be achieved using the device's built-in speaker or an external audio device.

[0073] As a concrete example, when a user moves their mouth to say "hello," the device's camera captures the movement, and the analysis means identifies the phonemes corresponding to "konni ch iwa." This sequence of phonemes is then synthesized into smooth speech by the speech generation means and output as "hello" from the speaker.

[0074] An example of a prompt for a generative AI model might be an instruction such as, "Generate natural-sounding speech that responds instantly based on the user's mouth movements." This prompt allows the system to process data quickly and generate speech.

[0075] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0076] Step 1:

[0077] The device captures the user's mouth in real time using an imaging device. The input is video data acquired through the camera, which captures the subtle movements of the user's mouth frame by frame. The output is video data representing the sequential movements of the mouth. Specifically, the camera tracks the user's mouth movements and continuously acquires frames according to those movements.

[0078] Step 2:

[0079] The terminal's analysis method receives the video data obtained in step 1 as input and analyzes phonemes using a machine learning model. Specifically, it extracts features from the video data using a neural network and compares them with known phoneme models. The output is phoneme data corresponding to the user's mouth movements. The operations here involve video data preprocessing, feature extraction, and phoneme classification.

[0080] Step 3:

[0081] If analyzed phoneme data exists, the terminal offloads the analysis process to the server, utilizing the server's advanced computing resources. The input is the intermediate result of the phoneme analysis sent from the terminal, and the output is the final phoneme data precisely analyzed by the server. The server uses cloud computing technology to perform the analysis quickly.

[0082] Step 4:

[0083] The device's speech generation mechanism receives the final phoneme data as input and generates speech data. It performs adjustments to smooth transitions between phonemes and synthesizes natural-sounding speech. The output is speech data that reproduces what the user intended to express. Specifically, the operation is speech synthesis through the concatenation and adjustment of phonemes.

[0084] Step 5:

[0085] The generated audio data is output to external audio equipment via the terminal's output mechanism. The input is the synthesized audio data, and the output is the audio emitted through the speaker. The terminal operates to reproduce the audio clearly using a Bluetooth speaker or similar device. In this step, the audio is converted from digital to analog and the actual audio output takes place.

[0086] (Application Example 1)

[0087] 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."

[0088] Traditionally, technologies that generate speech from lip movements for users who have difficulty communicating verbally have faced challenges in real-time performance and natural-sounding speech generation. Furthermore, the need for smooth voice communication in virtual spaces necessitates more advanced technological solutions.

[0089] 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.

[0090] In this invention, the server includes an imaging device that detects mouth movements, an analysis device that compares the mouth movements obtained by the imaging device with phonological data, a speech generation device that converts the phonological data determined by the analysis device into acoustic data, an output device that outputs the acoustic data generated by the speech generation device, and a communication device that dynamically provides a customer service experience via a visual device and transmits the acoustic data to other users via remote communication. This makes it possible to provide natural voice in real time even in a virtual space.

[0091] "Mouth movements" refer to changes in muscle movement in specific areas of the face, particularly those related to the articulation of sounds.

[0092] An "imaging device" is a device that optically captures an object and acquires its image data. Specifically, this includes cameras and sensors.

[0093] "Phonological data" refers to data related to phonemes and syllables, which are the basic units that make up human speech.

[0094] An "analysis device" is a device used to extract and interpret specific information based on acquired data. It performs analysis using machine learning models, among other methods.

[0095] A "speech generation device" is a device that generates synthesized speech based on analyzed data.

[0096] "Acoustic data" refers to data that represents sound as vibrations of air as digital or analog signals.

[0097] An "output device" is a device used to present generated audio or video information to the user. This includes speakers and displays.

[0098] A "visual device" is a device used to display or present images or videos. This includes commonly used displays and smart glasses.

[0099] A "communication device" is a device that connects audio or video data to other devices or networks and transmits and receives data.

[0100] The system for carrying out this invention operates by combining multiple devices and components. The server operates an imaging device, such as a camera, to detect the user's mouth movements. The imaging device captures video in real time to accurately understand the user's mouth movements. The obtained video data is processed by an analysis device located on the server. This analysis device uses a machine learning model to compare the mouth movements with phonological data and obtain appropriate phonological data.

[0101] The analyzed phoneme data is converted into acoustic data by a speech generator. This speech generator produces effective and natural-sounding speech, making it available for real-time use. The audio data is then transmitted through an output device so that it can be heard by the user and others.

[0102] The system also includes visual and communication devices, enabling the distribution of audio data to other users in remote locations. The visual devices allow for dynamic customer service experiences in virtual spaces. For example, smart glasses can be used to provide natural responses within a virtual store.

[0103] As a concrete example, consider a scenario where a user wants to receive an explanation about a product in a virtual store. When the avatar of the store clerk moves its mouth, the system analyzes the movement and emits voice in real time. Through this mechanism, the user can experience smooth communication within the virtual environment.

[0104] An example of a prompt to input into the generation AI model is: "Design a system that generates speech from the mouth movements of a customer service avatar in a virtual store and transmits it to other users in real time." Based on this prompt, the system performs detailed processing to achieve real-time and natural speech generation.

[0105] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0106] Step 1:

[0107] The server acquires real-time video data of the user's mouth area using an imaging device. The input is a video stream from the camera, and the output is digitized video data. This process captures the video, providing data necessary for subsequent analysis by capturing the user's mouth area in detail.

[0108] Step 2:

[0109] The server transmits the acquired video data to the analysis device. The acquired video data is used as input and passed through a machine learning model to convert mouth movements into phonetic data. The output is the generated phonetic data. The analysis device incorporates a generative AI model, which compares and analyzes the captured video with pre-registered reference data.

[0110] Step 3:

[0111] The server receives phonological data from the analysis device and transfers it to the speech generation device. The phonological data is used as input, and the speech generation device converts it into acoustic data to produce natural-sounding speech. The output consists of acoustic data. In this step, prompts are used to ensure smooth speech generation.

[0112] Step 4:

[0113] The server transmits the generated audio data to the output device, which then outputs it through speakers or headsets in a way that can be heard by the user and others. Its main role is to take audio data as input and output sound. The output device also adjusts the volume and sound quality to achieve optimal audio output.

[0114] Step 5:

[0115] Visual and communication devices in the user's environment support responses using generated acoustic data. Acoustic data is used as input to provide visual feedback and synchronize with remote users. For example, smart glasses can be used to display text on a virtual screen or to share audio with other users.

[0116] 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.

[0117] The system according to the present invention aims to generate speech for users who have difficulty with voice communication by utilizing mouth movements and the user's emotional state. This system mainly includes "imaging means," "analysis means," "speech generation means," "output means," and "emotion engine."

[0118] The user prepares for voice generation using a device equipped with the system, utilizing the movement of their mouth. The device's imaging device captures the user's mouth in real time, acquiring video data to understand the user's desire to speak. In addition, an emotion engine analyzes the user's entire face to determine their emotional state and collect emotional data.

[0119] On the device, the analysis means uses the video data of the mouth obtained by the imaging means and identifies phonemes by comparing it with pre-registered reference data. A machine learning model is involved in this analysis. Furthermore, the emotion engine uses an emotion database to estimate the user's current emotional state from their facial expressions.

[0120] The phonemes identified by the analysis means are converted into speech data by the speech generation means. At that time, the tone and intonation of the generated speech are adjusted based on the user's emotions identified by the emotion engine, resulting in more natural and emotionally rich speech.

[0121] Ultimately, the generated audio data is provided externally through an output device and output in a format that can be heard by the user and those around them. For example, if the user moves their mouth to express gratitude by saying "thank you," the system recognizes that emotion and generates and plays back the voice saying "thank you" in a tone that reflects the nuance of gratitude. In this way, the system is able to accurately convey the user's intentions.

[0122] This system has the potential to improve users' quality of life by providing communication support that takes both voice and emotion into consideration.

[0123] The following describes the processing flow.

[0124] Step 1:

[0125] The user picks up the device and launches the "LipTalk" app. The device activates its camera and prepares to capture the user's mouth and entire face.

[0126] Step 2:

[0127] The device's imaging capabilities capture the user's mouth movements and entire face in real time, generating video data of their movements and facial expressions.

[0128] Step 3:

[0129] The terminal sends the captured video data to the analysis device. The analysis device starts processing to identify phonemes from the video of the mouth. In this process, a machine learning model recognizes phoneme patterns using pre-registered reference data.

[0130] Step 4:

[0131] The emotion engine within the device analyzes video data of the entire face and estimates the user's emotional state from their facial expressions. The emotion engine uses an emotion database to perform facial recognition.

[0132] Step 5:

[0133] Phoneme data identified by the analysis means and emotion information estimated by the emotion engine are integrated. The terminal's voice generation means generates voice data based on this data. During this generation process, tone and intonation adjustments are made to reflect the emotion information.

[0134] Step 6:

[0135] The terminal's output method plays the generated audio data through its speaker. The user can then confirm that the audio, which reflects their intentions and emotions, has been appropriately output.

[0136] Step 7:

[0137] Depending on the situation, users can review system feedback and make adjustments to improve the accuracy of future analyses.

[0138] (Example 2)

[0139] 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".

[0140] Traditionally, there has been a lack of effective voice-based communication methods for users who have difficulty with voice communication. Furthermore, accurately expressing users' intentions and emotions has been challenging, leading to a decline in the quality of communication.

[0141] 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.

[0142] In this invention, the server includes imaging means for detecting the user's mouth movements, analysis means for comparing the mouth movements obtained by the imaging means with phoneme data, and emotion analysis means for analyzing the user's emotional state in order to adjust the tone and intonation of the generated speech data. This enables the generation of natural and emotionally rich speech using the user's mouth movements and emotional state, making more effective communication possible.

[0143] The "imaging means" refers to hardware that detects the movement of the user's mouth and acquires it as video data.

[0144] "Analysis means" refers to software or hardware that has the function of analyzing the movement of the mouth obtained by the imaging means and comparing it with phoneme data for identification.

[0145] "Speech generation means" refers to a process or apparatus for converting phoneme data identified by analysis means into speech data.

[0146] "Emotion analysis means" refers to a part of a system that has the function of analyzing facial expressions from the user's entire face and estimating their emotional state.

[0147] "Audio data" refers to data in digital or analog format that has been converted into an acoustic signal.

[0148] "Output means" refers to a device or process for transmitting the generated audio data to an external source.

[0149] The system according to the present invention provides a technology that generates natural speech using mouth movements and emotional states for users who have difficulty with voice communication. This system mainly includes "imaging means," "analysis means," "speech generation means," "emotion analysis means," and "output means."

[0150] The device is equipped with an imaging device that captures the user's mouth in real time and acquires video data. A high-resolution camera is used for the imaging device, making it possible to accurately capture the user's desire to speak. In addition, the device captures the user's entire face for emotion analysis and provides that data to the emotion analysis device.

[0151] The terminal's analysis method identifies phonemes by comparing the video data of the mouth obtained by the imaging method with pre-registered reference data. This analysis involves a machine learning model, enabling high-precision phoneme identification.

[0152] The emotion analysis method uses an emotion database to estimate the user's current emotional state from their facial expressions. For example, if the user is smiling, it recognizes the emotion of "happiness."

[0153] The terminal generates audio data using a speech generation method based on phonemes obtained by an analysis method. Here, the tone and intonation of the generated audio are adjusted based on data from the emotion analysis method. This results in the generation of natural-sounding audio that reflects the user's emotions.

[0154] Ultimately, the terminal's output method provides the generated audio data through an external device such as a speaker. This allows the user to faithfully convey their intentions and emotions.

[0155] For example, if a user moves their mouth to express gratitude and say "thank you," the system analyzes that movement and emotion, and generates a voice message saying "thank you" in a grateful tone.

[0156] An example of a prompt message could be: "Recognize the user's mouth movements, analyze their emotions, and generate the following emotionally expressive voice: (Specific mouth movements of the input, phonemes based on reference video data), emotional state (gratitude)."

[0157] In this way, this system realizes communication support that integrates voice and emotion, providing technology that helps users lead richer lives.

[0158] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0159] Step 1:

[0160] The terminal uses an imaging device to capture the user's mouth in real time. The input is the movement of the user's mouth, and the output is high-resolution video data. The imaging device captures the user's urge to speak and transmits this data to the next processing step.

[0161] Step 2:

[0162] The device analyzes the user's entire face using emotion analysis means and estimates their current emotional state. The input is video data of the face captured by the imaging means, and the output is emotion data based on the user's facial expressions. The emotion analysis means compares this data with an emotion database to estimate the user's emotional state.

[0163] Step 3:

[0164] The terminal's analysis method takes in the video data of the mouth obtained in step 1 and uses a machine learning model to identify phonemes. The input is the video data of the mouth, and the output is the identified phoneme data. This analysis procedure performs accurate phoneme recognition by comparing it with existing reference data.

[0165] Step 4:

[0166] The terminal's voice generation means generates voice data using phoneme data obtained from the analysis means. The input is the analyzed phoneme data and emotion data from the emotion analysis means, and the output is synthesized voice data. Based on the data from the emotion analysis means, the tone and intonation of the voice are adjusted to generate natural-sounding voice that reflects the user's emotions.

[0167] Step 5:

[0168] The terminal provides audio data generated by the voice generation means to the outside world through the output means. The input is synthesized audio data, and the output is audio played back through the speaker. This allows the user to communicate their intentions and emotions to those around them.

[0169] (Application Example 2)

[0170] 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".

[0171] In autonomous vehicles, there is a lack of means for the driver to give instructions to the vehicle without speaking. Furthermore, there is a need to achieve safer and more comfortable driving by implementing driving control that reflects the driver's emotional state.

[0172] 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.

[0173] In this invention, the server includes means for detecting mouth movements using a video device implemented in the communication vehicle to assist the user's speech, analysis means for comparing the mouth movements obtained by the video device with phoneme information, and adjustment means for reflecting the user's emotional state in the driving control system. This enables the driver to communicate their intentions to the vehicle without speaking and to control the vehicle in accordance with their emotions.

[0174] A "communication vehicle" is a vehicle equipped with digital devices that can communicate and process data in real time.

[0175] A "video device" is a device that has a shooting function to capture the user's movements and collect visual data.

[0176] "Phoneme information" refers to data that represents the basic units that make up speech.

[0177] "Analysis means" refers to technical methods used to identify specific information based on collected data.

[0178] "Audio information" refers to information that represents the characteristics and content of audio as digital data.

[0179] "Presentation means" refers to a device or mechanism for physically outputting generated audio information.

[0180] "Adjustment means" refers to means that have the function of appropriately changing or adjusting the operation of the system based on the analysis results.

[0181] The system for realizing this invention consists of multiple devices mounted on a communication vehicle. The server first captures the user's mouth movements using a video device. This video device uses an in-vehicle camera, and hardware for analyzing the data in real time, specifically NVIDIA's Jetson, is used.

[0182] The data acquired from the user's movements is processed by an analysis system. This analysis uses a Python program, OpenCV for image processing, and Tensorflow® for phoneme analysis using a machine learning model. The analysis system generates speech information based on the obtained phoneme information, and an adjustment system adjusts the tone and intonation of the speech by incorporating sentiment analysis as needed. This sentiment analysis utilizes a sentiment model that has been pre-trained using deep learning.

[0183] The generated audio information is output by a presentation device and provided to the user and passengers through the car's speakers. For example, if the user moves their mouth to indicate "I want to turn right at the next intersection," the system analyzes that movement and generates an audio message saying "Turning right at the next intersection" which then plays from the speakers, and the vehicle takes control accordingly. Furthermore, if the user is smiling, the system reflects that nuance and outputs the audio in a brighter tone. The prompt in this case is instructed to the model as follows: "Explain how the car should respond when the user gives the next instruction with a smile."

[0184] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0185] Step 1:

[0186] The server uses video equipment to capture the user's mouth movements in real time. The input is video data from cameras installed inside the vehicle. The server acquires this data and converts it into an analyzable format.

[0187] Step 2:

[0188] The server analyzes mouth movements from video data acquired using the analysis method. The input is the video data obtained in step 1. Movement tracking is performed on this data using OpenCV, and phoneme analysis is performed using a TensorFlow machine learning model. The output is the classified phoneme data.

[0189] Step 3:

[0190] The server analyzes the user's facial expressions simultaneously to determine their emotional state. The input is the video data obtained in step 1. Using an emotion engine, a deep learning model is used to analyze facial expressions and derive the user's emotional data. The output of this step is the emotional state determined by the analysis.

[0191] Step 4:

[0192] The server converts the phoneme data based on the analysis into speech information, reflecting the emotional state in the tone and intonation of the speech. The input is the phoneme data from step 2 and the emotional state from step 3. A speech generation means is used to generate speech that includes natural emotional expression. The output of this step is the generated speech data.

[0193] Step 5:

[0194] The server outputs the generated audio data using a presentation device. The input is the audio data created in step 4. The audio is output using the in-car speakers and heard by the user and passengers. Specifically, it transmits vehicle operation commands (e.g., right turn instruction) based on the audio to the vehicle control system. The output consists of the audio presentation and the vehicle operation commands.

[0195] 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.

[0196] Data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (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.

[0197] 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.

[0198] [Second Embodiment]

[0199] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0200] 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.

[0201] 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).

[0202] 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.

[0203] 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.

[0204] 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).

[0205] 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.

[0206] 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.

[0207] 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.

[0208] 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.

[0209] 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.

[0210] 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".

[0211] The system according to the present invention provides a means for generating speech using mouth movements for users who have difficulty communicating using voice. This system mainly consists of an "imaging means," an "analysis means," a "speech generation means," and an "output means."

[0212] The user uses a terminal equipped with the system to prepare for recognition of their mouth movements. The terminal's imaging device captures the user's mouth in real time and acquires the video data. This ensures that the user's mouth movements are captured in a timely manner and prepared for the next processing step.

[0213] Next, the analysis mechanism operates on the terminal. The video data of mouth movements acquired by the imaging mechanism is compared by the analysis mechanism with reference data registered in advance. This comparison uses a machine learning model to determine which phoneme corresponds to the user's mouth movements. If analysis is difficult on the terminal side, it cooperates with a server and utilizes the server's analysis capabilities.

[0214] The phonemes identified by the analysis means are converted into speech data by the speech generation means. Based on the analysis results, the speech generation means concatenates the phonemes to generate the speech that the user intends to express. During this process, adjustments are also made to make the speech sound more natural.

[0215] Ultimately, the generated audio data is output through an output device in a format that can be heard by the user and those around them. For example, when a user moves their mouth to say "hello," the device analyzes that movement, generates the voice of "hello," and outputs it through the speaker. This system enables smooth voice communication.

[0216] This system can be used to facilitate voice communication in various everyday situations and to expand users' communication abilities.

[0217] The following describes the processing flow.

[0218] Step 1:

[0219] The user launches the "LipTalk" app on their device and prepares to point the camera at their mouth. The device then activates the camera and prepares to capture the user's mouth.

[0220] Step 2:

[0221] The device's imaging capabilities capture the user's mouth movements and acquire them as video frames. This generates real-time data of the user's mouth movements.

[0222] Step 3:

[0223] The terminal passes the captured video data to the analysis device. The analysis device applies a machine learning model to compare it with pre-registered reference data and identifies phonemes from the user's mouth movements.

[0224] Step 4:

[0225] If the analysis is complex on the terminal, the terminal sends the data to the server. The server uses its high-performance analysis capabilities to perform optimal phoneme identification. The results are then returned to the terminal.

[0226] Step 5:

[0227] The terminal's voice generation mechanism concatenates phonemes based on the analysis results to generate audible audio data. At this stage, adjustments are made to ensure smooth audio output.

[0228] Step 6:

[0229] The terminal's output mechanism plays the generated audio data through its speaker. The user can receive the output as audio along with visual feedback.

[0230] Step 7:

[0231] Users can utilize a function that allows them to review the content of the outputted audio and the accuracy of the analysis, and to adjust or correct the system for future use, as needed.

[0232] (Example 1)

[0233] 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."

[0234] In conventional voice communication systems, voice generation relies on the accuracy of speech recognition, which sometimes prevents users from accurately conveying what they want to express. Furthermore, for users who have difficulty communicating by voice, the real-time and natural-sounding voice generation was lacking. This, in turn, hindered the smoothness of user communication.

[0235] 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.

[0236] In this invention, the server includes imaging means for detecting mouth movements to assist the user's speech, analysis means for analyzing the acquired data using a machine learning model, and speech generation means for generating natural-sounding speech based on the analysis results. This makes it possible to quickly generate accurate and natural-sounding speech for the content that the user wants to express in voice-based communication.

[0237] The "imaging means" is a device for detecting the movement of the user's mouth and has the function of acquiring video data in real time.

[0238] An "analysis means" is a device that analyzes data obtained by an imaging means and performs phoneme-corresponding analysis using a machine learning model.

[0239] A "speech generation means" is a device that generates natural and smooth speech based on phoneme data identified by an analysis means.

[0240] "Output means" refers to a device that converts the audio data generated by the audio generation means into a format that can be heard by the user and those around them, and outputs it.

[0241] A "neural network" is a type of machine learning model that uses multiple layers to learn complex patterns.

[0242] "Cloud computing resources" refer to computing resources that utilize servers and storage distributed across multiple data centers to perform advanced computational processing.

[0243] A "generative AI model" is an artificial intelligence model that learns from a large amount of data in advance and then makes predictions and generates new data.

[0244] This invention is a system that generates speech based on mouth movements for users who have difficulty communicating using voice. This system mainly consists of a terminal and a server.

[0245] The user uses a device to detect mouth movements. The device is equipped with an imaging device that captures the user's mouth movements in real time. A general-purpose camera or a dedicated sensor can be used for this imaging.

[0246] The acquired video data is transmitted to the terminal's analysis system. The analysis system uses a neural network to analyze the image data and converts mouth movements into phonemes. If analysis is difficult to perform within the terminal, a server assists with the analysis. The server uses computing resources in the cloud to perform the analysis quickly and efficiently.

[0247] The data, converted into phonemes, is then transformed into natural-sounding speech by a speech generation system. Known speech synthesis software is used for speech generation, and adjustments are made between phonemes to provide high-quality, natural-sounding speech.

[0248] Finally, the generated audio is output externally through the device's output mechanism, allowing the user and those nearby to hear it. Output can be achieved using the device's built-in speaker or an external audio device.

[0249] As a concrete example, when a user moves their mouth to say "hello," the device's camera captures the movement, and the analysis means identifies the phonemes corresponding to "konni ch iwa." This sequence of phonemes is then synthesized into smooth speech by the speech generation means and output as "hello" from the speaker.

[0250] An example of a prompt for a generative AI model might be an instruction such as, "Generate natural-sounding speech that responds instantly based on the user's mouth movements." This prompt allows the system to process data quickly and generate speech.

[0251] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0252] Step 1:

[0253] The device captures the user's mouth in real time using an imaging device. The input is video data acquired through the camera, which captures the subtle movements of the user's mouth frame by frame. The output is video data representing the sequential movements of the mouth. Specifically, the camera tracks the user's mouth movements and continuously acquires frames according to those movements.

[0254] Step 2:

[0255] The terminal's analysis method receives the video data obtained in step 1 as input and analyzes phonemes using a machine learning model. Specifically, it extracts features from the video data using a neural network and compares them with known phoneme models. The output is phoneme data corresponding to the user's mouth movements. The operations here involve video data preprocessing, feature extraction, and phoneme classification.

[0256] Step 3:

[0257] If analyzed phoneme data exists, the terminal offloads the analysis process to the server, utilizing the server's advanced computing resources. The input is the intermediate result of the phoneme analysis sent from the terminal, and the output is the final phoneme data precisely analyzed by the server. The server uses cloud computing technology to perform the analysis quickly.

[0258] Step 4:

[0259] The device's speech generation mechanism receives the final phoneme data as input and generates speech data. It performs adjustments to smooth transitions between phonemes and synthesizes natural-sounding speech. The output is speech data that reproduces what the user intended to express. Specifically, the operation is speech synthesis through the concatenation and adjustment of phonemes.

[0260] Step 5:

[0261] The generated audio data is output to external audio equipment via the terminal's output mechanism. The input is the synthesized audio data, and the output is the audio emitted through the speaker. The terminal operates to reproduce the audio clearly using a Bluetooth speaker or similar device. In this step, the audio is converted from digital to analog and the actual audio output takes place.

[0262] (Application Example 1)

[0263] 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."

[0264] Traditionally, technologies that generate speech from lip movements for users who have difficulty communicating verbally have faced challenges in real-time performance and natural-sounding speech generation. Furthermore, the need for smooth voice communication in virtual spaces necessitates more advanced technological solutions.

[0265] 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.

[0266] In this invention, the server includes an imaging device that detects mouth movements, an analysis device that compares the mouth movements obtained by the imaging device with phonological data, a speech generation device that converts the phonological data determined by the analysis device into acoustic data, an output device that outputs the acoustic data generated by the speech generation device, and a communication device that dynamically provides a customer service experience via a visual device and transmits the acoustic data to other users via remote communication. This makes it possible to provide natural voice in real time even in a virtual space.

[0267] "Mouth movements" refer to changes in muscle movement in specific areas of the face, particularly those related to the articulation of sounds.

[0268] An "imaging device" is a device that optically captures an object and acquires its image data. Specifically, this includes cameras and sensors.

[0269] "Phonological data" refers to data related to phonemes and syllables, which are the basic units that make up human speech.

[0270] An "analysis device" is a device used to extract and interpret specific information based on acquired data. It performs analysis using machine learning models, among other methods.

[0271] A "speech generation device" is a device that generates synthesized speech based on analyzed data.

[0272] "Acoustic data" refers to data that represents sound as vibrations of air as digital or analog signals.

[0273] An "output device" is a device used to present generated audio or video information to the user. This includes speakers and displays.

[0274] A "visual device" is a device used to display or present images or videos. This includes commonly used displays and smart glasses.

[0275] A "communication device" is a device that connects audio or video data to other devices or networks and transmits and receives data.

[0276] The system for carrying out this invention operates by combining multiple devices and components. The server operates an imaging device, such as a camera, to detect the user's mouth movements. The imaging device captures video in real time to accurately understand the user's mouth movements. The obtained video data is processed by an analysis device located on the server. This analysis device uses a machine learning model to compare the mouth movements with phonological data and obtain appropriate phonological data.

[0277] The analyzed phoneme data is converted into acoustic data by a speech generator. This speech generator produces effective and natural-sounding speech, making it available for real-time use. The audio data is then transmitted through an output device so that it can be heard by the user and others.

[0278] The system also includes visual and communication devices, enabling the distribution of audio data to other users in remote locations. The visual devices allow for dynamic customer service experiences in virtual spaces. For example, smart glasses can be used to provide natural responses within a virtual store.

[0279] As a concrete example, consider a scenario where a user wants to receive an explanation about a product in a virtual store. When the avatar of the store clerk moves its mouth, the system analyzes the movement and emits voice in real time. Through this mechanism, the user can experience smooth communication within the virtual environment.

[0280] An example of a prompt sentence to be input into the generation AI model is "Please design a system that generates audio from the lip movements of a customer service avatar in a virtual store and transmits it to other users in real time." Based on this prompt sentence, the system performs detailed processing to achieve real-time and natural audio generation.

[0281] The flow of the specific process in Application Example 1 will be described using FIG. 12.

[0282] Step 1:

[0283] The server acquires real-time video data of the user's lips by an imaging device. As input, there is a video stream from the camera, and as output, digitized video data is obtained. In this process, video capture is performed, and by capturing the user's lips in detail, data necessary for subsequent analysis is provided.

[0284] Step 2:

[0285] The server transmits the acquired video data to an analysis device. At this time, taking the acquired video data as input, it passes through a machine learning model to convert the lip movement into phonetic data. As output, phonetic data is generated. The analysis device incorporates a generation AI model, and here, the captured video and pre-registered reference data are compared and analyzed.

[0286] Step 3:

[0287] The server receives the phonetic data from the analysis device and transfers it to an audio generation device. Taking the phonetic data as input, it converts this into acoustic data in the audio generation device to generate natural audio. As output, acoustic data is constituted. In this step, the prompt sentence is used to smoothly generate audio.

[0288] Step 4:

[0289] The server transmits the generated audio data to the output device, which then outputs it through speakers or headsets in a way that can be heard by the user and others. Its main role is to take audio data as input and output sound. The output device also adjusts the volume and sound quality to achieve optimal audio output.

[0290] Step 5:

[0291] Visual and communication devices in the user's environment support responses using generated acoustic data. Acoustic data is used as input to provide visual feedback and synchronize with remote users. For example, smart glasses can be used to display text on a virtual screen or to share audio with other users.

[0292] 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.

[0293] The system according to the present invention aims to generate speech for users who have difficulty with voice communication by utilizing mouth movements and the user's emotional state. This system mainly includes "imaging means," "analysis means," "speech generation means," "output means," and "emotion engine."

[0294] The user prepares for voice generation using a device equipped with the system, utilizing the movement of their mouth. The device's imaging device captures the user's mouth in real time, acquiring video data to understand the user's desire to speak. In addition, an emotion engine analyzes the user's entire face to determine their emotional state and collect emotional data.

[0295] On the device, the analysis means uses the video data of the mouth obtained by the imaging means and identifies phonemes by comparing it with pre-registered reference data. A machine learning model is involved in this analysis. Furthermore, the emotion engine uses an emotion database to estimate the user's current emotional state from their facial expressions.

[0296] The phonemes identified by the analysis means are converted into speech data by the speech generation means. At that time, the tone and intonation of the generated speech are adjusted based on the user's emotions identified by the emotion engine, resulting in more natural and emotionally rich speech.

[0297] Ultimately, the generated audio data is provided externally through an output device and output in a format that can be heard by the user and those around them. For example, if the user moves their mouth to express gratitude by saying "thank you," the system recognizes that emotion and generates and plays back the voice saying "thank you" in a tone that reflects the nuance of gratitude. In this way, the system is able to accurately convey the user's intentions.

[0298] This system has the potential to improve users' quality of life by providing communication support that takes both voice and emotion into consideration.

[0299] The following describes the processing flow.

[0300] Step 1:

[0301] The user picks up the device and launches the "LipTalk" app. The device activates its camera and prepares to capture the user's mouth and entire face.

[0302] Step 2:

[0303] The device's imaging capabilities capture the user's mouth movements and entire face in real time, generating video data of their movements and facial expressions.

[0304] Step 3:

[0305] The terminal sends the captured video data to the analysis means. The analysis means starts the process of discriminating phonemes from the video of the mouth area. In this process, using the pre-registered reference data, the machine learning model recognizes the pattern of phonemes.

[0306] Step 4:

[0307] The emotion engine in the terminal analyzes the video data of the entire face and estimates the user's emotional state from the expression. The emotion engine utilizes the emotion database to perform expression recognition.

[0308] Step 5:

[0309] The phoneme data discriminated by the analysis means and the emotion information estimated by the emotion engine are integrated. The voice generation means of the terminal generates voice data based on these data. In this generation process, the tone and intonation are adjusted to reflect the emotion information.

[0310] Step 6:

[0311] The output means of the terminal plays the generated voice data through the speaker. The user can confirm that the voice reflecting their own intention and emotion is appropriately output.

[0312] Step 7:

[0313] In some cases, the user can check the system feedback and make adjustments to improve the analysis accuracy for subsequent times.

[0314] (Example 2)

[0315] Next, Example 2 will be described. In the following description, the data processing device 12 is referred to as the "server", and the smart glasses 214 are referred to as the "terminal".

[0316] Traditionally, there has been a lack of effective voice-based communication methods for users who have difficulty with voice communication. Furthermore, accurately expressing users' intentions and emotions has been challenging, leading to a decline in the quality of communication.

[0317] 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.

[0318] In this invention, the server includes imaging means for detecting the user's mouth movements, analysis means for comparing the mouth movements obtained by the imaging means with phoneme data, and emotion analysis means for analyzing the user's emotional state in order to adjust the tone and intonation of the generated speech data. This enables the generation of natural and emotionally rich speech using the user's mouth movements and emotional state, making more effective communication possible.

[0319] The "imaging means" refers to hardware that detects the movement of the user's mouth and acquires it as video data.

[0320] "Analysis means" refers to software or hardware that has the function of analyzing the movement of the mouth obtained by the imaging means and comparing it with phoneme data for identification.

[0321] "Speech generation means" refers to a process or apparatus for converting phoneme data identified by analysis means into speech data.

[0322] "Emotion analysis means" refers to a part of a system that has the function of analyzing facial expressions from the user's entire face and estimating their emotional state.

[0323] "Audio data" refers to data in digital or analog format that has been converted into an acoustic signal.

[0324] "Output means" refers to a device or process for transmitting the generated audio data to an external source.

[0325] The system according to the present invention provides a technology that generates natural speech using mouth movements and emotional states for users who have difficulty with voice communication. This system mainly includes "imaging means," "analysis means," "speech generation means," "emotion analysis means," and "output means."

[0326] The device is equipped with an imaging device that captures the user's mouth in real time and acquires video data. A high-resolution camera is used for the imaging device, making it possible to accurately capture the user's desire to speak. In addition, the device captures the user's entire face for emotion analysis and provides that data to the emotion analysis device.

[0327] The terminal's analysis method identifies phonemes by comparing the video data of the mouth obtained by the imaging method with pre-registered reference data. This analysis involves a machine learning model, enabling high-precision phoneme identification.

[0328] The emotion analysis method uses an emotion database to estimate the user's current emotional state from their facial expressions. For example, if the user is smiling, it recognizes the emotion of "happiness."

[0329] The terminal generates audio data using a speech generation method based on phonemes obtained by an analysis method. Here, the tone and intonation of the generated audio are adjusted based on data from the emotion analysis method. This results in the generation of natural-sounding audio that reflects the user's emotions.

[0330] Ultimately, the terminal's output method provides the generated audio data through an external device such as a speaker. This allows the user to faithfully convey their intentions and emotions.

[0331] For example, if a user moves their mouth to express gratitude and say "thank you," the system analyzes that movement and emotion, and generates a voice message saying "thank you" in a grateful tone.

[0332] An example of a prompt message could be: "Recognize the user's mouth movements, analyze their emotions, and generate the following emotionally expressive voice: (Specific mouth movements of the input, phonemes based on reference video data), emotional state (gratitude)."

[0333] In this way, this system realizes communication support that integrates voice and emotion, providing technology that helps users lead richer lives.

[0334] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0335] Step 1:

[0336] The terminal uses an imaging device to capture the user's mouth in real time. The input is the movement of the user's mouth, and the output is high-resolution video data. The imaging device captures the user's urge to speak and transmits this data to the next processing step.

[0337] Step 2:

[0338] The device analyzes the user's entire face using emotion analysis means and estimates their current emotional state. The input is video data of the face captured by the imaging means, and the output is emotion data based on the user's facial expressions. The emotion analysis means compares this data with an emotion database to estimate the user's emotional state.

[0339] Step 3:

[0340] The terminal's analysis method takes in the video data of the mouth obtained in step 1 and uses a machine learning model to identify phonemes. The input is the video data of the mouth, and the output is the identified phoneme data. This analysis procedure performs accurate phoneme recognition by comparing it with existing reference data.

[0341] Step 4:

[0342] The terminal's voice generation means generates voice data using phoneme data obtained from the analysis means. The input is the analyzed phoneme data and emotion data from the emotion analysis means, and the output is synthesized voice data. Based on the data from the emotion analysis means, the tone and intonation of the voice are adjusted to generate natural-sounding voice that reflects the user's emotions.

[0343] Step 5:

[0344] The terminal provides audio data generated by the voice generation means to the outside world through the output means. The input is synthesized audio data, and the output is audio played back through the speaker. This allows the user to communicate their intentions and emotions to those around them.

[0345] (Application Example 2)

[0346] 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."

[0347] In autonomous vehicles, there is a lack of means for the driver to give instructions to the vehicle without speaking. Furthermore, there is a need to achieve safer and more comfortable driving by implementing driving control that reflects the driver's emotional state.

[0348] 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.

[0349] In this invention, the server includes means for detecting mouth movements using a video device implemented in the communication vehicle to assist the user's speech, analysis means for comparing the mouth movements obtained by the video device with phoneme information, and adjustment means for reflecting the user's emotional state in the driving control system. This enables the driver to communicate their intentions to the vehicle without speaking and to control the vehicle in accordance with their emotions.

[0350] A "communication vehicle" is a vehicle equipped with digital devices that can communicate and process data in real time.

[0351] A "video device" is a device that has a shooting function to capture the user's movements and collect visual data.

[0352] "Phoneme information" refers to data that represents the basic units that make up speech.

[0353] "Analysis means" refers to technical methods used to identify specific information based on collected data.

[0354] "Audio information" refers to information that represents the characteristics and content of audio as digital data.

[0355] "Presentation means" refers to a device or mechanism for physically outputting generated audio information.

[0356] "Adjustment means" refers to means that have the function of appropriately changing or adjusting the operation of the system based on the analysis results.

[0357] The system for realizing this invention consists of multiple devices mounted on a communication vehicle. The server first captures the user's mouth movements using a video device. This video device uses an in-vehicle camera, and hardware for analyzing the data in real time, specifically NVIDIA's Jetson, is used.

[0358] The data acquired from the user's movements is processed by an analysis system. This analysis uses a Python program, OpenCV for image processing, and TensorFlow for phoneme analysis using a machine learning model. The analysis system generates speech information based on the obtained phoneme information, and an adjustment system adjusts the tone and intonation of the speech by incorporating sentiment analysis as needed. This sentiment analysis utilizes a sentiment model that has been pre-trained using deep learning.

[0359] The generated audio information is output by a presentation device and provided to the user and passengers through the car's speakers. For example, if the user moves their mouth to indicate "I want to turn right at the next intersection," the system analyzes that movement and generates an audio message saying "Turning right at the next intersection" which then plays from the speakers, and the vehicle takes control accordingly. Furthermore, if the user is smiling, the system reflects that nuance and outputs the audio in a brighter tone. The prompt in this case is instructed to the model as follows: "Explain how the car should respond when the user gives the next instruction with a smile."

[0360] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0361] Step 1:

[0362] The server uses video equipment to capture the user's mouth movements in real time. The input is video data from cameras installed inside the vehicle. The server acquires this data and converts it into an analyzable format.

[0363] Step 2:

[0364] The server analyzes mouth movements from video data acquired using the analysis method. The input is the video data obtained in step 1. Movement tracking is performed on this data using OpenCV, and phoneme analysis is performed using a TensorFlow machine learning model. The output is the classified phoneme data.

[0365] Step 3:

[0366] The server analyzes the user's facial expressions simultaneously to determine their emotional state. The input is the video data obtained in step 1. Using an emotion engine, a deep learning model is used to analyze facial expressions and derive the user's emotional data. The output of this step is the emotional state determined by the analysis.

[0367] Step 4:

[0368] The server converts the phoneme data based on the analysis into speech information, reflecting the emotional state in the tone and intonation of the speech. The input is the phoneme data from step 2 and the emotional state from step 3. A speech generation means is used to generate speech that includes natural emotional expression. The output of this step is the generated speech data.

[0369] Step 5:

[0370] The server outputs the generated audio data using a presentation device. The input is the audio data created in step 4. The audio is output using the in-car speakers and heard by the user and passengers. Specifically, it transmits vehicle operation commands (e.g., right turn instruction) based on the audio to the vehicle control system. The output consists of the audio presentation and the vehicle operation commands.

[0371] 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.

[0372] 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.

[0373] 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.

[0374] [Third Embodiment]

[0375] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0376] 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.

[0377] 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).

[0378] 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.

[0379] 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.

[0380] 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).

[0381] 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.

[0382] 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.

[0383] 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.

[0384] 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.

[0385] 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.

[0386] 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".

[0387] The system according to the present invention provides a means for generating speech using mouth movements for users who have difficulty communicating using voice. This system mainly consists of an "imaging means," an "analysis means," a "speech generation means," and an "output means."

[0388] The user uses a terminal equipped with the system to prepare for recognition of their mouth movements. The terminal's imaging device captures the user's mouth in real time and acquires the video data. This ensures that the user's mouth movements are captured in a timely manner and prepared for the next processing step.

[0389] Next, the analysis mechanism operates on the terminal. The video data of mouth movements acquired by the imaging mechanism is compared by the analysis mechanism with reference data registered in advance. This comparison uses a machine learning model to determine which phoneme corresponds to the user's mouth movements. If analysis is difficult on the terminal side, it cooperates with a server and utilizes the server's analysis capabilities.

[0390] The phonemes identified by the analysis means are converted into speech data by the speech generation means. Based on the analysis results, the speech generation means concatenates the phonemes to generate the speech that the user intends to express. During this process, adjustments are also made to make the speech sound more natural.

[0391] Ultimately, the generated audio data is output through an output device in a format that can be heard by the user and those around them. For example, when a user moves their mouth to say "hello," the device analyzes that movement, generates the voice of "hello," and outputs it through the speaker. This system enables smooth voice communication.

[0392] This system can be used to facilitate voice communication in various everyday situations and to expand users' communication abilities.

[0393] The following describes the processing flow.

[0394] Step 1:

[0395] The user launches the "LipTalk" app on their device and prepares to point the camera at their mouth. The device then activates the camera and prepares to capture the user's mouth.

[0396] Step 2:

[0397] The device's imaging capabilities capture the user's mouth movements and acquire them as video frames. This generates real-time data of the user's mouth movements.

[0398] Step 3:

[0399] The terminal passes the captured video data to the analysis device. The analysis device applies a machine learning model to compare it with pre-registered reference data and identifies phonemes from the user's mouth movements.

[0400] Step 4:

[0401] If the analysis is complex on the terminal, the terminal sends the data to the server. The server uses its high-performance analysis capabilities to perform optimal phoneme identification. The results are then returned to the terminal.

[0402] Step 5:

[0403] The terminal's voice generation mechanism concatenates phonemes based on the analysis results to generate audible audio data. At this stage, adjustments are made to ensure smooth audio output.

[0404] Step 6:

[0405] The terminal's output mechanism plays the generated audio data through its speaker. The user can receive the output as audio along with visual feedback.

[0406] Step 7:

[0407] Users can utilize a function that allows them to review the content of the outputted audio and the accuracy of the analysis, and to adjust or correct the system for future use, as needed.

[0408] (Example 1)

[0409] 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."

[0410] In conventional voice communication systems, voice generation relies on the accuracy of speech recognition, which sometimes prevents users from accurately conveying what they want to express. Furthermore, for users who have difficulty communicating by voice, the real-time and natural-sounding voice generation was lacking. This, in turn, hindered the smoothness of user communication.

[0411] 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.

[0412] In this invention, the server includes imaging means for detecting mouth movements to assist the user's speech, analysis means for analyzing the acquired data using a machine learning model, and speech generation means for generating natural-sounding speech based on the analysis results. This makes it possible to quickly generate accurate and natural-sounding speech for the content that the user wants to express in voice-based communication.

[0413] The "imaging means" is a device for detecting the movement of the user's mouth and has the function of acquiring video data in real time.

[0414] An "analysis means" is a device that analyzes data obtained by an imaging means and performs phoneme-corresponding analysis using a machine learning model.

[0415] A "speech generation means" is a device that generates natural and smooth speech based on phoneme data identified by an analysis means.

[0416] "Output means" refers to a device that converts the audio data generated by the audio generation means into a format that can be heard by the user and those around them, and outputs it.

[0417] A "neural network" is a type of machine learning model that uses multiple layers to learn complex patterns.

[0418] "Cloud computing resources" refer to computing resources that utilize servers and storage distributed across multiple data centers to perform advanced computational processing.

[0419] A "generative AI model" is an artificial intelligence model that learns from a large amount of data in advance and then makes predictions and generates new data.

[0420] This invention is a system that generates speech based on mouth movements for users who have difficulty communicating using voice. This system mainly consists of a terminal and a server.

[0421] The user uses a device to detect mouth movements. The device is equipped with an imaging device that captures the user's mouth movements in real time. A general-purpose camera or a dedicated sensor can be used for this imaging.

[0422] The acquired video data is transmitted to the terminal's analysis system. The analysis system uses a neural network to analyze the image data and converts mouth movements into phonemes. If analysis is difficult to perform within the terminal, a server assists with the analysis. The server uses computing resources in the cloud to perform the analysis quickly and efficiently.

[0423] The data, converted into phonemes, is then transformed into natural-sounding speech by a speech generation system. Known speech synthesis software is used for speech generation, and adjustments are made between phonemes to provide high-quality, natural-sounding speech.

[0424] Finally, the generated audio is output externally through the device's output mechanism, allowing the user and those nearby to hear it. Output can be achieved using the device's built-in speaker or an external audio device.

[0425] As a concrete example, when a user moves their mouth to say "hello," the device's camera captures the movement, and the analysis means identifies the phonemes corresponding to "konni ch iwa." This sequence of phonemes is then synthesized into smooth speech by the speech generation means and output as "hello" from the speaker.

[0426] An example of a prompt for a generative AI model might be an instruction such as, "Generate natural-sounding speech that responds instantly based on the user's mouth movements." This prompt allows the system to process data quickly and generate speech.

[0427] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0428] Step 1:

[0429] The device captures the user's mouth in real time using an imaging device. The input is video data acquired through the camera, which captures the subtle movements of the user's mouth frame by frame. The output is video data representing the sequential movements of the mouth. Specifically, the camera tracks the user's mouth movements and continuously acquires frames according to those movements.

[0430] Step 2:

[0431] The terminal's analysis method receives the video data obtained in step 1 as input and analyzes phonemes using a machine learning model. Specifically, it extracts features from the video data using a neural network and compares them with known phoneme models. The output is phoneme data corresponding to the user's mouth movements. The operations here involve video data preprocessing, feature extraction, and phoneme classification.

[0432] Step 3:

[0433] If analyzed phoneme data exists, the terminal offloads the analysis process to the server, utilizing the server's advanced computing resources. The input is the intermediate result of the phoneme analysis sent from the terminal, and the output is the final phoneme data precisely analyzed by the server. The server uses cloud computing technology to perform the analysis quickly.

[0434] Step 4:

[0435] The device's speech generation mechanism receives the final phoneme data as input and generates speech data. It performs adjustments to smooth transitions between phonemes and synthesizes natural-sounding speech. The output is speech data that reproduces what the user intended to express. Specifically, the operation is speech synthesis through the concatenation and adjustment of phonemes.

[0436] Step 5:

[0437] The generated audio data is output to external audio equipment via the terminal's output mechanism. The input is the synthesized audio data, and the output is the audio emitted through the speaker. The terminal operates to reproduce the audio clearly using a Bluetooth speaker or similar device. In this step, the audio is converted from digital to analog and the actual audio output takes place.

[0438] (Application Example 1)

[0439] 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."

[0440] Traditionally, technologies that generate speech from lip movements for users who have difficulty communicating verbally have faced challenges in real-time performance and natural-sounding speech generation. Furthermore, the need for smooth voice communication in virtual spaces necessitates more advanced technological solutions.

[0441] 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.

[0442] In this invention, the server includes an imaging device that detects mouth movements, an analysis device that compares the mouth movements obtained by the imaging device with phonological data, a speech generation device that converts the phonological data determined by the analysis device into acoustic data, an output device that outputs the acoustic data generated by the speech generation device, and a communication device that dynamically provides a customer service experience via a visual device and transmits the acoustic data to other users via remote communication. This makes it possible to provide natural voice in real time even in a virtual space.

[0443] "Mouth movements" refer to changes in muscle movement in specific areas of the face, particularly those related to the articulation of sounds.

[0444] An "imaging device" is a device that optically captures an object and acquires its image data. Specifically, this includes cameras and sensors.

[0445] "Phonological data" refers to data related to phonemes and syllables, which are the basic units that make up human speech.

[0446] An "analysis device" is a device used to extract and interpret specific information based on acquired data. It performs analysis using machine learning models, among other methods.

[0447] A "speech generation device" is a device that generates synthesized speech based on analyzed data.

[0448] "Acoustic data" refers to data that represents sound as vibrations of air as digital or analog signals.

[0449] An "output device" is a device used to present generated audio or video information to the user. This includes speakers and displays.

[0450] A "visual device" is a device used to display or present images or videos. This includes commonly used displays and smart glasses.

[0451] A "communication device" is a device that connects audio or video data to other devices or networks and transmits and receives data.

[0452] The system for carrying out this invention operates by combining multiple devices and components. The server operates an imaging device, such as a camera, to detect the user's mouth movements. The imaging device captures video in real time to accurately understand the user's mouth movements. The obtained video data is processed by an analysis device located on the server. This analysis device uses a machine learning model to compare the mouth movements with phonological data and obtain appropriate phonological data.

[0453] The analyzed phoneme data is converted into acoustic data by a speech generator. This speech generator produces effective and natural-sounding speech, making it available for real-time use. The audio data is then transmitted through an output device so that it can be heard by the user and others.

[0454] The system also includes visual and communication devices, enabling the distribution of audio data to other users in remote locations. The visual devices allow for dynamic customer service experiences in virtual spaces. For example, smart glasses can be used to provide natural responses within a virtual store.

[0455] As a concrete example, consider a scenario where a user wants to receive an explanation about a product in a virtual store. When the avatar of the store clerk moves its mouth, the system analyzes the movement and emits voice in real time. Through this mechanism, the user can experience smooth communication within the virtual environment.

[0456] An example of a prompt to input into the generation AI model is: "Design a system that generates speech from the mouth movements of a customer service avatar in a virtual store and transmits it to other users in real time." Based on this prompt, the system performs detailed processing to achieve real-time and natural speech generation.

[0457] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0458] Step 1:

[0459] The server acquires real-time video data of the user's mouth area using an imaging device. The input is a video stream from the camera, and the output is digitized video data. This process captures the video, providing data necessary for subsequent analysis by capturing the user's mouth area in detail.

[0460] Step 2:

[0461] The server transmits the acquired video data to the analysis device. The acquired video data is used as input and passed through a machine learning model to convert mouth movements into phonetic data. The output is the generated phonetic data. The analysis device incorporates a generative AI model, which compares and analyzes the captured video with pre-registered reference data.

[0462] Step 3:

[0463] The server receives phonological data from the analysis device and transfers it to the speech generation device. The phonological data is used as input, and the speech generation device converts it into acoustic data to produce natural-sounding speech. The output consists of acoustic data. In this step, prompts are used to ensure smooth speech generation.

[0464] Step 4:

[0465] The server transmits the generated audio data to the output device, which then outputs it through speakers or headsets in a way that can be heard by the user and others. Its main role is to take audio data as input and output sound. The output device also adjusts the volume and sound quality to achieve optimal audio output.

[0466] Step 5:

[0467] Visual and communication devices in the user's environment support responses using generated acoustic data. Acoustic data is used as input to provide visual feedback and synchronize with remote users. For example, smart glasses can be used to display text on a virtual screen or to share audio with other users.

[0468] 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.

[0469] The system according to the present invention aims to generate speech for users who have difficulty with voice communication by utilizing mouth movements and the user's emotional state. This system mainly includes "imaging means," "analysis means," "speech generation means," "output means," and "emotion engine."

[0470] The user prepares for voice generation using a device equipped with the system, utilizing the movement of their mouth. The device's imaging device captures the user's mouth in real time, acquiring video data to understand the user's desire to speak. In addition, an emotion engine analyzes the user's entire face to determine their emotional state and collect emotional data.

[0471] On the device, the analysis means uses the video data of the mouth obtained by the imaging means and identifies phonemes by comparing it with pre-registered reference data. A machine learning model is involved in this analysis. Furthermore, the emotion engine uses an emotion database to estimate the user's current emotional state from their facial expressions.

[0472] The phonemes identified by the analysis means are converted into speech data by the speech generation means. At that time, the tone and intonation of the generated speech are adjusted based on the user's emotions identified by the emotion engine, resulting in more natural and emotionally rich speech.

[0473] Ultimately, the generated audio data is provided externally through an output device and output in a format that can be heard by the user and those around them. For example, if the user moves their mouth to express gratitude by saying "thank you," the system recognizes that emotion and generates and plays back the voice saying "thank you" in a tone that reflects the nuance of gratitude. In this way, the system is able to accurately convey the user's intentions.

[0474] This system has the potential to improve users' quality of life by providing communication support that takes both voice and emotion into consideration.

[0475] The following describes the processing flow.

[0476] Step 1:

[0477] The user picks up the device and launches the "LipTalk" app. The device activates its camera and prepares to capture the user's mouth and entire face.

[0478] Step 2:

[0479] The device's imaging capabilities capture the user's mouth movements and entire face in real time, generating video data of their movements and facial expressions.

[0480] Step 3:

[0481] The terminal sends the captured video data to the analysis device. The analysis device starts processing to identify phonemes from the video of the mouth. In this process, a machine learning model recognizes phoneme patterns using pre-registered reference data.

[0482] Step 4:

[0483] The emotion engine within the device analyzes video data of the entire face and estimates the user's emotional state from their facial expressions. The emotion engine uses an emotion database to perform facial recognition.

[0484] Step 5:

[0485] Phoneme data identified by the analysis means and emotion information estimated by the emotion engine are integrated. The terminal's voice generation means generates voice data based on this data. During this generation process, tone and intonation adjustments are made to reflect the emotion information.

[0486] Step 6:

[0487] The terminal's output method plays the generated audio data through its speaker. The user can then confirm that the audio, which reflects their intentions and emotions, has been appropriately output.

[0488] Step 7:

[0489] Depending on the situation, users can review system feedback and make adjustments to improve the accuracy of future analyses.

[0490] (Example 2)

[0491] 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."

[0492] Traditionally, there has been a lack of effective voice-based communication methods for users who have difficulty with voice communication. Furthermore, accurately expressing users' intentions and emotions has been challenging, leading to a decline in the quality of communication.

[0493] 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.

[0494] In this invention, the server includes imaging means for detecting the user's mouth movements, analysis means for comparing the mouth movements obtained by the imaging means with phoneme data, and emotion analysis means for analyzing the user's emotional state in order to adjust the tone and intonation of the generated speech data. This enables the generation of natural and emotionally rich speech using the user's mouth movements and emotional state, making more effective communication possible.

[0495] The "imaging means" refers to hardware that detects the movement of the user's mouth and acquires it as video data.

[0496] "Analysis means" refers to software or hardware that has the function of analyzing the movement of the mouth obtained by the imaging means and comparing it with phoneme data for identification.

[0497] "Speech generation means" refers to a process or apparatus for converting phoneme data identified by analysis means into speech data.

[0498] "Emotion analysis means" refers to a part of a system that has the function of analyzing facial expressions from the user's entire face and estimating their emotional state.

[0499] "Audio data" refers to data in digital or analog format that has been converted into an acoustic signal.

[0500] "Output means" refers to a device or process for transmitting the generated audio data to an external source.

[0501] The system according to the present invention provides a technology that generates natural speech using mouth movements and emotional states for users who have difficulty with voice communication. This system mainly includes "imaging means," "analysis means," "speech generation means," "emotion analysis means," and "output means."

[0502] The device is equipped with an imaging device that captures the user's mouth in real time and acquires video data. A high-resolution camera is used for the imaging device, making it possible to accurately capture the user's desire to speak. In addition, the device captures the user's entire face for emotion analysis and provides that data to the emotion analysis device.

[0503] The terminal's analysis method identifies phonemes by comparing the video data of the mouth obtained by the imaging method with pre-registered reference data. This analysis involves a machine learning model, enabling high-precision phoneme identification.

[0504] The emotion analysis method uses an emotion database to estimate the user's current emotional state from their facial expressions. For example, if the user is smiling, it recognizes the emotion of "happiness."

[0505] The terminal generates audio data using a speech generation method based on phonemes obtained by an analysis method. Here, the tone and intonation of the generated audio are adjusted based on data from the emotion analysis method. This results in the generation of natural-sounding audio that reflects the user's emotions.

[0506] Ultimately, the terminal's output method provides the generated audio data through an external device such as a speaker. This allows the user to faithfully convey their intentions and emotions.

[0507] For example, if a user moves their mouth to express gratitude and say "thank you," the system analyzes that movement and emotion, and generates a voice message saying "thank you" in a grateful tone.

[0508] An example of a prompt message could be: "Recognize the user's mouth movements, analyze their emotions, and generate the following emotionally expressive voice: (Specific mouth movements of the input, phonemes based on reference video data), emotional state (gratitude)."

[0509] In this way, this system realizes communication support that integrates voice and emotion, providing technology that helps users lead richer lives.

[0510] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0511] Step 1:

[0512] The terminal uses an imaging device to capture the user's mouth in real time. The input is the movement of the user's mouth, and the output is high-resolution video data. The imaging device captures the user's urge to speak and transmits this data to the next processing step.

[0513] Step 2:

[0514] The device analyzes the user's entire face using emotion analysis means and estimates their current emotional state. The input is video data of the face captured by the imaging means, and the output is emotion data based on the user's facial expressions. The emotion analysis means compares this data with an emotion database to estimate the user's emotional state.

[0515] Step 3:

[0516] The terminal's analysis method takes in the video data of the mouth obtained in step 1 and uses a machine learning model to identify phonemes. The input is the video data of the mouth, and the output is the identified phoneme data. This analysis procedure performs accurate phoneme recognition by comparing it with existing reference data.

[0517] Step 4:

[0518] The terminal's voice generation means generates voice data using phoneme data obtained from the analysis means. The input is the analyzed phoneme data and emotion data from the emotion analysis means, and the output is synthesized voice data. Based on the data from the emotion analysis means, the tone and intonation of the voice are adjusted to generate natural-sounding voice that reflects the user's emotions.

[0519] Step 5:

[0520] The terminal provides audio data generated by the voice generation means to the outside world through the output means. The input is synthesized audio data, and the output is audio played back through the speaker. This allows the user to communicate their intentions and emotions to those around them.

[0521] (Application Example 2)

[0522] 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."

[0523] In autonomous vehicles, there is a lack of means for the driver to give instructions to the vehicle without speaking. Furthermore, there is a need to achieve safer and more comfortable driving by implementing driving control that reflects the driver's emotional state.

[0524] 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.

[0525] In this invention, the server includes means for detecting mouth movements using a video device implemented in the communication vehicle to assist the user's speech, analysis means for comparing the mouth movements obtained by the video device with phoneme information, and adjustment means for reflecting the user's emotional state in the driving control system. This enables the driver to communicate their intentions to the vehicle without speaking and to control the vehicle in accordance with their emotions.

[0526] A "communication vehicle" is a vehicle equipped with digital devices that can communicate and process data in real time.

[0527] A "video device" is a device that has a shooting function to capture the user's movements and collect visual data.

[0528] "Phoneme information" refers to data that represents the basic units that make up speech.

[0529] "Analysis means" refers to technical methods used to identify specific information based on collected data.

[0530] "Audio information" refers to information that represents the characteristics and content of audio as digital data.

[0531] "Presentation means" refers to a device or mechanism for physically outputting generated audio information.

[0532] "Adjustment means" refers to means that have the function of appropriately changing or adjusting the operation of the system based on the analysis results.

[0533] The system for realizing this invention consists of multiple devices mounted on a communication vehicle. The server first captures the user's mouth movements using a video device. This video device uses an in-vehicle camera, and hardware for analyzing the data in real time, specifically NVIDIA's Jetson, is used.

[0534] The data acquired from the user's movements is processed by an analysis system. This analysis uses a Python program, OpenCV for image processing, and TensorFlow for phoneme analysis using a machine learning model. The analysis system generates speech information based on the obtained phoneme information, and an adjustment system adjusts the tone and intonation of the speech by incorporating sentiment analysis as needed. This sentiment analysis utilizes a sentiment model that has been pre-trained using deep learning.

[0535] The generated audio information is output by a presentation device and provided to the user and passengers through the car's speakers. For example, if the user moves their mouth to indicate "I want to turn right at the next intersection," the system analyzes that movement and generates an audio message saying "Turning right at the next intersection" which then plays from the speakers, and the vehicle takes control accordingly. Furthermore, if the user is smiling, the system reflects that nuance and outputs the audio in a brighter tone. The prompt in this case is instructed to the model as follows: "Explain how the car should respond when the user gives the next instruction with a smile."

[0536] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0537] Step 1:

[0538] The server uses video equipment to capture the user's mouth movements in real time. The input is video data from cameras installed inside the vehicle. The server acquires this data and converts it into an analyzable format.

[0539] Step 2:

[0540] The server analyzes mouth movements from video data acquired using the analysis method. The input is the video data obtained in step 1. Movement tracking is performed on this data using OpenCV, and phoneme analysis is performed using a TensorFlow machine learning model. The output is the classified phoneme data.

[0541] Step 3:

[0542] The server analyzes the user's facial expressions simultaneously to determine their emotional state. The input is the video data obtained in step 1. Using an emotion engine, a deep learning model is used to analyze facial expressions and derive the user's emotional data. The output of this step is the emotional state determined by the analysis.

[0543] Step 4:

[0544] The server converts the phoneme data based on the analysis into speech information, reflecting the emotional state in the tone and intonation of the speech. The input is the phoneme data from step 2 and the emotional state from step 3. A speech generation means is used to generate speech that includes natural emotional expression. The output of this step is the generated speech data.

[0545] Step 5:

[0546] The server outputs the generated audio data using a presentation device. The input is the audio data created in step 4. The audio is output using the in-car speakers and heard by the user and passengers. Specifically, it transmits vehicle operation commands (e.g., right turn instruction) based on the audio to the vehicle control system. The output consists of the audio presentation and the vehicle operation commands.

[0547] 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.

[0548] 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.

[0549] 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.

[0550] [Fourth Embodiment]

[0551] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[0552] 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.

[0553] 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).

[0554] 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.

[0555] 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.

[0556] 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).

[0557] 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.

[0558] 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.

[0559] 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.

[0560] 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.

[0561] 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.

[0562] 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.

[0563] 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".

[0564] The system according to the present invention provides a means for generating speech using mouth movements for users who have difficulty communicating using voice. This system mainly consists of an "imaging means," an "analysis means," a "speech generation means," and an "output means."

[0565] The user uses a terminal equipped with the system to prepare for recognition of their mouth movements. The terminal's imaging device captures the user's mouth in real time and acquires the video data. This ensures that the user's mouth movements are captured in a timely manner and prepared for the next processing step.

[0566] Next, the analysis mechanism operates on the terminal. The video data of mouth movements acquired by the imaging mechanism is compared by the analysis mechanism with reference data registered in advance. This comparison uses a machine learning model to determine which phoneme corresponds to the user's mouth movements. If analysis is difficult on the terminal side, it cooperates with a server and utilizes the server's analysis capabilities.

[0567] The phonemes identified by the analysis means are converted into speech data by the speech generation means. Based on the analysis results, the speech generation means concatenates the phonemes to generate the speech that the user intends to express. During this process, adjustments are also made to make the speech sound more natural.

[0568] Ultimately, the generated audio data is output through an output device in a format that can be heard by the user and those around them. For example, when a user moves their mouth to say "hello," the device analyzes that movement, generates the voice of "hello," and outputs it through the speaker. This system enables smooth voice communication.

[0569] This system can be used to facilitate voice communication in various everyday situations and to expand users' communication abilities.

[0570] The following describes the processing flow.

[0571] Step 1:

[0572] The user launches the "LipTalk" app on their device and prepares to point the camera at their mouth. The device then activates the camera and prepares to capture the user's mouth.

[0573] Step 2:

[0574] The device's imaging capabilities capture the user's mouth movements and acquire them as video frames. This generates real-time data of the user's mouth movements.

[0575] Step 3:

[0576] The terminal passes the captured video data to the analysis device. The analysis device applies a machine learning model to compare it with pre-registered reference data and identifies phonemes from the user's mouth movements.

[0577] Step 4:

[0578] If the analysis is complex on the terminal, the terminal sends the data to the server. The server uses its high-performance analysis capabilities to perform optimal phoneme identification. The results are then returned to the terminal.

[0579] Step 5:

[0580] The terminal's voice generation mechanism concatenates phonemes based on the analysis results to generate audible audio data. At this stage, adjustments are made to ensure smooth audio output.

[0581] Step 6:

[0582] The terminal's output mechanism plays the generated audio data through its speaker. The user can receive the output as audio along with visual feedback.

[0583] Step 7:

[0584] Users can utilize a function that allows them to review the content of the outputted audio and the accuracy of the analysis, and to adjust or correct the system for future use, as needed.

[0585] (Example 1)

[0586] 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".

[0587] In conventional voice communication systems, voice generation relies on the accuracy of speech recognition, which sometimes prevents users from accurately conveying what they want to express. Furthermore, for users who have difficulty communicating by voice, the real-time and natural-sounding voice generation was lacking. This, in turn, hindered the smoothness of user communication.

[0588] 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.

[0589] In this invention, the server includes imaging means for detecting mouth movements to assist the user's speech, analysis means for analyzing the acquired data using a machine learning model, and speech generation means for generating natural-sounding speech based on the analysis results. This makes it possible to quickly generate accurate and natural-sounding speech for the content that the user wants to express in voice-based communication.

[0590] The "imaging means" is a device for detecting the movement of the user's mouth and has the function of acquiring video data in real time.

[0591] An "analysis means" is a device that analyzes data obtained by an imaging means and performs phoneme-corresponding analysis using a machine learning model.

[0592] A "speech generation means" is a device that generates natural and smooth speech based on phoneme data identified by an analysis means.

[0593] "Output means" refers to a device that converts the audio data generated by the audio generation means into a format that can be heard by the user and those around them, and outputs it.

[0594] A "neural network" is a type of machine learning model that uses multiple layers to learn complex patterns.

[0595] "Cloud computing resources" refer to computing resources that utilize servers and storage distributed across multiple data centers to perform advanced computational processing.

[0596] A "generative AI model" is an artificial intelligence model that learns from a large amount of data in advance and then makes predictions and generates new data.

[0597] This invention is a system that generates speech based on mouth movements for users who have difficulty communicating using voice. This system mainly consists of a terminal and a server.

[0598] The user uses a device to detect mouth movements. The device is equipped with an imaging device that captures the user's mouth movements in real time. A general-purpose camera or a dedicated sensor can be used for this imaging.

[0599] The acquired video data is transmitted to the terminal's analysis system. The analysis system uses a neural network to analyze the image data and converts mouth movements into phonemes. If analysis is difficult to perform within the terminal, a server assists with the analysis. The server uses computing resources in the cloud to perform the analysis quickly and efficiently.

[0600] The data, converted into phonemes, is then transformed into natural-sounding speech by a speech generation system. Known speech synthesis software is used for speech generation, and adjustments are made between phonemes to provide high-quality, natural-sounding speech.

[0601] Finally, the generated audio is output externally through the device's output mechanism, allowing the user and those nearby to hear it. Output can be achieved using the device's built-in speaker or an external audio device.

[0602] As a concrete example, when a user moves their mouth to say "hello," the device's camera captures the movement, and the analysis means identifies the phonemes corresponding to "konni ch iwa." This sequence of phonemes is then synthesized into smooth speech by the speech generation means and output as "hello" from the speaker.

[0603] An example of a prompt for a generative AI model might be an instruction such as, "Generate natural-sounding speech that responds instantly based on the user's mouth movements." This prompt allows the system to process data quickly and generate speech.

[0604] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0605] Step 1:

[0606] The device captures the user's mouth in real time using an imaging device. The input is video data acquired through the camera, which captures the subtle movements of the user's mouth frame by frame. The output is video data representing the sequential movements of the mouth. Specifically, the camera tracks the user's mouth movements and continuously acquires frames according to those movements.

[0607] Step 2:

[0608] The terminal's analysis method receives the video data obtained in step 1 as input and analyzes phonemes using a machine learning model. Specifically, it extracts features from the video data using a neural network and compares them with known phoneme models. The output is phoneme data corresponding to the user's mouth movements. The operations here involve video data preprocessing, feature extraction, and phoneme classification.

[0609] Step 3:

[0610] If analyzed phoneme data exists, the terminal offloads the analysis process to the server, utilizing the server's advanced computing resources. The input is the intermediate result of the phoneme analysis sent from the terminal, and the output is the final phoneme data precisely analyzed by the server. The server uses cloud computing technology to perform the analysis quickly.

[0611] Step 4:

[0612] The device's speech generation mechanism receives the final phoneme data as input and generates speech data. It performs adjustments to smooth transitions between phonemes and synthesizes natural-sounding speech. The output is speech data that reproduces what the user intended to express. Specifically, the operation is speech synthesis through the concatenation and adjustment of phonemes.

[0613] Step 5:

[0614] The generated audio data is output to external audio equipment via the terminal's output mechanism. The input is the synthesized audio data, and the output is the audio emitted through the speaker. The terminal operates to reproduce the audio clearly using a Bluetooth speaker or similar device. In this step, the audio is converted from digital to analog and the actual audio output takes place.

[0615] (Application Example 1)

[0616] 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".

[0617] Traditionally, technologies that generate speech from lip movements for users who have difficulty communicating verbally have faced challenges in real-time performance and natural-sounding speech generation. Furthermore, the need for smooth voice communication in virtual spaces necessitates more advanced technological solutions.

[0618] 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.

[0619] In this invention, the server includes an imaging device that detects mouth movements, an analysis device that compares the mouth movements obtained by the imaging device with phonological data, a speech generation device that converts the phonological data determined by the analysis device into acoustic data, an output device that outputs the acoustic data generated by the speech generation device, and a communication device that dynamically provides a customer service experience via a visual device and transmits the acoustic data to other users via remote communication. This makes it possible to provide natural voice in real time even in a virtual space.

[0620] "Mouth movements" refer to changes in muscle movement in specific areas of the face, particularly those related to the articulation of sounds.

[0621] An "imaging device" is a device that optically captures an object and acquires its image data. Specifically, this includes cameras and sensors.

[0622] "Phonological data" refers to data related to phonemes and syllables, which are the basic units that make up human speech.

[0623] An "analysis device" is a device used to extract and interpret specific information based on acquired data. It performs analysis using machine learning models, among other methods.

[0624] A "speech generation device" is a device that generates synthesized speech based on analyzed data.

[0625] "Acoustic data" refers to data that represents sound as vibrations of air as digital or analog signals.

[0626] An "output device" is a device used to present generated audio or video information to the user. This includes speakers and displays.

[0627] A "visual device" is a device used to display or present images or videos. This includes commonly used displays and smart glasses.

[0628] A "communication device" is a device that connects audio or video data to other devices or networks and transmits and receives data.

[0629] The system for carrying out this invention operates by combining multiple devices and components. The server operates an imaging device, such as a camera, to detect the user's mouth movements. The imaging device captures video in real time to accurately understand the user's mouth movements. The obtained video data is processed by an analysis device located on the server. This analysis device uses a machine learning model to compare the mouth movements with phonological data and obtain appropriate phonological data.

[0630] The analyzed phoneme data is converted into acoustic data by a speech generator. This speech generator produces effective and natural-sounding speech, making it available for real-time use. The audio data is then transmitted through an output device so that it can be heard by the user and others.

[0631] The system also includes visual and communication devices, enabling the distribution of audio data to other users in remote locations. The visual devices allow for dynamic customer service experiences in virtual spaces. For example, smart glasses can be used to provide natural responses within a virtual store.

[0632] As a concrete example, consider a scenario where a user wants to receive an explanation about a product in a virtual store. When the avatar of the store clerk moves its mouth, the system analyzes the movement and emits voice in real time. Through this mechanism, the user can experience smooth communication within the virtual environment.

[0633] An example of a prompt to input into the generation AI model is: "Design a system that generates speech from the mouth movements of a customer service avatar in a virtual store and transmits it to other users in real time." Based on this prompt, the system performs detailed processing to achieve real-time and natural speech generation.

[0634] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0635] Step 1:

[0636] The server acquires real-time video data of the user's mouth area using an imaging device. The input is a video stream from the camera, and the output is digitized video data. This process captures the video, providing data necessary for subsequent analysis by capturing the user's mouth area in detail.

[0637] Step 2:

[0638] The server transmits the acquired video data to the analysis device. The acquired video data is used as input and passed through a machine learning model to convert mouth movements into phonetic data. The output is the generated phonetic data. The analysis device incorporates a generative AI model, which compares and analyzes the captured video with pre-registered reference data.

[0639] Step 3:

[0640] The server receives phonological data from the analysis device and transfers it to the speech generation device. The phonological data is used as input, and the speech generation device converts it into acoustic data to produce natural-sounding speech. The output consists of acoustic data. In this step, prompts are used to ensure smooth speech generation.

[0641] Step 4:

[0642] The server transmits the generated audio data to the output device, which then outputs it through speakers or headsets in a way that can be heard by the user and others. Its main role is to take audio data as input and output sound. The output device also adjusts the volume and sound quality to achieve optimal audio output.

[0643] Step 5:

[0644] Visual and communication devices in the user's environment support responses using generated acoustic data. Acoustic data is used as input to provide visual feedback and synchronize with remote users. For example, smart glasses can be used to display text on a virtual screen or to share audio with other users.

[0645] 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.

[0646] The system according to the present invention aims to generate speech for users who have difficulty with voice communication by utilizing mouth movements and the user's emotional state. This system mainly includes "imaging means," "analysis means," "speech generation means," "output means," and "emotion engine."

[0647] The user prepares for voice generation using a device equipped with the system, utilizing the movement of their mouth. The device's imaging device captures the user's mouth in real time, acquiring video data to understand the user's desire to speak. In addition, an emotion engine analyzes the user's entire face to determine their emotional state and collect emotional data.

[0648] On the device, the analysis means uses the video data of the mouth obtained by the imaging means and identifies phonemes by comparing it with pre-registered reference data. A machine learning model is involved in this analysis. Furthermore, the emotion engine uses an emotion database to estimate the user's current emotional state from their facial expressions.

[0649] The phonemes identified by the analysis means are converted into speech data by the speech generation means. At that time, the tone and intonation of the generated speech are adjusted based on the user's emotions identified by the emotion engine, resulting in more natural and emotionally rich speech.

[0650] Ultimately, the generated audio data is provided externally through an output device and output in a format that can be heard by the user and those around them. For example, if the user moves their mouth to express gratitude by saying "thank you," the system recognizes that emotion and generates and plays back the voice saying "thank you" in a tone that reflects the nuance of gratitude. In this way, the system is able to accurately convey the user's intentions.

[0651] This system has the potential to improve users' quality of life by providing communication support that takes both voice and emotion into consideration.

[0652] The following describes the processing flow.

[0653] Step 1:

[0654] The user picks up the device and launches the "LipTalk" app. The device activates its camera and prepares to capture the user's mouth and entire face.

[0655] Step 2:

[0656] The device's imaging capabilities capture the user's mouth movements and entire face in real time, generating video data of their movements and facial expressions.

[0657] Step 3:

[0658] The terminal sends the captured video data to the analysis device. The analysis device starts processing to identify phonemes from the video of the mouth. In this process, a machine learning model recognizes phoneme patterns using pre-registered reference data.

[0659] Step 4:

[0660] The emotion engine within the device analyzes video data of the entire face and estimates the user's emotional state from their facial expressions. The emotion engine uses an emotion database to perform facial recognition.

[0661] Step 5:

[0662] Phoneme data identified by the analysis means and emotion information estimated by the emotion engine are integrated. The terminal's voice generation means generates voice data based on this data. During this generation process, tone and intonation adjustments are made to reflect the emotion information.

[0663] Step 6:

[0664] The terminal's output method plays the generated audio data through its speaker. The user can then confirm that the audio, which reflects their intentions and emotions, has been appropriately output.

[0665] Step 7:

[0666] Depending on the situation, users can review system feedback and make adjustments to improve the accuracy of future analyses.

[0667] (Example 2)

[0668] 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".

[0669] Traditionally, there has been a lack of effective voice-based communication methods for users who have difficulty with voice communication. Furthermore, accurately expressing users' intentions and emotions has been challenging, leading to a decline in the quality of communication.

[0670] 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.

[0671] In this invention, the server includes imaging means for detecting the user's mouth movements, analysis means for comparing the mouth movements obtained by the imaging means with phoneme data, and emotion analysis means for analyzing the user's emotional state in order to adjust the tone and intonation of the generated speech data. This enables the generation of natural and emotionally rich speech using the user's mouth movements and emotional state, making more effective communication possible.

[0672] The "imaging means" refers to hardware that detects the movement of the user's mouth and acquires it as video data.

[0673] "Analysis means" refers to software or hardware that has the function of analyzing the movement of the mouth obtained by the imaging means and comparing it with phoneme data for identification.

[0674] "Speech generation means" refers to a process or apparatus for converting phoneme data identified by analysis means into speech data.

[0675] "Emotion analysis means" refers to a part of a system that has the function of analyzing facial expressions from the user's entire face and estimating their emotional state.

[0676] "Audio data" refers to data in digital or analog format that has been converted into an acoustic signal.

[0677] "Output means" refers to a device or process for transmitting the generated audio data to an external source.

[0678] The system according to the present invention provides a technology that generates natural speech using mouth movements and emotional states for users who have difficulty with voice communication. This system mainly includes "imaging means," "analysis means," "speech generation means," "emotion analysis means," and "output means."

[0679] The device is equipped with an imaging device that captures the user's mouth in real time and acquires video data. A high-resolution camera is used for the imaging device, making it possible to accurately capture the user's desire to speak. In addition, the device captures the user's entire face for emotion analysis and provides that data to the emotion analysis device.

[0680] The terminal's analysis method identifies phonemes by comparing the video data of the mouth obtained by the imaging method with pre-registered reference data. This analysis involves a machine learning model, enabling high-precision phoneme identification.

[0681] The emotion analysis method uses an emotion database to estimate the user's current emotional state from their facial expressions. For example, if the user is smiling, it recognizes the emotion of "happiness."

[0682] The terminal generates audio data using a speech generation method based on phonemes obtained by an analysis method. Here, the tone and intonation of the generated audio are adjusted based on data from the emotion analysis method. This results in the generation of natural-sounding audio that reflects the user's emotions.

[0683] Ultimately, the terminal's output method provides the generated audio data through an external device such as a speaker. This allows the user to faithfully convey their intentions and emotions.

[0684] For example, if a user moves their mouth to express gratitude and say "thank you," the system analyzes that movement and emotion, and generates a voice message saying "thank you" in a grateful tone.

[0685] An example of a prompt message could be: "Recognize the user's mouth movements, analyze their emotions, and generate the following emotionally expressive voice: (Specific mouth movements of the input, phonemes based on reference video data), emotional state (gratitude)."

[0686] In this way, this system realizes communication support that integrates voice and emotion, providing technology that helps users lead richer lives.

[0687] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0688] Step 1:

[0689] The terminal uses an imaging device to capture the user's mouth in real time. The input is the movement of the user's mouth, and the output is high-resolution video data. The imaging device captures the user's urge to speak and transmits this data to the next processing step.

[0690] Step 2:

[0691] The device analyzes the user's entire face using emotion analysis means and estimates their current emotional state. The input is video data of the face captured by the imaging means, and the output is emotion data based on the user's facial expressions. The emotion analysis means compares this data with an emotion database to estimate the user's emotional state.

[0692] Step 3:

[0693] The terminal's analysis method takes in the video data of the mouth obtained in step 1 and uses a machine learning model to identify phonemes. The input is the video data of the mouth, and the output is the identified phoneme data. This analysis procedure performs accurate phoneme recognition by comparing it with existing reference data.

[0694] Step 4:

[0695] The terminal's voice generation means generates voice data using phoneme data obtained from the analysis means. The input is the analyzed phoneme data and emotion data from the emotion analysis means, and the output is synthesized voice data. Based on the data from the emotion analysis means, the tone and intonation of the voice are adjusted to generate natural-sounding voice that reflects the user's emotions.

[0696] Step 5:

[0697] The terminal provides audio data generated by the voice generation means to the outside world through the output means. The input is synthesized audio data, and the output is audio played back through the speaker. This allows the user to communicate their intentions and emotions to those around them.

[0698] (Application Example 2)

[0699] 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".

[0700] In autonomous vehicles, there is a lack of means for the driver to give instructions to the vehicle without speaking. Furthermore, there is a need to achieve safer and more comfortable driving by implementing driving control that reflects the driver's emotional state.

[0701] 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.

[0702] In this invention, the server includes means for detecting mouth movements using a video device implemented in the communication vehicle to assist the user's speech, analysis means for comparing the mouth movements obtained by the video device with phoneme information, and adjustment means for reflecting the user's emotional state in the driving control system. This enables the driver to communicate their intentions to the vehicle without speaking and to control the vehicle in accordance with their emotions.

[0703] A "communication vehicle" is a vehicle equipped with digital devices that can communicate and process data in real time.

[0704] A "video device" is a device that has a shooting function to capture the user's movements and collect visual data.

[0705] "Phoneme information" refers to data that represents the basic units that make up speech.

[0706] "Analysis means" refers to technical methods used to identify specific information based on collected data.

[0707] "Audio information" refers to information that represents the characteristics and content of audio as digital data.

[0708] "Presentation means" refers to a device or mechanism for physically outputting generated audio information.

[0709] "Adjustment means" refers to means that have the function of appropriately changing or adjusting the operation of the system based on the analysis results.

[0710] The system for realizing this invention consists of multiple devices mounted on a communication vehicle. The server first captures the user's mouth movements using a video device. This video device uses an in-vehicle camera, and hardware for analyzing the data in real time, specifically NVIDIA's Jetson, is used.

[0711] The data acquired from the user's movements is processed by an analysis system. This analysis uses a Python program, OpenCV for image processing, and TensorFlow for phoneme analysis using a machine learning model. The analysis system generates speech information based on the obtained phoneme information, and an adjustment system adjusts the tone and intonation of the speech by incorporating sentiment analysis as needed. This sentiment analysis utilizes a sentiment model that has been pre-trained using deep learning.

[0712] The generated audio information is output by a presentation device and provided to the user and passengers through the car's speakers. For example, if the user moves their mouth to indicate "I want to turn right at the next intersection," the system analyzes that movement and generates an audio message saying "Turning right at the next intersection" which then plays from the speakers, and the vehicle takes control accordingly. Furthermore, if the user is smiling, the system reflects that nuance and outputs the audio in a brighter tone. The prompt in this case is instructed to the model as follows: "Explain how the car should respond when the user gives the next instruction with a smile."

[0713] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0714] Step 1:

[0715] The server uses video equipment to capture the user's mouth movements in real time. The input is video data from cameras installed inside the vehicle. The server acquires this data and converts it into an analyzable format.

[0716] Step 2:

[0717] The server analyzes mouth movements from video data acquired using the analysis method. The input is the video data obtained in step 1. Movement tracking is performed on this data using OpenCV, and phoneme analysis is performed using a TensorFlow machine learning model. The output is the classified phoneme data.

[0718] Step 3:

[0719] The server analyzes the user's facial expressions simultaneously to determine their emotional state. The input is the video data obtained in step 1. Using an emotion engine, a deep learning model is used to analyze facial expressions and derive the user's emotional data. The output of this step is the emotional state determined by the analysis.

[0720] Step 4:

[0721] The server converts the phoneme data based on the analysis into speech information, reflecting the emotional state in the tone and intonation of the speech. The input is the phoneme data from step 2 and the emotional state from step 3. A speech generation means is used to generate speech that includes natural emotional expression. The output of this step is the generated speech data.

[0722] Step 5:

[0723] The server outputs the generated audio data using a presentation device. The input is the audio data created in step 4. The audio is output using the in-car speakers and heard by the user and passengers. Specifically, it transmits vehicle operation commands (e.g., right turn instruction) based on the audio to the vehicle control system. The output consists of the audio presentation and the vehicle operation commands.

[0724] 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.

[0725] 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.

[0726] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[0727] 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.

[0728] 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. In the upper and lower directions of the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. Also, the upper side of the concentric circles is where "pleasant" emotions are located, and the lower side is where "unpleasant" emotions are located. In this way, 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.

[0729] 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.

[0730] 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.

[0731] 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.

[0732] 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."

[0733] 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.

[0734] 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.

[0735] 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.

[0736] 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.

[0737] 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.

[0738] 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.

[0739] 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.

[0740] 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.

[0741] 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.

[0742] 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.

[0743] 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.

[0744] 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.

[0745] The following is further disclosed regarding the embodiments described above.

[0746] (Claim 1)

[0747] To assist the user's speech, an imaging means for detecting mouth movements is provided,

[0748] An analysis means for comparing the mouth movements obtained by the imaging means with phoneme data,

[0749] A speech generation means that converts phoneme data identified by an analysis means into speech data,

[0750] An output means that outputs audio data generated by the audio generation means,

[0751] A system that includes this.

[0752] (Claim 2)

[0753] The system according to claim 1, wherein the imaging means captures the movements of the user's mouth in real time and transmits them to the analysis means.

[0754] (Claim 3)

[0755] The system according to claim 1, wherein the analysis means determines mouth movements using pre-registered reference data and a machine learning model.

[0756] "Example 1"

[0757] (Claim 1)

[0758] To assist the user's speech, an imaging means for detecting mouth movements is provided,

[0759] An analysis means that compares the mouth movements obtained by the imaging means with phoneme data and analyzes them using a machine learning model including a neural network,

[0760] A speech generation means that adjusts phoneme data identified by an analysis means to generate natural speech,

[0761] An output means that outputs audio data generated by the audio generation means,

[0762] A system that includes this.

[0763] (Claim 2)

[0764] The system according to claim 1, wherein the imaging means captures the movements of the user's mouth in real time and transmits them to the analysis means, including coordination with a server.

[0765] (Claim 3)

[0766] The system according to claim 1, wherein the analysis means utilizes computing resources on the cloud and uses pre-registered reference data and a generated AI model to determine the movement of the mouth.

[0767] "Application Example 1"

[0768] (Claim 1)

[0769] An imaging device that detects mouth movements,

[0770] An analysis device that compares mouth movements obtained by an imaging device with phonetic data,

[0771] A speech generation device that converts phonological data identified by an analysis device into acoustic data,

[0772] An output device that outputs acoustic data generated by a sound generation device,

[0773] A communication device that provides a dynamic customer service experience through a visual device and transmits audio data to other users via remote communication,

[0774] A system that includes this.

[0775] (Claim 2)

[0776] The system according to claim 1, wherein the imaging device captures the movements of the user's mouth in real time and transmits them to the analysis device.

[0777] (Claim 3)

[0778] The system according to claim 1, wherein the analysis device determines the movement of the mouth using pre-registered reference data and a generated AI model.

[0779] "Example 2 of combining an emotion engine"

[0780] (Claim 1)

[0781] To assist the user's speech, an imaging means for detecting mouth movements is provided,

[0782] An analysis means for comparing the mouth movements obtained by the imaging means with phoneme data,

[0783] A speech generation means that converts phoneme data identified by an analysis means into speech data,

[0784] An emotion analysis means for analyzing the user's emotional state in order to adjust the tone and intonation of the generated audio data,

[0785] An output means that outputs audio data generated by the audio generation means,

[0786] A system that includes this.

[0787] (Claim 2)

[0788] The system according to claim 1, wherein the imaging means captures the movements of the user's mouth in real time and transmits the video data, including the user's emotional state, to the analysis means.

[0789] (Claim 3)

[0790] The system according to claim 1, wherein the analysis means determines mouth movements using pre-registered reference data and a machine learning model, and the emotion analysis means estimates the user's emotional state using an emotion database.

[0791] "Application example 2 when combining with an emotional engine"

[0792] (Claim 1)

[0793] To assist the user's speech, a means for detecting mouth movements using a video device installed in a communication vehicle,

[0794] An analysis means for comparing mouth movements obtained by a video device with phoneme information,

[0795] A speech generation means that converts phoneme information identified by an analysis means into speech information,

[0796] A presentation means that outputs audio information generated by a speech generation means,

[0797] An adjustment mechanism that reflects the user's emotional state in the driving control system,

[0798] A system that includes this.

[0799] (Claim 2)

[0800] The system according to claim 1, wherein the video device instantly captures the movements of the user's mouth and transmits them to the analysis means.

[0801] (Claim 3)

[0802] The system according to claim 1, wherein the analysis means determines mouth movements using pre-registered reference data and a learning model. [Explanation of Symbols]

[0803] 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. To assist the user's speech, an imaging means for detecting mouth movements is provided, An analysis means for comparing the mouth movements obtained by the imaging means with phoneme data, A speech generation means that converts phoneme data identified by an analysis means into speech data, An output means that outputs audio data generated by the audio generation means, A system that includes this.

2. The system according to claim 1, wherein the imaging means captures the movements of the user's mouth in real time and transmits them to the analysis means.

3. The system according to claim 1, wherein the analysis means determines mouth movements using pre-registered reference data and a machine learning model.

Images