System

Abstract

To provide a system for emphasizing the voice of a specific speaker by removing ambient noise and unnecessary voice.SOLUTION: The system includes means for receiving speech data from a user's terminal, means for preprocessing the speech data, means for identifying speech of a conversational partner using a large language model for analyzing the preprocessed speech data, means for filtering ambient noise and unwanted speech from the identified speech data, and means for transmitting the filtered speech data to the user's terminal.SELECTED DRAWING: None

Description

[Technical Field]

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

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] In current online conferences and calls, background noise and the voices of other speakers can disrupt conversations. Furthermore, it can be difficult to clearly hear a specific speaker's voice, especially in noisy environments or for people with hearing impairments. In such cases, the quality of communication declines, hindering productivity and social participation. Therefore, there is a need for technology that can remove background noise and unwanted sounds while emphasizing the voice of a specific speaker. [Means for solving the problem]

[0005] The present invention solves the above problems by providing a system having the following configuration.

[0006] The system includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying a conversation partner's voice using a large-scale language model for analyzing the preprocessed voice data, means for filtering ambient noise and unwanted voices from the identified voice data, and means for transmitting the filtered voice data to the user terminal.

[0007] Furthermore, by including a means for acquiring audio data from a microphone, a means for transmitting the acquired audio data to a server, and a means for playing back the filtered audio data received from the server, users can enjoy clear audio in noisy environments and online conferences.

[0008] Also included is a means for the user to adjust specific settings and a means for adjusting the filtering of audio data based on said settings, allowing the user to adjust the accuracy of filtering to suit their needs.

[0009] "Voice data" is data that digitally represents the user's speech and surrounding sounds.

[0010] A "terminal" is a device used by a user, such as a computer, smartphone, or tablet.

[0011] "Preprocessing" refers to processes such as noise reduction and standardization of the sampling rate to make the audio data easier to analyze.

[0012] A "large-scale language model" is an artificial intelligence model that is trained based on large amounts of text data and performs advanced analysis of speech and text characteristics.

[0013] "Identification" is the process of using large-scale language models to identify and isolate the voice of a particular speaker within speech data.

[0014] "Filtering" is the process of removing ambient noise and unwanted sounds from the identified audio data.

[0015] A "server" is a central computer system that operates large-scale language models and processes speech data.

[0016] "Transmission" refers to transferring data from a terminal to a server or from a server to a terminal via a network.

[0017] "Playback" refers to the process of outputting audio data received by the terminal from a speaker or headphones in a form that can be heard by the user.

[0018] "Settings" are various options and parameters that a user adjusts within an application, such as the accuracy of voice filtering or the emphasis on specific voices.

[0019] A "microphone" is a device that converts sound waves into electrical signals and is used to capture audio data. [Brief explanation of the drawings]

[0020] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8]FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0021] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0022] First, the terms used in the following description will be explained.

[0023] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0024] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0025] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0026] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0027] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0028] [First embodiment]

[0029] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0030] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0031] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0032] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0033] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0034] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0035] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0036] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0037] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0038] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0039] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0040] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0041] The present invention is a system for realizing clear voice communication in online conferences and in noisy environments. This system operates through cooperation between a server, a terminal, and a user.

[0042] Server Roles

[0043] The server plays a central role in receiving voice data sent from the user's device and analyzing it using a large-scale language model (LLM). The server has the following functions:

[0044] 1. Receiving audio data

[0045] The server receives the voice data from the user's terminal and temporarily stores the received voice data in the server's memory.

[0046] 2. Preprocessing of audio data

[0047] The server pre-processes the received audio data, which includes applying noise reduction filters and normalizing the sampling rate.

[0048] 3. Analysis using large-scale language models

[0049] The server analyzes the pre-processed speech data using a large-scale language model, which identifies the speech of the interlocutor and distinguishes it from background noise and unwanted speech.

[0050] 4. Audio Filtering

[0051] The server filters the identified audio data to emphasize important audio components, and the filtered audio data is re-encoded and converted into a format that can be transmitted efficiently.

[0052] 5. Sending filtered audio data

[0053] The server sends the filtered audio data to the user's device, returning it in real time so the user receives clear audio.

[0054] Device Role

[0055] The device acquires audio data, sends it to the server, receives filtered audio from the server, and plays it back. Specific functions include:

[0056] 1. Acquiring audio data

[0057] The device acquires audio data from a built-in or external microphone, capturing the user's voice and environmental sounds via a microphone driver.

[0058] 2. Sending audio data

[0059] The acquired voice data is sent from the device to the server, which divides the voice data into packets and sends them to the server using a network protocol.

[0060] 3. Receiving filtered audio

[0061] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[0062] 4. Playing back received audio

[0063] The reconstructed audio data is played back by the device through speakers or headphones, and the volume and quality of the audio can be adjusted to provide the user with clear audio.

[0064] User operations

[0065] Users interact with the system and adjust settings to take advantage of noise cancellation and specific voice enhancement, including:

[0066] 1. Booting the system

[0067] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[0068] 2. Adjust your settings

[0069] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[0070] 3. Start a conversation

[0071] Users can start an online meeting or call and enjoy the benefits of Noise Cancellation 2.0 in real time while continuing the conversation.

[0072] Specific examples

[0073] For example, if a user is holding a remote meeting at a coffee shop, the following happens:

[0074] 1. The user launches the application and enables noise cancellation.

[0075] 2. The device captures the noise in the cafe and the user's voice and sends them to the server.

[0076] 3. The server receives the voice data and uses LLM to identify the user's voice and remove noise.

[0077] 4. The server sends the filtered audio to the device.

[0078] 5. The device plays the received clear audio to the user in real time.

[0079] This significantly reduces noise in the cafe, allowing conversations to proceed smoothly, enabling users to hold meetings more efficiently.In this way, the system proposed by this invention improves the quality of communication in online meetings and noisy environments.

[0080] The processing flow will be explained below.

[0081] Server Processing Steps

[0082] Step 1:

[0083] The server receives the audio stream from the user's terminal, and the received audio data is temporarily stored in the server's memory.

[0084] Step 2:

[0085] The server pre-processes the received audio data, which may include applying noise reduction filters and normalizing the sampling rate.

[0086] Step 3:

[0087] The server inputs the preprocessed speech data into a large-scale language model (LLM), which analyzes the speech data and extracts features to identify the speech of the conversation partner.

[0088] Step 4:

[0089] The server filters the identified audio data, a process that emphasizes important audio components and reduces ambient noise and unwanted voices.

[0090] Step 5:

[0091] The server encodes the filtered audio data into a format that can be transmitted efficiently.

[0092] Step 6:

[0093] The server transmits the encoded and filtered audio data to the user's terminal in real time via the Internet.

[0094] Terminal processing steps

[0095] Step 1:

[0096] The device acquires audio data from a built-in or external microphone. The device's microphone driver captures the user's voice and environmental sounds.

[0097] Step 2:

[0098] The terminal divides the acquired voice data into packets and transmits them to the server using a network protocol.

[0099] Step 3:

[0100] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[0101] Step 4:

[0102] The device plays the reconstructed audio data through speakers or headphones, adjusting the volume and quality of the audio to provide the user with clear audio.

[0103] User operation steps

[0104] Step 1:

[0105] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[0106] Step 2:

[0107] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[0108] Step 3:

[0109] The user starts an online meeting or call and continues the conversation while enjoying the real-time benefits of Noise Cancellation 2.0.

[0110] Specific examples

[0111] Step 1:

[0112] The user starts an application to hold a remote conference at a cafe and enables noise canceling.

[0113] Step 2:

[0114] The device captures the noise of the cafe and the user's voice and sends it to the server.

[0115] Step 3:

[0116] The server receives the audio data, identifies the user's voice with a large language model, and removes noise.

[0117] Step 4:

[0118] The server transmits the filtered audio to the terminal.

[0119] Step 5:

[0120] The terminal then plays the received clear audio to the user in real time.

[0121] This significantly reduces noise in the cafe, allowing conversations to proceed smoothly, allowing users to hold meetings efficiently.

[0122] Example 1

[0123] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0124] In conventional online meetings and noisy environments, voice communication is often hindered by ambient noise and unwanted sounds. This makes it difficult to achieve clear, crisp voice communication. In particular, in remote meetings and conference calls, if the speaker's voice cannot be clearly heard, the quality of communication declines and work efficiency also deteriorates. The present invention aims to solve these problems and provide a system that effectively clears voices even in noisy environments.

[0125] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0126] In this invention, the server includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying a conversation partner's voice using a generative AI model for analyzing the preprocessed voice data, means for filtering ambient noise and unwanted voices from the identified voice data, and means for transmitting the filtered voice data to the user terminal, thereby enabling the provision of clear, noise-free voice in real time.

[0127] "Audio data" refers to data that represents the user's voice and environmental sounds in digital form.

[0128] "Terminal" refers to a device that acquires voice data and communicates with the server. Examples include PCs, smartphones, and tablets.

[0129] "Server" is a central system for receiving, pre-processing, analyzing, filtering, and transmitting audio data.

[0130] "Preprocessing" refers to processing for applying a noise reduction filter to the received audio data and standardizing the sampling rate.

[0131] A "generative AI model" is a type of artificial intelligence model that uses a large-scale language model to analyze voice data, identify the voice of the conversation partner, and distinguish between unwanted voices and noise.

[0132] "Filtering" refers to processing for removing ambient noise and unwanted sounds from identified audio data and for emphasizing important audio components.

[0133] "Playback" refers to the process of making the filtered audio data audible to the user through the device's speakers or headphones.

[0134] "Settings" are options that allow users to customize the system's behavior, such as noise cancellation or emphasis on specific sounds.

[0135] The present invention is a system for realizing clear voice communication in online conferences and in noisy environments. This system operates through cooperation between a server, a terminal, and a user.

[0136] The server plays a central role in receiving voice data sent from the user's device and analyzing it using a large-scale language model (generative AI model). The server has the following functions:

[0137] 1. Receiving audio data

[0138] The server receives the voice data from the user's device. The received voice data is temporarily stored in the server's memory. For example, the data can be received using open source voice communication software.

[0139] 2. Preprocessing of audio data

[0140] The server preprocesses the received audio data, applying noise reduction filters and standardizing the sampling rate. This process uses tools such as LibROSA, a Python audio processing library, to reduce noise.

[0141] 3. Analysis using large-scale language models

[0142] The server then analyzes the preprocessed audio data using a generative AI model that identifies the voice of the conversation partner and distinguishes it from background noise and unwanted speech. At this stage, the LLM can be, for example, GPT-4 (registered trademark).

[0143] 4. Audio Filtering

[0144] The server filters the identified audio data to emphasize important audio components, and then re-encodes the filtered audio data into a format that can be transmitted efficiently. Specifically, the audio data can be compressed using the OPUS codec.

[0145] 5. Sending filtered audio data

[0146] The server sends the filtered audio data to the user's device, returning it in real time so the user receives clear audio.

[0147] The device's role is to acquire audio data, send it to the server, receive filtered audio from the server, and play it back. Specific functions include:

[0148] 1. Acquiring audio data

[0149] The device acquires audio data from a built-in or external microphone, for example, a USB microphone or a built-in microphone to capture the user's voice and environmental sounds.

[0150] 2. Sending audio data

[0151] The acquired voice data is sent from the terminal to the server, where it is divided into packets and sent to the server using the TCP / IP protocol.

[0152] 3. Receiving filtered audio

[0153] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[0154] 4. Playing back received audio

[0155] The reconstructed audio data is played back by the device through speakers or headphones, specifically via the device's audio driver.

[0156] Users can interact with the system and adjust settings to take advantage of noise cancellation and specific voice enhancement, including:

[0157] 1. Booting the system

[0158] The user launches the noise cancelling application on their device and clicks on the application icon to open the software.

[0159] 2. Adjust your settings

[0160] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[0161] 3. Start a conversation

[0162] Users can start an online meeting or call and enjoy the noise-canceling effect in real time while continuing the conversation.

[0163] Specific examples

[0164] For example, if users are having a remote meeting at a coffee shop, the process would be as follows:

[0165] 1. The user launches the application and enables noise cancellation.

[0166] 2. The device captures the noise of the cafe and the user's voice from the microphone and sends them to the server.

[0167] 3. The server receives the audio data and applies a noise reduction filter as preprocessing.

[0168] 4. The server uses a generative AI model (e.g., GPT-4) to identify the user's voice and remove noise.

[0169] 5. The server sends the filtered audio to the device.

[0170] 6. The device reconstructs the received audio data and plays it over speakers or headphones.

[0171] This significantly reduces the noise in the cafe, allowing for clearer audio during remote meetings. An example prompt is, "Design a system to efficiently cancel noise and make the user's voice clear in a situation where a remote meeting is being held in a cafe. Explain how the user's device captures audio data and communicates it with the server. Also, explain how the server processes the audio data and sends the filtered data back to the device."

[0172] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0173] Step 1: Acquire audio data (device)

[0174] The device uses a built-in or external microphone to capture voice data. This process captures the user's voice and environmental sounds. For example, when a user speaks in a cafe, the device's microphone picks up the user's voice and background noise at the same time. The input is an analog voice signal, and the output is digital voice data.

[0175] Step 2: Sending audio data (device)

[0176] The device sends the captured audio data to the server via the network. During this process, the audio data is divided into packets and transmitted using the TCP / IP protocol. The specific operations include packetizing the data and transmitting it over the network. The input is the digital audio data, and the output is the transmitted data packets.

[0177] Step 3: Receiving audio data (server)

[0178] The server receives voice data sent from the terminal via the network. The received data is temporarily stored in the server's memory. Specific operations include receiving data via the network interface and storing it in memory. The input is the transmitted data packet, and the output is the digital voice data stored in memory.

[0179] Step 4: Preprocessing the audio data (server)

[0180] The server applies a noise reduction filter to the received audio data and standardizes the sampling rate. Specifically, noise reduction is performed using LibROSA, a Python audio processing library. The input is digital audio data stored in memory, and the output is audio data with reduced noise and a standardized sampling rate.

[0181] Step 5: Analysis using large-scale language models (server)

[0182] The server analyzes the preprocessed audio data using a generative AI model. LLM distinguishes between the voice of the conversation partner and background noise from the audio data. In this process, GPT-4 can be used as the model. The input is the preprocessed audio data, and the output is the identified audio components.

[0183] Step 6: Audio Filtering (Server)

[0184] The server uses LLM to filter the identified audio data and emphasize important audio components. This filtering process applies multiple filters to emphasize specific frequency bands. The input is the identified audio components, and the output is clear filtered audio data.

[0185] Step 7: Sending the filtered audio data (server)

[0186] The server sends the filtered audio data to the device. The audio data is efficiently compressed and transferred in real time. Specifically, the OPUS codec is used to compress the data and send it over the network. The input is the filtered audio data, and the output is compressed audio data packets.

[0187] Step 8: Receive filtered audio (terminal)

[0188] The terminal receives the filtered voice data sent from the server and reconstructs the received data packets into a single voice data packet. Specific operations include receiving packets from the network and reconstructing them. The input is the compressed voice data packet, and the output is the reconstructed digital voice data.

[0189] Step 9: Playing back received audio (device)

[0190] The reconstructed audio data is then played back through the device's audio driver on speakers or headphones, with the volume and sound quality adjusted appropriately to provide clear audio to the user. The input is the reconstructed digital audio data, and the output is audio that the user can hear.

[0191] Step 10: User configuration operations (User)

[0192] A user uses a noise-canceling application to adjust settings, including whether to emphasize specific audio components and adjusting the level of noise cancellation. The input is the user's actions and settings, and the output is the customized system settings.

[0193] (Application example 1)

[0194] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0195] With conventional factory robots, voice communication between workers and robots is often hindered by loud noise in the factory, posing problems that affect work efficiency and safety. Furthermore, the lack of appropriate noise filtering methods increases the risk of workers misunderstanding instructions and causing operational errors. Therefore, there is a need for a method that enables clear voice communication even in noisy environments, allowing robots to work safely and efficiently.

[0196] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0197] In this invention, the server includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying the voice of a conversation partner using a large-scale language model for analyzing the preprocessed voice data, means for filtering ambient noise and unwanted voices from the identified voice data, means for transmitting the filtered voice data to the user terminal, and means for realizing voice communication between a worker and a robot in a noisy factory environment, thereby enabling clear voice communication between the worker and the robot even in a noisy factory environment.

[0198] "Voice data" means electronically recorded sound information, particularly data intended for communication by human voice.

[0199] A "user terminal" is a device for transmitting and receiving voice data, and is equipped with a microphone and a speaker.

[0200] "Preprocessing" refers to the initial data processing operations performed on received audio data, specifically noise reduction and standardization of the sampling rate.

[0201] A "large-scale language model" is a machine learning model that learns from large amounts of data for a specific task and analyzes or generates natural language.

[0202] "Interviewer voice identification" refers to the process of detecting and recognizing a specific person's voice from audio data.

[0203] "Noise" refers to background sounds and interfering sounds other than the target voice, and is a factor that reduces the quality of voice communication.

[0204] "Filtering" refers to the process of removing unnecessary components from audio data and extracting and emphasizing necessary parts.

[0205] "Noise environment within a factory" refers to a workplace environment where there is a lot of machinery and other work noise, and where noise levels are generally high.

[0206] "Workers" refers to personnel who operate machinery and equipment within a factory.

[0207] A "robot" is a programmable mechanical device capable of performing specific tasks automatically.

[0208] "Voice communication" refers to the transmission of information between people or machines via voice.

[0209] "Misoperation" refers to an inappropriate response by a robot to unintended operations or instructions, which has a negative impact on production efficiency and safety.

[0210] This invention is a system that realizes clear voice communication between a worker and a robot in a noisy factory environment. This system operates through cooperation between a server, a terminal, and a robot.

[0211] Server Roles

[0212] The server is responsible for the central processing of voice data. Specifically, it has the following functions:

[0213] 1. Receiving audio data

[0214] The server receives the voice data transmitted from the worker's terminal and temporarily stores the received voice data in the server's memory.

[0215] 2. Preprocessing of audio data

[0216] The server applies a noise reduction filter to the received audio data and normalizes the sampling rate, improving the quality of the data.

[0217] 3. Analysis using large-scale language models

[0218] The server uses the pre-processed speech data to analyze it with a large-scale language model to identify the worker's voice, and specific voice instructions are extracted from the analysis.

[0219] 4. Audio Filtering

[0220] The server filters out ambient noise from the identified audio data and emphasizes the important audio components. The filtered audio data is then re-encoded and converted into a format that can be transmitted efficiently.

[0221] 5. Sending filtered audio data

[0222] The server sends the filtered audio data to the robot, which then initiates actions based on it.

[0223] Device Role

[0224] The terminal is responsible for receiving and transmitting audio data. Specifically, it has the following functions:

[0225] 1. Acquiring audio data

[0226] The device acquires audio data from a built-in or external microphone, which is captured via a microphone driver.

[0227] 2. Sending audio data

[0228] The acquired voice data is sent from the device to the server, where it is divided into packets and transmitted using a network protocol.

[0229] 3. Receiving and playing filtered audio

[0230] The terminal receives the filtered voice data sent from the server and plays it through a speaker or headphones, allowing the worker to hear clear audio.

[0231] The role of robots

[0232] The robot operates based on filtered voice data from the server, and has the following functions:

[0233] 1. Analysis of audio data

[0234] The robot analyzes the filtered voice data sent from the server and begins to act as instructed.

[0235] 2. Real-time response

[0236] The robot responds to the worker's instructions in real time and carries out tasks efficiently.

[0237] Specific examples

[0238] For example, when a worker in a factory commands a robot to "start the next task," the flow is as follows:

[0239] 1. The worker captures the voice with the microphone on the terminal and sends it to the server.

[0240] 2. The server analyzes the received voice data and identifies the specified instructions.

[0241] 3. The server sends the filtered audio data to the robot.

[0242] 4. The robot follows the instructions and begins the next task.

[0243] Prompt Sentence Examples

[0244] "Design a voice instruction analysis system for a noisy factory environment. The system must filter out the noise in the factory and clearly convey the operator's instructions to the robot."

[0245] In this way, the system of the present invention enables clear voice communication between workers and robots even in noisy factory environments.

[0246] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0247] Step 1:

[0248] Acquiring audio data

[0249] The terminal uses a built-in or external microphone to acquire the worker's voice data. The input is a voice signal, and the output is digital voice data. This data is captured via a microphone driver.

[0250] Step 2:

[0251] Sending audio data

[0252] The device captures audio data and sends it to the server. The input is digital audio data, and the output is packet data sent to the server. The audio data is divided into packets and sent to the server using a network protocol.

[0253] Step 3:

[0254] Receiving audio data

[0255] The server receives the voice data sent from the terminal. The input is packet data, and the output is digital voice data temporarily stored in memory.

[0256] Step 4:

[0257] Audio data preprocessing

[0258] The server applies a noise reduction filter to the received audio data and normalizes the sampling rate. The input is digital audio data and the output is pre-processed digital audio data. Software filtering techniques are used for noise reduction and the sampling rate is changed to a normalized value.

[0259] Step 5:

[0260] Analysis using large-scale language models

[0261] The server analyzes the preprocessed voice data using a large-scale language model. The input is the preprocessed digital voice data, and the output is the identified voice instruction data. The large-scale language model analyzes the voice characteristics and extracts specific voice instructions.

[0262] Step 6:

[0263] Audio Filtering

[0264] The server removes ambient noise from the identified voice data and emphasizes the important voice components. The input is the identified voice instruction data, and the output is the filtered digital voice data. The filtering uses automatic noise reduction technology to emphasize only the necessary parts.

[0265] Step 7:

[0266] Sending filtered audio data

[0267] The server sends the filtered audio data to the robot. The input is filtered digital audio data, and the output is packets of data sent to the robot. This data is encoded and converted into a format that can be transmitted efficiently.

[0268] Step 8:

[0269] Analyzes voice data and starts operation

[0270] The robot analyzes the filtered voice data sent from the server and starts to act based on the instructions. The input is filtered digital voice data, and the output is the robot's specific action. The robot processes the received voice data and performs the appropriate action.

[0271] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0272] This invention is a system that realizes clear voice communication in online conferences and noisy environments, and also recognizes and provides feedback on the user's emotions. In addition to cooperation between the server, terminals, and users, this system also includes an emotion engine, providing more advanced functionality.

[0273] Server Roles

[0274] The server plays a central role in receiving voice data sent from the user's device and analyzing it using a large-scale language model (LLM). It also uses an emotion engine to recognize the user's emotional state and reflect it in voice filtering. Specific functions include:

[0275] 1. Receiving audio data

[0276] The server receives the voice data from the user's terminal and temporarily stores the received voice data in the server's memory.

[0277] 2. Preprocessing of audio data

[0278] The server pre-processes the received audio data, which includes applying noise reduction filters and normalizing the sampling rate.

[0279] 3. Analysis using large-scale language models

[0280] The server then analyzes the preprocessed speech data using a large-scale language model, which extracts features to identify the speech of the conversation partner and distinguish it from background noise and unwanted speech.

[0281] 4. Emotion Recognition by Emotion Engine

[0282] The server uses an emotion engine to recognize the user's emotion from the voice data. The emotion engine analyzes voice characteristics such as tone, pitch, and rate to identify the emotional state.

[0283] 5. Audio filtering and conditioning

[0284] The server filters the identified audio data, emphasizes important audio components, and adjusts the filtering and volume based on the emotion engine to provide optimal audio feedback according to the user's emotional state.

[0285] 6. Sending filtered audio data

[0286] The server encodes the filtered audio data and converts it into a format that can be transmitted efficiently, then transmits it in real time over the Internet to the user's device.

[0287] Device Role

[0288] The device is responsible for capturing audio data, sending it to the server, and receiving and playing back the filtered audio from the server. Specific functions include:

[0289] 1. Acquiring audio data

[0290] The device acquires audio data from a built-in or external microphone. The device's microphone driver captures the user's voice and environmental sounds.

[0291] 2. Sending audio data

[0292] The acquired voice data is sent from the device to the server, which divides the voice data into packets and sends them to the server using a network protocol.

[0293] 3. Receiving filtered audio

[0294] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[0295] 4. Playing back received audio

[0296] The reconstructed audio data is played back by the device through speakers or headphones, and the device adjusts the volume and quality of the audio to provide clear audio to the user.

[0297] 5. Displaying Emotional Feedback

[0298] The device receives the output of the emotion engine sent from the server and provides emotion feedback to the user, allowing the user to check their own emotional state in real time.

[0299] User operations

[0300] Users interact with the system and adjust settings to enable noise cancellation, emphasis on specific voices, and emotional feedback, including:

[0301] 1. Booting the system

[0302] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[0303] 2. Adjust your settings

[0304] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[0305] 3. Start a conversation

[0306] The user starts an online meeting or call and, during the meeting, proceeds with the conversation while receiving real-time emotional feedback and the effects of Noise Cancellation 2.0.

[0307] Specific examples

[0308] For example, if a user is holding a remote meeting at a coffee shop, the following happens:

[0309] 1. The user launches the application and enables noise cancellation.

[0310] 2. The device captures the noise in the cafe and the user's voice and sends them to the server.

[0311] 3. The server receives the voice data and uses LLM to identify the user's voice and remove noise.

[0312] 4. The server uses an emotion engine to recognize the user's emotions and reflect them in filtering.

[0313] 5. The server sends the filtered audio and emotional feedback to the device.

[0314] 6. The device receives clear audio and provides emotional feedback to the user in real time.

[0315] This not only significantly reduces noise in the cafe and allows conversations to proceed smoothly, but also enables optimal voice communication that is tailored to the user's emotional state.

[0316] The processing flow will be explained below.

[0317] Server Processing Steps

[0318] Step 1:

[0319] The server receives the audio stream from the user's terminal, and the received audio data is temporarily stored in the server's memory.

[0320] Step 2:

[0321] The server pre-processes the received audio data, which may include applying noise reduction filters and normalizing the sampling rate.

[0322] Step 3:

[0323] The server inputs the preprocessed speech data into a large-scale language model (LLM), which analyzes the speech data and extracts features to identify the speech of the conversation partner.

[0324] Step 4:

[0325] The server uses the LLM analysis results to identify and filter out ambient noise and unwanted sounds, highlighting important voice components.

[0326] Step 5:

[0327] The server inputs the filtered voice data into an emotion engine, which analyzes voice characteristics such as tone, pitch, and rate to identify the user's emotional state.

[0328] Step 6:

[0329] The server fine-tunes the voice filtering settings based on the emotional information obtained from the emotion engine. For example, if the user is feeling stressed, the server further reduces ambient noise.

[0330] Step 7:

[0331] The server encodes the filtered audio data and the output of the emotion engine into a format that can be transmitted efficiently.

[0332] Step 8:

[0333] The server transmits the encoded filtered voice data and emotional feedback to the user's device in real time via the Internet.

[0334] Terminal processing steps

[0335] Step 1:

[0336] The device acquires audio data from a built-in or external microphone. The device's microphone driver captures the user's voice and environmental sounds.

[0337] Step 2:

[0338] The terminal divides the acquired voice data into packets and transmits them to the server using a network protocol.

[0339] Step 3:

[0340] The device receives the filtered voice data and emotion feedback sent from the server, and reconstructs the received packets into a single voice data.

[0341] Step 4:

[0342] The device plays the reconstructed audio data in real time over speakers or headphones, adjusting the volume and quality of the audio to provide the user with clear audio.

[0343] Step 5:

[0344] The device displays the received emotional feedback to the user, allowing the user to check their own emotional state in real time.

[0345] User operation steps

[0346] Step 1:

[0347] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[0348] Step 2:

[0349] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[0350] Step 3:

[0351] The user starts an online meeting or call and, during the meeting, proceeds with the conversation while receiving real-time emotional feedback and the effects of Noise Cancellation 2.0.

[0352] Specific examples

[0353] Step 1:

[0354] The user starts an application to hold a remote conference at a cafe and enables noise canceling.

[0355] Step 2:

[0356] The device captures the noise of the cafe and the user's voice and sends it to the server.

[0357] Step 3:

[0358] The server receives the voice data, identifies the user's voice using LLM, and removes noise.

[0359] Step 4:

[0360] The server runs the filtered voice data through an emotion engine to analyze the user's emotional state.

[0361] Step 5:

[0362] The server fine-tunes the voice filtering based on the results of the emotion engine and generates optimal voice data.

[0363] Step 6:

[0364] The server transmits the filtered voice data and the emotional feedback to the terminal.

[0365] Step 7:

[0366] The device provides the user with clear received audio and emotional feedback in real time.

[0367] This not only significantly reduces noise in the cafe and allows conversations to proceed smoothly, but also enables optimal voice communication that is tailored to the user's emotional state.

[0368] Example 2

[0369] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0370] In conventional online conference and call systems, clear communication is difficult due to poor audio quality in noisy environments. Furthermore, the lack of a function to recognize and provide feedback on the user's emotional state in real time hinders smooth communication.

[0371] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0372] In this invention, the server includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying the voice of a conversation partner using a large-scale language model for analyzing the preprocessed voice data, means for using an emotion engine for recognizing emotions from the voice data, means for adjusting the voice data based on the emotional state, and means for transmitting the filtered and adjusted voice data to the user terminal, thereby enabling clear voice communication even in noisy environments and further enabling smooth communication by providing feedback on the user's emotional state in real time.

[0373] "Audio data" refers to audio information acquired from and transmitted by a user's terminal.

[0374] "Preprocessing" refers to processing to improve the quality of audio data, such as applying a noise reduction filter or standardizing the sampling rate.

[0375] A "large-scale language model" is an advanced machine learning model that analyzes text and audio data to understand the context and meaning of conversations.

[0376] An "emotion engine" is a system that analyzes voice characteristics such as tone, pitch, and speed to identify a speaker's emotional state.

[0377] "Filtering" refers to the process of removing unnecessary noise and interference sounds from audio data and emphasizing important audio components.

[0378] "Terminal" refers to a device used to capture, send, receive, and play audio data.

[0379] "Encoding" refers to the process of converting audio data into a format that can be transmitted efficiently.

[0380] "Real-time" refers to immediate processing and feedback.

[0381] "Emotional feedback" refers to informing users of their emotional state and providing information to improve the quality of communication.

[0382] "Settings" are adjustments that a user makes to adjust the system's behavior or options.

[0383] This invention is a system that realizes clear voice communication in online conferences and noisy environments, and also recognizes and provides feedback on the user's emotions. In addition to cooperation between the server, terminals, and users, this system also includes an emotion engine, providing more advanced functionality.

[0384] Server Roles

[0385] The server plays a central role in receiving, analyzing, and providing feedback on voice data sent from the user's device. Specific hardware used includes a high-performance CPU, memory, and network interface. Software includes a noise reduction filter (e.g., RNNoise), a large-scale language model (e.g., OpenAI® GPT-3®), and an emotion engine (e.g., IBM Watson® Tone Analyzer). These software and hardware components are used to perform the following processes:

[0386] 1. Receiving audio data

[0387] The server receives the voice data sent from the user's terminal and temporarily stores it in memory.

[0388] 2. Preprocessing of audio data

[0389] The server applies noise reduction filters to clean up the audio data and also standardizes the sampling rate to equalize the quality of the data.

[0390] 3. Analysis using large-scale language models

[0391] The server analyzes the preprocessed speech data using a large-scale language model to extract features to identify the speech of the conversation partner and distinguish it from ambient noise and unwanted speech.

[0392] 4. Emotion Recognition by Emotion Engine

[0393] The server uses an emotion engine to recognize the user's emotion from the voice data, which analyzes voice characteristics such as tone, pitch, and rate to identify the emotional state.

[0394] 5. Filtering and conditioning of audio data

[0395] The server emphasizes the identified important audio components and adjusts the volume and texture based on the output of the emotion engine.

[0396] 6. Encoding and transmitting filtered audio data

[0397] The server converts the filtered audio data into a format that can be transmitted efficiently and transmits it to the terminal via the network.

[0398] Device Role

[0399] The device is responsible for capturing audio data, sending it to the server, and receiving and playing back the filtered audio from the server. The hardware used includes an internal or external microphone, speakers or headphones, and a network interface. The software includes an audio capture driver and playback software. It performs the following tasks:

[0400] 1. Acquiring audio data

[0401] The device captures audio data using a microphone, and the captured data is temporarily stored in the device's memory.

[0402] 2. Sending audio data

[0403] The terminal transmits the acquired voice data to the server using a network protocol.

[0404] 3. Receiving filtered audio

[0405] The terminal receives the filtered audio data sent from the server and reconstructs it.

[0406] 4. Playing back received audio

[0407] The device then plays the reconstructed audio data over speakers or headphones, providing the user with clear audio.

[0408] 5. Displaying Emotional Feedback

[0409] The terminal receives the output of the emotion engine sent from the server and displays emotion feedback to the user in real time.

[0410] User operations

[0411] The user operates the system and configures the appropriate settings, including noise cancellation, voice emphasis, and whether or not emotional feedback is enabled. The user then launches the Noise Cancellation 2.0 application on the device, adjusts the settings, and starts an online meeting or call.

[0412] Specific examples

[0413] For example, if a user is holding a remote meeting at a coffee shop, the following happens:

[0414] 1. The user launches the Noise Cancelling 2.0 app and enables noise cancellation and voice enhancement.

[0415] 2. The device captures the noise of the cafe and the user's voice through its built-in microphone and sends the audio data to the server.

[0416] 3. The server receives the audio data and preprocesses it with a noise reduction filter.

[0417] 4. The server uses a large-scale language model to distinguish between the user's voice and noise.

[0418] 5. The server uses an emotion engine to identify the user's emotion from the voice data and perform filtering and adjustment.

[0419] 6. The server sends the filtered audio to the device.

[0420] 7. The device receives and plays the filtered audio data.

[0421] 8. The device displays emotional feedback to the user in real time.

[0422] Prompt Sentence Examples

[0423] Explain how Noise Cancellation 2.0 can be used to ensure clear voice communication and provide real-time feedback on user emotions during a remote meeting in a coffee shop.

[0424] As described above, the system of the present invention enables clear voice communication even in noisy environments, and furthermore, by recognizing and feeding back the user's emotions in real time, smooth communication can be realized.

[0425] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0426] Step 1:

[0427] Acquiring audio data

[0428] The device acquires audio data using a built-in or external microphone. The input is an audio signal from the microphone, and the output is digital audio data. The microphone driver starts up and captures the user's voice and environmental sounds in real time. This data is temporarily stored in the device's memory.

[0429] Step 2:

[0430] Sending audio data

[0431] The device sends the captured audio data to the server using a network protocol (e.g. TCP / IP). The input is digital audio data, which is converted into packets that are sent to the server as output. The audio data is divided into small packets and sent over the internet to the server.

[0432] Step 3:

[0433] Receiving voice data and storing it in memory

[0434] The server receives the voice data sent from the terminal and temporarily stores it in memory. The input is data packets, and the output is continuous digital voice data. The receiving module combines the data packets and reconstructs them into a single voice data.

[0435] Step 4:

[0436] Audio data preprocessing

[0437] The server applies a noise reduction filter (e.g., RNNoise) to clean up the audio data. It takes continuous digital audio data as input and produces denoised audio data as output. Noise is removed sample by sample and the sampling rate is normalized.

[0438] Step 5:

[0439] Analysis using large-scale language models

[0440] The server analyzes the preprocessed voice data using a large-scale language model (e.g., OpenAI GPT-3). The input is noise-removed voice data, and the output is the voice identity of the conversation partner. The language model performs contextual analysis and extracts features to distinguish the conversation partner's voice from noise and unwanted sounds.

[0441] Step 6:

[0442] Emotion recognition by emotion engine

[0443] The server uses an emotion engine (e.g., IBM Watson Tone Analyzer) to recognize the user's emotion from the voice data. Feature data is input, and emotional state information is obtained as output. The tone, pitch, and rate of the voice are analyzed to identify the emotional state.

[0444] Step 7:

[0445] Filtering and conditioning audio data

[0446] The server emphasizes the identified voice components and adjusts the volume and texture based on the output of the emotion engine. The inputs are emotional state information and voice identification information, and the output is adjusted voice data. The voice filter emphasizes important voice components and optimizes the volume and tone according to the emotion.

[0447] Step 8:

[0448] Encoding and transmitting filtered audio data

[0449] The server converts the filtered audio data into a format that can be transmitted efficiently (e.g., Opus codec) and sends it over the network to the device. The input is the conditioned audio data, and the output is the encoded audio data. The data is split into packets and transmitted again.

[0450] Step 9:

[0451] Receive and play filtered audio

[0452] The device receives the filtered audio data sent from the server and reconstructs it. The input is the encoded audio data, and the output is the reconstructed audio data. The reconstructed audio data is then played back through speakers or headphones, providing the user with clear audio.

[0453] Step 10:

[0454] Displaying Emotional Feedback

[0455] The device receives the output of the emotion engine sent from the server and displays emotional feedback to the user in real time. Emotional state information is input, and emotional feedback display is obtained as output. This allows the user to check their own emotional state in real time.

[0456] (Application example 2)

[0457] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0458] In recent years, with the increase in online meetings and remote work, the quality of voice communication in noisy environments has become an issue. Maintaining clear voice communication in noisy environments is also difficult during field work such as delivery work. Furthermore, there are currently limited methods for understanding delivery workers' emotions and stress levels in real time and providing appropriate feedback.

[0459] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving voice data from a user's information device, means for preprocessing the voice data, means for identifying the voice of the conversation partner using a large-scale language model for analyzing the preprocessed voice data, means for filtering environmental noise and unnecessary voice from the identified voice data, means for transmitting the filtered voice data to the user's information device, means for using an emotion engine for recognizing the user's emotion from the voice data, means for adjusting the voice data according to the user's emotional state, and means for displaying the adjusted voice data and emotion feedback on the user's information device. This not only enables clear voice communication even in noisy environments, but also makes it possible to grasp the user's emotional state in real time and provide appropriate feedback.

[0460] "Audio Data" means a digital recording of sound, typically captured using an audio input device such as a microphone.

[0461] "User's information equipment" refers to electronic devices used by users, such as smartphones, tablets, and laptops.

[0462] "Preprocessing" refers to performing processes such as noise reduction and standardization of the sampling rate on the acquired audio data.

[0463] A "large-scale language model" refers to a natural language processing model trained on a large amount of data, which enables the analysis and semantic understanding of speech data.

[0464] "Means for identifying the voice of a conversation partner" refers to a method or algorithm for extracting the voice of a specific speaker from multiple voice data.

[0465] "Environmental noise" refers to background noise or unwanted sounds that can be present during a conversation, such as traffic or wind noise.

[0466] "Filtering" refers to a process for emphasizing specific audio components and removing unwanted components.

[0467] "Emotion engine" refers to algorithms and technologies that analyze characteristics of voice data, such as tone, pitch, and speed, to identify a user's emotional state.

[0468] "Emotional feedback" refers to information that analyzes a user's emotional state and provides the results to the user in real time.

[0469] This invention is a system that improves the quality of voice communication in food delivery and simultaneously grasps the emotional state of delivery personnel in real time. The specific configuration and processing method of the system are described below.

[0470] Server Roles

[0471] The server plays a central role in receiving voice data from the user's information device (e.g., a smartphone) and recognizing emotions using an emotion engine. Specific processes include preprocessing the voice data, analyzing it using a large-scale language model, filtering environmental noise, and transmitting the voice data and emotional feedback.

[0472] 1. Receiving audio data:

[0473] The server receives the voice data sent from the delivery person's smartphone. At this stage, the voice data is received in packet format using a network protocol.

[0474] 2. Preprocessing the audio data:

[0475] The server processes the received audio data with a noise reduction filter and normalizes the sampling rate. For data processing, it uses an audio processing library (e.g., librosa).

[0476] 3. Large-scale Language Model (LLM) Analysis:

[0477] The preprocessed speech data is then analyzed using a large-scale language model, which extracts features from the speech data and identifies the speech of the conversation partner. The model uses a generative AI model (e.g., GPT-based).

[0478] 4. Emotion Recognition with Emotion Engine:

[0479] The server uses an emotion engine to recognize the delivery person's emotions from the voice data, which includes algorithms that analyze voice characteristics such as tone, pitch, and speed to identify emotional states.

[0480] 5. Audio filtering and conditioning:

[0481] The system removes environmental noise and unwanted sounds from the identified audio data to generate clear audio data, and also adjusts the audio data (e.g., volume) based on the results of emotion recognition.

[0482] 6. Sending filtered audio data and emotional feedback:

[0483] The server re-encodes the filtered audio data and transmits it to the user's information device in real time, and simultaneously transmits the emotional feedback and displays it on the information device.

[0484] Device Role

[0485] The terminal (i.e., the delivery person's smartphone) acquires, sends, and receives voice data, plays back filtered voice, and displays emotional feedback.

[0486] 1. Acquire audio data:

[0487] The device captures audio data from the built-in microphone and uses the device's microphone driver for audio input.

[0488] 2. Sending audio data:

[0489] The acquired voice data is divided into packets and sent to a server using a network protocol.

[0490] 3. Receiving and playing filtered audio:

[0491] It receives the filtered audio data sent from the server, reconstructs it, and plays it over speakers or headphones.

[0492] 4. Displaying Emotional Feedback:

[0493] Emotional feedback is displayed on the delivery person's smartphone, allowing them to check their own emotional state in real time.

[0494] User operations

[0495] The user (i.e., delivery person) activates the system, adjusts settings, and navigates the system through a process of capturing and verifying voice data.

[0496] 1. Boot the system:

[0497] The user launches the noise canceling application on the smartphone.

[0498] 2. Adjust the settings:

[0499] Users can adjust noise cancellation levels, enhance specific sounds, and other customization settings in the app's settings screen.

[0500] 3. Start a conversation:

[0501] Users can use the phone and voice guidance during delivery tasks, and the system provides real-time noise cancellation and emotional feedback during conversations, ensuring clear communication even in noisy environments.

[0502] Examples of specific examples and prompts

[0503] For example, when a delivery person is delivering a pizza, the system can provide clear voice communication even in a noisy traffic environment.The system also monitors the delivery person's emotional state in real time, and if the delivery person is under high stress, the support center will provide appropriate feedback.

[0504] Example prompt sentence:

[0505] Describe a smartphone app that maintains clear voice communication even when the user is in a noisy environment, and recognizes and provides feedback on the user's emotional state. The app performs voice data acquisition, noise reduction, emotion recognition, voice filtering, and emotional feedback display.

[0506] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0507] Step 1:

[0508] The user launches the application on their smartphone, which displays the initial application screen and allows them to adjust the noise cancellation level and other settings.

[0509] Step 2:

[0510] The device (smartphone) acquires voice data using a built-in microphone. Specifically, it captures the user's voice and surrounding environmental sounds (noise) and saves them as digital voice data. The acquired voice data is temporarily stored in the device's memory.

[0511] Step 3:

[0512] The terminal sends the acquired voice data to the server. The voice data is divided into packets and sent to the server using a network protocol (e.g., TCP / IP). The input is the raw voice data, and the output is the voice packets that arrive at the server.

[0513] Step 4:

[0514] The server receives the audio data sent from the device and preprocesses it by applying a noise reduction filter, which removes environmental noise and unwanted sounds in the audio signal. The input is raw audio data, and the output is preprocessed audio data with noise removed.

[0515] Step 5:

[0516] The server analyzes the preprocessed speech data using a large-scale language model (LLM) to identify the speech of the conversation partner. This analysis involves extracting and classifying speech features. The input is the preprocessed speech data, and the output is the identified speech segments of the conversation partner.

[0517] Step 6:

[0518] The server filters out environmental noise and unwanted sounds from the identified audio data by applying a filtering algorithm to highlight only the audio components of interest. This process results in clear audio data. The input is the identified audio data, and the output is the filtered, clear audio data.

[0519] Step 7:

[0520] The server uses an emotion engine to recognize the user's emotion from the voice data. It analyzes acoustic characteristics such as tone, pitch, and rate of the voice to identify the user's emotional state. The input is the filtered voice data, and the output is data indicating the user's emotional state.

[0521] Step 8:

[0522] The server adjusts the audio data according to the user's emotional state and generates appropriate audio feedback. For example, if the user is feeling stressed, it adjusts the volume or applies a specific filter. The input is the emotional state data and the filtered audio data, and the output is the adjusted audio data.

[0523] Step 9:

[0524] The server encodes the filtered and adjusted voice data and sends it to the user's device, and simultaneously sends emotional feedback to be displayed on the device. The input is the adjusted voice data and emotional feedback data, and the output is the data that arrives at the user's device.

[0525] Step 10:

[0526] The terminal reconstructs the received filtered and adjusted audio data and plays it over speakers or headphones. The input is the audio packets received from the server, and the output is the clear audio being played.

[0527] Step 11:

[0528] The device displays emotional feedback to the user, which is displayed on the smartphone screen as an icon or message indicating the user's emotional state. The input is the user's emotional feedback data, and the output is the displayed emotional feedback.

[0529] Through specific processing steps, users can maintain clear voice communication even in noisy environments and understand their own emotional state in real time. This system is very useful in food delivery situations.

[0530] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0531] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (registered trademark) (Internet search engine).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0532] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0533] [Second embodiment]

[0534] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0535] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0536] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0537] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0538] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0539] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0540] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0541] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0542] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0543] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0544] In the smart glasses 214, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0545] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0546] The present invention is a system for realizing clear voice communication in online conferences and in noisy environments. This system operates through cooperation between a server, a terminal, and a user.

[0547] Server Roles

[0548] The server plays a central role in receiving voice data sent from the user's device and analyzing it using a large-scale language model (LLM). The server has the following functions:

[0549] 1. Receiving audio data

[0550] The server receives the voice data from the user's terminal and temporarily stores the received voice data in the server's memory.

[0551] 2. Preprocessing of audio data

[0552] The server pre-processes the received audio data, which includes applying noise reduction filters and normalizing the sampling rate.

[0553] 3. Analysis using large-scale language models

[0554] The server analyzes the pre-processed speech data using a large-scale language model, which identifies the speech of the interlocutor and distinguishes it from background noise and unwanted speech.

[0555] 4. Audio Filtering

[0556] The server filters the identified audio data to emphasize important audio components, and the filtered audio data is re-encoded and converted into a format that can be transmitted efficiently.

[0557] 5. Sending filtered audio data

[0558] The server sends the filtered audio data to the user's device, returning it in real time so the user receives clear audio.

[0559] Device Role

[0560] The device acquires audio data, sends it to the server, receives filtered audio from the server, and plays it back. Specific functions include:

[0561] 1. Acquiring audio data

[0562] The device acquires audio data from a built-in or external microphone, capturing the user's voice and environmental sounds via a microphone driver.

[0563] 2. Sending audio data

[0564] The acquired voice data is sent from the device to the server, which divides the voice data into packets and sends them to the server using a network protocol.

[0565] 3. Receiving filtered audio

[0566] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[0567] 4. Playing back received audio

[0568] The reconstructed audio data is played back by the device through speakers or headphones, and the volume and quality of the audio can be adjusted to provide the user with clear audio.

[0569] User operations

[0570] Users interact with the system and adjust settings to take advantage of noise cancellation and specific voice enhancement, including:

[0571] 1. Booting the system

[0572] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[0573] 2. Adjust your settings

[0574] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[0575] 3. Start a conversation

[0576] Users can start an online meeting or call and enjoy the benefits of Noise Cancellation 2.0 in real time while continuing the conversation.

[0577] Specific examples

[0578] For example, if a user is holding a remote meeting at a coffee shop, the following happens:

[0579] 1. The user launches the application and enables noise cancellation.

[0580] 2. The device captures the noise in the cafe and the user's voice and sends them to the server.

[0581] 3. The server receives the voice data and uses LLM to identify the user's voice and remove noise.

[0582] 4. The server sends the filtered audio to the device.

[0583] 5. The device plays the received clear audio to the user in real time.

[0584] This significantly reduces noise in the cafe, allowing conversations to proceed smoothly, enabling users to hold meetings more efficiently.In this way, the system proposed by this invention improves the quality of communication in online meetings and noisy environments.

[0585] The processing flow will be explained below.

[0586] Server Processing Steps

[0587] Step 1:

[0588] The server receives the audio stream from the user's terminal, and the received audio data is temporarily stored in the server's memory.

[0589] Step 2:

[0590] The server pre-processes the received audio data, which may include applying noise reduction filters and normalizing the sampling rate.

[0591] Step 3:

[0592] The server inputs the preprocessed speech data into a large-scale language model (LLM), which analyzes the speech data and extracts features to identify the speech of the conversation partner.

[0593] Step 4:

[0594] The server filters the identified audio data, a process that emphasizes important audio components and reduces ambient noise and unwanted voices.

[0595] Step 5:

[0596] The server encodes the filtered audio data into a format that can be transmitted efficiently.

[0597] Step 6:

[0598] The server transmits the encoded and filtered audio data to the user's terminal in real time via the Internet.

[0599] Terminal processing steps

[0600] Step 1:

[0601] The device acquires audio data from a built-in or external microphone. The device's microphone driver captures the user's voice and environmental sounds.

[0602] Step 2:

[0603] The terminal divides the acquired voice data into packets and transmits them to the server using a network protocol.

[0604] Step 3:

[0605] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[0606] Step 4:

[0607] The device plays the reconstructed audio data through speakers or headphones, adjusting the volume and quality of the audio to provide the user with clear audio.

[0608] User operation steps

[0609] Step 1:

[0610] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[0611] Step 2:

[0612] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[0613] Step 3:

[0614] The user starts an online meeting or call and continues the conversation while enjoying the real-time benefits of Noise Cancellation 2.0.

[0615] Specific examples

[0616] Step 1:

[0617] The user starts an application to hold a remote conference at a cafe and enables noise canceling.

[0618] Step 2:

[0619] The device captures the noise of the cafe and the user's voice and sends it to the server.

[0620] Step 3:

[0621] The server receives the audio data, identifies the user's voice with a large language model, and removes noise.

[0622] Step 4:

[0623] The server transmits the filtered audio to the terminal.

[0624] Step 5:

[0625] The terminal then plays the received clear audio to the user in real time.

[0626] This significantly reduces noise in the cafe, allowing conversations to proceed smoothly, allowing users to hold meetings efficiently.

[0627] Example 1

[0628] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0629] In conventional online meetings and noisy environments, voice communication is often hindered by ambient noise and unwanted sounds. This makes it difficult to achieve clear, crisp voice communication. In particular, in remote meetings and conference calls, if the speaker's voice cannot be clearly heard, the quality of communication declines and work efficiency also deteriorates. The present invention aims to solve these problems and provide a system that effectively clears voices even in noisy environments.

[0630] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0631] In this invention, the server includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying a conversation partner's voice using a generative AI model for analyzing the preprocessed voice data, means for filtering ambient noise and unwanted voices from the identified voice data, and means for transmitting the filtered voice data to the user terminal, thereby enabling the provision of clear, noise-free voice in real time.

[0632] "Audio data" refers to data that represents the user's voice and environmental sounds in digital form.

[0633] "Terminal" refers to a device that acquires voice data and communicates with the server. Examples include PCs, smartphones, and tablets.

[0634] "Server" is a central system for receiving, pre-processing, analyzing, filtering, and transmitting audio data.

[0635] "Preprocessing" refers to processing for applying a noise reduction filter to the received audio data and standardizing the sampling rate.

[0636] A "generative AI model" is a type of artificial intelligence model that uses a large-scale language model to analyze voice data, identify the voice of the conversation partner, and distinguish between unwanted voices and noise.

[0637] "Filtering" refers to processing for removing ambient noise and unwanted sounds from identified audio data and for emphasizing important audio components.

[0638] "Playback" refers to the process of making the filtered audio data audible to the user through the device's speakers or headphones.

[0639] "Settings" are options that allow users to customize the system's behavior, such as noise cancellation or emphasis on specific sounds.

[0640] The present invention is a system for realizing clear voice communication in online conferences and in noisy environments. This system operates through cooperation between a server, a terminal, and a user.

[0641] The server plays a central role in receiving voice data sent from the user's device and analyzing it using a large-scale language model (generative AI model). The server has the following functions:

[0642] 1. Receiving audio data

[0643] The server receives the voice data from the user's device. The received voice data is temporarily stored in the server's memory. For example, the data can be received using open source voice communication software.

[0644] 2. Preprocessing of audio data

[0645] The server preprocesses the received audio data, applying noise reduction filters and standardizing the sampling rate. This process uses tools such as LibROSA, a Python audio processing library, to reduce noise.

[0646] 3. Analysis using large-scale language models

[0647] The server then analyzes the preprocessed audio data using a generative AI model, which identifies the voice of the person speaking and distinguishes it from background noise and unwanted sounds. At this stage, the LLM can be, for example, GPT-4.

[0648] 4. Audio Filtering

[0649] The server filters the identified audio data to emphasize important audio components, and then re-encodes the filtered audio data into a format that can be transmitted efficiently. Specifically, the audio data can be compressed using the OPUS codec.

[0650] 5. Sending filtered audio data

[0651] The server sends the filtered audio data to the user's device, returning it in real time so the user receives clear audio.

[0652] The device's role is to acquire audio data, send it to the server, receive filtered audio from the server, and play it back. Specific functions include:

[0653] 1. Acquiring audio data

[0654] The device acquires audio data from a built-in or external microphone, for example, a USB microphone or a built-in microphone to capture the user's voice and environmental sounds.

[0655] 2. Sending audio data

[0656] The acquired voice data is sent from the terminal to the server, where it is divided into packets and sent to the server using the TCP / IP protocol.

[0657] 3. Receiving filtered audio

[0658] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[0659] 4. Playing back received audio

[0660] The reconstructed audio data is played back by the device through speakers or headphones, specifically via the device's audio driver.

[0661] Users can interact with the system and adjust settings to take advantage of noise cancellation and specific voice enhancement, including:

[0662] 1. Booting the system

[0663] The user launches the noise cancelling application on their device and clicks on the application icon to open the software.

[0664] 2. Adjust your settings

[0665] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[0666] 3. Start a conversation

[0667] Users can start an online meeting or call and enjoy the noise-canceling effect in real time while continuing the conversation.

[0668] Specific examples

[0669] For example, if users are having a remote meeting at a coffee shop, the process would be as follows:

[0670] 1. The user launches the application and enables noise cancellation.

[0671] 2. The device captures the noise of the cafe and the user's voice from the microphone and sends them to the server.

[0672] 3. The server receives the audio data and applies a noise reduction filter as preprocessing.

[0673] 4. The server uses a generative AI model (e.g., GPT-4) to identify the user's voice and remove noise.

[0674] 5. The server sends the filtered audio to the device.

[0675] 6. The device reconstructs the received audio data and plays it over speakers or headphones.

[0676] This significantly reduces the noise in the cafe, allowing for clearer audio during remote meetings. An example prompt is, "Design a system to efficiently cancel noise and make the user's voice clear in a situation where a remote meeting is being held in a cafe. Explain how the user's device captures audio data and communicates it with the server. Also, explain how the server processes the audio data and sends the filtered data back to the device."

[0677] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0678] Step 1: Acquire audio data (device)

[0679] The device uses a built-in or external microphone to capture voice data. This process captures the user's voice and environmental sounds. For example, when a user speaks in a cafe, the device's microphone picks up the user's voice and background noise at the same time. The input is an analog voice signal, and the output is digital voice data.

[0680] Step 2: Sending audio data (device)

[0681] The device sends the captured audio data to the server via the network. During this process, the audio data is divided into packets and transmitted using the TCP / IP protocol. The specific operations include packetizing the data and transmitting it over the network. The input is the digital audio data, and the output is the transmitted data packets.

[0682] Step 3: Receiving audio data (server)

[0683] The server receives voice data sent from the terminal via the network. The received data is temporarily stored in the server's memory. Specific operations include receiving data via the network interface and storing it in memory. The input is the transmitted data packet, and the output is the digital voice data stored in memory.

[0684] Step 4: Preprocessing the audio data (server)

[0685] The server applies a noise reduction filter to the received audio data and standardizes the sampling rate. Specifically, noise reduction is performed using LibROSA, a Python audio processing library. The input is digital audio data stored in memory, and the output is audio data with reduced noise and a standardized sampling rate.

[0686] Step 5: Analysis using large-scale language models (server)

[0687] The server analyzes the preprocessed audio data using a generative AI model. LLM distinguishes between the voice of the conversation partner and background noise from the audio data. In this process, GPT-4 can be used as the model. The input is the preprocessed audio data, and the output is the identified audio components.

[0688] Step 6: Audio Filtering (Server)

[0689] The server uses LLM to filter the identified audio data and emphasize important audio components. This filtering process applies multiple filters to emphasize specific frequency bands. The input is the identified audio components, and the output is clear filtered audio data.

[0690] Step 7: Sending the filtered audio data (server)

[0691] The server sends the filtered audio data to the device. The audio data is efficiently compressed and transferred in real time. Specifically, the OPUS codec is used to compress the data and send it over the network. The input is the filtered audio data, and the output is compressed audio data packets.

[0692] Step 8: Receive filtered audio (terminal)

[0693] The terminal receives the filtered voice data sent from the server and reconstructs the received data packets into a single voice data packet. Specific operations include receiving packets from the network and reconstructing them. The input is the compressed voice data packet, and the output is the reconstructed digital voice data.

[0694] Step 9: Playing back received audio (device)

[0695] The reconstructed audio data is then played back through the device's audio driver on speakers or headphones, with the volume and sound quality adjusted appropriately to provide clear audio to the user. The input is the reconstructed digital audio data, and the output is audio that the user can hear.

[0696] Step 10: User configuration operations (User)

[0697] A user uses a noise-canceling application to adjust settings, including whether to emphasize specific audio components and adjusting the level of noise cancellation. The input is the user's actions and settings, and the output is the customized system settings.

[0698] (Application example 1)

[0699] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0700] With conventional factory robots, voice communication between workers and robots is often hindered by loud noise in the factory, posing problems that affect work efficiency and safety. Furthermore, the lack of appropriate noise filtering methods increases the risk of workers misunderstanding instructions and causing operational errors. Therefore, there is a need for a method that enables clear voice communication even in noisy environments, allowing robots to work safely and efficiently.

[0701] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0702] In this invention, the server includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying the voice of a conversation partner using a large-scale language model for analyzing the preprocessed voice data, means for filtering ambient noise and unwanted voices from the identified voice data, means for transmitting the filtered voice data to the user terminal, and means for realizing voice communication between a worker and a robot in a noisy factory environment, thereby enabling clear voice communication between the worker and the robot even in a noisy factory environment.

[0703] "Voice data" means electronically recorded sound information, particularly data intended for communication by human voice.

[0704] A "user terminal" is a device for transmitting and receiving voice data, and is equipped with a microphone and a speaker.

[0705] "Preprocessing" refers to the initial data processing operations performed on received audio data, specifically noise reduction and standardization of the sampling rate.

[0706] A "large-scale language model" is a machine learning model that learns from large amounts of data for a specific task and analyzes or generates natural language.

[0707] "Interviewer voice identification" refers to the process of detecting and recognizing a specific person's voice from audio data.

[0708] "Noise" refers to background sounds and interfering sounds other than the target voice, and is a factor that reduces the quality of voice communication.

[0709] "Filtering" refers to the process of removing unnecessary components from audio data and extracting and emphasizing necessary parts.

[0710] "Noise environment within a factory" refers to a workplace environment where there is a lot of machinery and other work noise, and where noise levels are generally high.

[0711] "Workers" refers to personnel who operate machinery and equipment within a factory.

[0712] A "robot" is a programmable mechanical device capable of performing specific tasks automatically.

[0713] "Voice communication" refers to the transmission of information between people or machines via voice.

[0714] "Misoperation" refers to an inappropriate response by a robot to unintended operations or instructions, which has a negative impact on production efficiency and safety.

[0715] This invention is a system that realizes clear voice communication between a worker and a robot in a noisy factory environment. This system operates through cooperation between a server, a terminal, and a robot.

[0716] Server Roles

[0717] The server is responsible for the central processing of voice data. Specifically, it has the following functions:

[0718] 1. Receiving audio data

[0719] The server receives the voice data transmitted from the worker's terminal and temporarily stores the received voice data in the server's memory.

[0720] 2. Preprocessing of audio data

[0721] The server applies a noise reduction filter to the received audio data and normalizes the sampling rate, improving the quality of the data.

[0722] 3. Analysis using large-scale language models

[0723] The server uses the pre-processed speech data to analyze it with a large-scale language model to identify the worker's voice, and specific voice instructions are extracted from the analysis.

[0724] 4. Audio Filtering

[0725] The server filters out ambient noise from the identified audio data and emphasizes the important audio components. The filtered audio data is then re-encoded and converted into a format that can be transmitted efficiently.

[0726] 5. Sending filtered audio data

[0727] The server sends the filtered audio data to the robot, which then initiates actions based on it.

[0728] Device Role

[0729] The terminal is responsible for receiving and transmitting audio data. Specifically, it has the following functions:

[0730] 1. Acquiring audio data

[0731] The device acquires audio data from a built-in or external microphone, which is captured via a microphone driver.

[0732] 2. Sending audio data

[0733] The acquired voice data is sent from the device to the server, where it is divided into packets and transmitted using a network protocol.

[0734] 3. Receiving and playing filtered audio

[0735] The terminal receives the filtered voice data sent from the server and plays it through a speaker or headphones, allowing the worker to hear clear audio.

[0736] The role of robots

[0737] The robot operates based on filtered voice data from the server, and has the following functions:

[0738] 1. Analysis of audio data

[0739] The robot analyzes the filtered voice data sent from the server and begins to act as instructed.

[0740] 2. Real-time response

[0741] The robot responds to the worker's instructions in real time and carries out tasks efficiently.

[0742] Specific examples

[0743] For example, when a worker in a factory commands a robot to "start the next task," the flow is as follows:

[0744] 1. The worker captures the voice with the microphone on the terminal and sends it to the server.

[0745] 2. The server analyzes the received voice data and identifies the specified instructions.

[0746] 3. The server sends the filtered audio data to the robot.

[0747] 4. The robot follows the instructions and begins the next task.

[0748] Prompt Sentence Examples

[0749] "Design a voice instruction analysis system for a noisy factory environment. The system must filter out the noise in the factory and clearly convey the operator's instructions to the robot."

[0750] In this way, the system of the present invention enables clear voice communication between workers and robots even in noisy factory environments.

[0751] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0752] Step 1:

[0753] Acquiring audio data

[0754] The terminal uses a built-in or external microphone to acquire the worker's voice data. The input is a voice signal, and the output is digital voice data. This data is captured via a microphone driver.

[0755] Step 2:

[0756] Sending audio data

[0757] The device captures audio data and sends it to the server. The input is digital audio data, and the output is packet data sent to the server. The audio data is divided into packets and sent to the server using a network protocol.

[0758] Step 3:

[0759] Receiving audio data

[0760] The server receives the voice data sent from the terminal. The input is packet data, and the output is digital voice data temporarily stored in memory.

[0761] Step 4:

[0762] Audio data preprocessing

[0763] The server applies a noise reduction filter to the received audio data and normalizes the sampling rate. The input is digital audio data and the output is pre-processed digital audio data. Software filtering techniques are used for noise reduction and the sampling rate is changed to a normalized value.

[0764] Step 5:

[0765] Analysis using large-scale language models

[0766] The server analyzes the preprocessed voice data using a large-scale language model. The input is the preprocessed digital voice data, and the output is the identified voice instruction data. The large-scale language model analyzes the voice characteristics and extracts specific voice instructions.

[0767] Step 6:

[0768] Audio Filtering

[0769] The server removes ambient noise from the identified voice data and emphasizes the important voice components. The input is the identified voice instruction data, and the output is the filtered digital voice data. The filtering uses automatic noise reduction technology to emphasize only the necessary parts.

[0770] Step 7:

[0771] Sending filtered audio data

[0772] The server sends the filtered audio data to the robot. The input is filtered digital audio data, and the output is packets of data sent to the robot. This data is encoded and converted into a format that can be transmitted efficiently.

[0773] Step 8:

[0774] Analyzes voice data and starts operation

[0775] The robot analyzes the filtered voice data sent from the server and starts to act based on the instructions. The input is filtered digital voice data, and the output is the robot's specific action. The robot processes the received voice data and performs the appropriate action.

[0776] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0777] This invention is a system that realizes clear voice communication in online conferences and noisy environments, and also recognizes and provides feedback on the user's emotions. In addition to cooperation between the server, terminals, and users, this system also includes an emotion engine, providing more advanced functionality.

[0778] Server Roles

[0779] The server plays a central role in receiving voice data sent from the user's device and analyzing it using a large-scale language model (LLM). It also uses an emotion engine to recognize the user's emotional state and reflect it in voice filtering. Specific functions include:

[0780] 1. Receiving audio data

[0781] The server receives the voice data from the user's terminal and temporarily stores the received voice data in the server's memory.

[0782] 2. Preprocessing of audio data

[0783] The server pre-processes the received audio data, which includes applying noise reduction filters and normalizing the sampling rate.

[0784] 3. Analysis using large-scale language models

[0785] The server then analyzes the preprocessed speech data using a large-scale language model, which extracts features to identify the speech of the conversation partner and distinguish it from background noise and unwanted speech.

[0786] 4. Emotion Recognition by Emotion Engine

[0787] The server uses an emotion engine to recognize the user's emotion from the voice data. The emotion engine analyzes voice characteristics such as tone, pitch, and rate to identify the emotional state.

[0788] 5. Audio filtering and conditioning

[0789] The server filters the identified audio data, emphasizes important audio components, and adjusts the filtering and volume based on the emotion engine to provide optimal audio feedback according to the user's emotional state.

[0790] 6. Sending filtered audio data

[0791] The server encodes the filtered audio data and converts it into a format that can be transmitted efficiently, then transmits it in real time over the Internet to the user's device.

[0792] Device Role

[0793] The device is responsible for capturing audio data, sending it to the server, and receiving and playing back the filtered audio from the server. Specific functions include:

[0794] 1. Acquiring audio data

[0795] The device acquires audio data from a built-in or external microphone. The device's microphone driver captures the user's voice and environmental sounds.

[0796] 2. Sending audio data

[0797] The acquired voice data is sent from the device to the server, which divides the voice data into packets and sends them to the server using a network protocol.

[0798] 3. Receiving filtered audio

[0799] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[0800] 4. Playing back received audio

[0801] The reconstructed audio data is played back by the device through speakers or headphones, and the device adjusts the volume and quality of the audio to provide clear audio to the user.

[0802] 5. Displaying Emotional Feedback

[0803] The device receives the output of the emotion engine sent from the server and provides emotion feedback to the user, allowing the user to check their own emotional state in real time.

[0804] User operations

[0805] Users interact with the system and adjust settings to enable noise cancellation, emphasis on specific voices, and emotional feedback, including:

[0806] 1. Booting the system

[0807] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[0808] 2. Adjust your settings

[0809] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[0810] 3. Start a conversation

[0811] The user starts an online meeting or call and, during the meeting, proceeds with the conversation while receiving real-time emotional feedback and the effects of Noise Cancellation 2.0.

[0812] Specific examples

[0813] For example, if a user is holding a remote meeting at a coffee shop, the following happens:

[0814] 1. The user launches the application and enables noise cancellation.

[0815] 2. The device captures the noise in the cafe and the user's voice and sends them to the server.

[0816] 3. The server receives the voice data and uses LLM to identify the user's voice and remove noise.

[0817] 4. The server uses an emotion engine to recognize the user's emotions and reflect them in filtering.

[0818] 5. The server sends the filtered audio and emotional feedback to the device.

[0819] 6. The device receives clear audio and provides emotional feedback to the user in real time.

[0820] This not only significantly reduces noise in the cafe and allows conversations to proceed smoothly, but also enables optimal voice communication that is tailored to the user's emotional state.

[0821] The processing flow will be explained below.

[0822] Server Processing Steps

[0823] Step 1:

[0824] The server receives the audio stream from the user's terminal, and the received audio data is temporarily stored in the server's memory.

[0825] Step 2:

[0826] The server pre-processes the received audio data, which may include applying noise reduction filters and normalizing the sampling rate.

[0827] Step 3:

[0828] The server inputs the preprocessed speech data into a large-scale language model (LLM), which analyzes the speech data and extracts features to identify the speech of the conversation partner.

[0829] Step 4:

[0830] The server uses the LLM analysis results to identify and filter out ambient noise and unwanted sounds, highlighting important voice components.

[0831] Step 5:

[0832] The server inputs the filtered voice data into an emotion engine, which analyzes voice characteristics such as tone, pitch, and rate to identify the user's emotional state.

[0833] Step 6:

[0834] The server fine-tunes the voice filtering settings based on the emotional information obtained from the emotion engine. For example, if the user is feeling stressed, the server further reduces ambient noise.

[0835] Step 7:

[0836] The server encodes the filtered audio data and the output of the emotion engine into a format that can be transmitted efficiently.

[0837] Step 8:

[0838] The server transmits the encoded filtered voice data and emotional feedback to the user's device in real time via the Internet.

[0839] Terminal processing steps

[0840] Step 1:

[0841] The device acquires audio data from a built-in or external microphone. The device's microphone driver captures the user's voice and environmental sounds.

[0842] Step 2:

[0843] The terminal divides the acquired voice data into packets and transmits them to the server using a network protocol.

[0844] Step 3:

[0845] The device receives the filtered voice data and emotion feedback sent from the server, and reconstructs the received packets into a single voice data.

[0846] Step 4:

[0847] The device plays the reconstructed audio data in real time over speakers or headphones, adjusting the volume and quality of the audio to provide the user with clear audio.

[0848] Step 5:

[0849] The device displays the received emotional feedback to the user, allowing the user to check their own emotional state in real time.

[0850] User operation steps

[0851] Step 1:

[0852] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[0853] Step 2:

[0854] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[0855] Step 3:

[0856] The user starts an online meeting or call and, during the meeting, proceeds with the conversation while receiving real-time emotional feedback and the effects of Noise Cancellation 2.0.

[0857] Specific examples

[0858] Step 1:

[0859] The user starts an application to hold a remote conference at a cafe and enables noise canceling.

[0860] Step 2:

[0861] The device captures the noise of the cafe and the user's voice and sends it to the server.

[0862] Step 3:

[0863] The server receives the voice data, identifies the user's voice using LLM, and removes noise.

[0864] Step 4:

[0865] The server runs the filtered voice data through an emotion engine to analyze the user's emotional state.

[0866] Step 5:

[0867] The server fine-tunes the voice filtering based on the results of the emotion engine and generates optimal voice data.

[0868] Step 6:

[0869] The server transmits the filtered voice data and the emotional feedback to the terminal.

[0870] Step 7:

[0871] The device provides the user with clear received audio and emotional feedback in real time.

[0872] This not only significantly reduces noise in the cafe and allows conversations to proceed smoothly, but also enables optimal voice communication that is tailored to the user's emotional state.

[0873] Example 2

[0874] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0875] In conventional online conference and call systems, clear communication is difficult due to poor audio quality in noisy environments. Furthermore, the lack of a function to recognize and provide feedback on the user's emotional state in real time hinders smooth communication.

[0876] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0877] In this invention, the server includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying the voice of a conversation partner using a large-scale language model for analyzing the preprocessed voice data, means for using an emotion engine for recognizing emotions from the voice data, means for adjusting the voice data based on the emotional state, and means for transmitting the filtered and adjusted voice data to the user terminal, thereby enabling clear voice communication even in noisy environments and further enabling smooth communication by providing feedback on the user's emotional state in real time.

[0878] "Audio data" refers to audio information acquired from and transmitted by a user's terminal.

[0879] "Preprocessing" refers to processing to improve the quality of audio data, such as applying a noise reduction filter or standardizing the sampling rate.

[0880] A "large-scale language model" is an advanced machine learning model that analyzes text and audio data to understand the context and meaning of conversations.

[0881] An "emotion engine" is a system that analyzes voice characteristics such as tone, pitch, and speed to identify a speaker's emotional state.

[0882] "Filtering" refers to the process of removing unnecessary noise and interference sounds from audio data and emphasizing important audio components.

[0883] "Terminal" refers to a device used to capture, send, receive, and play audio data.

[0884] "Encoding" refers to the process of converting audio data into a format that can be transmitted efficiently.

[0885] "Real-time" refers to immediate processing and feedback.

[0886] "Emotional feedback" refers to informing users of their emotional state and providing information to improve the quality of communication.

[0887] "Settings" are adjustments that a user makes to adjust the system's behavior or options.

[0888] This invention is a system that realizes clear voice communication in online conferences and noisy environments, and also recognizes and provides feedback on the user's emotions. In addition to cooperation between the server, terminals, and users, this system also includes an emotion engine, providing more advanced functionality.

[0889] Server Roles

[0890] The server plays a central role in receiving, analyzing, and providing feedback on voice data sent from the user's device. Specific hardware used includes a high-performance CPU, memory, and network interface. Software includes a noise reduction filter (e.g., RNNoise), a large-scale language model (e.g., OpenAI GPT-3), and an emotion engine (e.g., IBM Watson Tone Analyzer). These software and hardware components are used to perform the following processes:

[0891] 1. Receiving audio data

[0892] The server receives the voice data sent from the user's terminal and temporarily stores it in memory.

[0893] 2. Preprocessing of audio data

[0894] The server applies noise reduction filters to clean up the audio data and also standardizes the sampling rate to equalize the quality of the data.

[0895] 3. Analysis using large-scale language models

[0896] The server analyzes the preprocessed speech data using a large-scale language model to extract features to identify the speech of the conversation partner and distinguish it from ambient noise and unwanted speech.

[0897] 4. Emotion Recognition by Emotion Engine

[0898] The server uses an emotion engine to recognize the user's emotion from the voice data, which analyzes voice characteristics such as tone, pitch, and rate to identify the emotional state.

[0899] 5. Filtering and conditioning of audio data

[0900] The server emphasizes the identified important audio components and adjusts the volume and texture based on the output of the emotion engine.

[0901] 6. Encoding and transmitting filtered audio data

[0902] The server converts the filtered audio data into a format that can be transmitted efficiently and transmits it to the terminal via the network.

[0903] Device Role

[0904] The device is responsible for capturing audio data, sending it to the server, and receiving and playing back the filtered audio from the server. The hardware used includes an internal or external microphone, speakers or headphones, and a network interface. The software includes an audio capture driver and playback software. It performs the following tasks:

[0905] 1. Acquiring audio data

[0906] The device captures audio data using a microphone, and the captured data is temporarily stored in the device's memory.

[0907] 2. Sending audio data

[0908] The terminal transmits the acquired voice data to the server using a network protocol.

[0909] 3. Receiving filtered audio

[0910] The terminal receives the filtered audio data sent from the server and reconstructs it.

[0911] 4. Playing back received audio

[0912] The device then plays the reconstructed audio data over speakers or headphones, providing the user with clear audio.

[0913] 5. Displaying Emotional Feedback

[0914] The terminal receives the output of the emotion engine sent from the server and displays emotion feedback to the user in real time.

[0915] User operations

[0916] The user operates the system and configures the appropriate settings, including noise cancellation, voice emphasis, and whether or not emotional feedback is enabled. The user then launches the Noise Cancellation 2.0 application on the device, adjusts the settings, and starts an online meeting or call.

[0917] Specific examples

[0918] For example, if a user is holding a remote meeting at a coffee shop, the following happens:

[0919] 1. The user launches the Noise Cancelling 2.0 app and enables noise cancellation and voice enhancement.

[0920] 2. The device captures the noise of the cafe and the user's voice through its built-in microphone and sends the audio data to the server.

[0921] 3. The server receives the audio data and preprocesses it with a noise reduction filter.

[0922] 4. The server uses a large-scale language model to distinguish between the user's voice and noise.

[0923] 5. The server uses an emotion engine to identify the user's emotion from the voice data and perform filtering and adjustment.

[0924] 6. The server sends the filtered audio to the device.

[0925] 7. The device receives and plays the filtered audio data.

[0926] 8. The device displays emotional feedback to the user in real time.

[0927] Prompt Sentence Examples

[0928] Explain how Noise Cancellation 2.0 can be used to ensure clear voice communication and provide real-time feedback on user emotions during a remote meeting in a coffee shop.

[0929] As described above, the system of the present invention enables clear voice communication even in noisy environments, and furthermore, by recognizing and feeding back the user's emotions in real time, smooth communication can be realized.

[0930] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0931] Step 1:

[0932] Acquiring audio data

[0933] The device acquires audio data using a built-in or external microphone. The input is an audio signal from the microphone, and the output is digital audio data. The microphone driver starts up and captures the user's voice and environmental sounds in real time. This data is temporarily stored in the device's memory.

[0934] Step 2:

[0935] Sending audio data

[0936] The device sends the captured audio data to the server using a network protocol (e.g. TCP / IP). The input is digital audio data, which is converted into packets that are sent to the server as output. The audio data is divided into small packets and sent over the internet to the server.

[0937] Step 3:

[0938] Receiving voice data and storing it in memory

[0939] The server receives the voice data sent from the terminal and temporarily stores it in memory. The input is data packets, and the output is continuous digital voice data. The receiving module combines the data packets and reconstructs them into a single voice data.

[0940] Step 4:

[0941] Audio data preprocessing

[0942] The server applies a noise reduction filter (e.g., RNNoise) to clean up the audio data. It takes continuous digital audio data as input and produces denoised audio data as output. Noise is removed sample by sample and the sampling rate is normalized.

[0943] Step 5:

[0944] Analysis using large-scale language models

[0945] The server analyzes the preprocessed voice data using a large-scale language model (e.g., OpenAI GPT-3). The input is noise-removed voice data, and the output is the voice identity of the conversation partner. The language model performs contextual analysis and extracts features to distinguish the conversation partner's voice from noise and unwanted sounds.

[0946] Step 6:

[0947] Emotion recognition by emotion engine

[0948] The server uses an emotion engine (e.g., IBM Watson Tone Analyzer) to recognize the user's emotion from the voice data. Feature data is input, and emotional state information is obtained as output. The tone, pitch, and rate of the voice are analyzed to identify the emotional state.

[0949] Step 7:

[0950] Filtering and conditioning audio data

[0951] The server emphasizes the identified voice components and adjusts the volume and texture based on the output of the emotion engine. The inputs are emotional state information and voice identification information, and the output is adjusted voice data. The voice filter emphasizes important voice components and optimizes the volume and tone according to the emotion.

[0952] Step 8:

[0953] Encoding and transmitting filtered audio data

[0954] The server converts the filtered audio data into a format that can be transmitted efficiently (e.g., Opus codec) and sends it over the network to the device. The input is the conditioned audio data, and the output is the encoded audio data. The data is split into packets and transmitted again.

[0955] Step 9:

[0956] Receive and play filtered audio

[0957] The device receives the filtered audio data sent from the server and reconstructs it. The input is the encoded audio data, and the output is the reconstructed audio data. The reconstructed audio data is then played back through speakers or headphones, providing the user with clear audio.

[0958] Step 10:

[0959] Displaying Emotional Feedback

[0960] The device receives the output of the emotion engine sent from the server and displays emotional feedback to the user in real time. Emotional state information is input, and emotional feedback display is obtained as output. This allows the user to check their own emotional state in real time.

[0961] (Application example 2)

[0962] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0963] In recent years, with the increase in online meetings and remote work, the quality of voice communication in noisy environments has become an issue. Maintaining clear voice communication in noisy environments is also difficult during field work such as delivery work. Furthermore, there are currently limited methods for understanding delivery workers' emotions and stress levels in real time and providing appropriate feedback.

[0964] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving voice data from a user's information device, means for preprocessing the voice data, means for identifying the voice of the conversation partner using a large-scale language model for analyzing the preprocessed voice data, means for filtering environmental noise and unnecessary voice from the identified voice data, means for transmitting the filtered voice data to the user's information device, means for using an emotion engine for recognizing the user's emotion from the voice data, means for adjusting the voice data according to the user's emotional state, and means for displaying the adjusted voice data and emotion feedback on the user's information device. This not only enables clear voice communication even in noisy environments, but also makes it possible to grasp the user's emotional state in real time and provide appropriate feedback.

[0965] "Audio Data" means a digital recording of sound, typically captured using an audio input device such as a microphone.

[0966] "User's information equipment" refers to electronic devices used by users, such as smartphones, tablets, and laptops.

[0967] "Preprocessing" refers to performing processes such as noise reduction and standardization of the sampling rate on the acquired audio data.

[0968] A "large-scale language model" refers to a natural language processing model trained on a large amount of data, which enables the analysis and semantic understanding of speech data.

[0969] "Means for identifying the voice of a conversation partner" refers to a method or algorithm for extracting the voice of a specific speaker from multiple voice data.

[0970] "Environmental noise" refers to background noise or unwanted sounds that can be present during a conversation, such as traffic or wind noise.

[0971] "Filtering" refers to a process for emphasizing specific audio components and removing unwanted components.

[0972] "Emotion engine" refers to algorithms and technologies that analyze characteristics of voice data, such as tone, pitch, and speed, to identify a user's emotional state.

[0973] "Emotional feedback" refers to information that analyzes a user's emotional state and provides the results to the user in real time.

[0974] This invention is a system that improves the quality of voice communication in food delivery and simultaneously grasps the emotional state of delivery personnel in real time. The specific configuration and processing method of the system are described below.

[0975] Server Roles

[0976] The server plays a central role in receiving voice data from the user's information device (e.g., a smartphone) and recognizing emotions using an emotion engine. Specific processes include preprocessing the voice data, analyzing it using a large-scale language model, filtering environmental noise, and transmitting the voice data and emotional feedback.

[0977] 1. Receiving audio data:

[0978] The server receives the voice data sent from the delivery person's smartphone. At this stage, the voice data is received in packet format using a network protocol.

[0979] 2. Preprocessing the audio data:

[0980] The server processes the received audio data with a noise reduction filter and normalizes the sampling rate. For data processing, it uses an audio processing library (e.g., librosa).

[0981] 3. Large-scale Language Model (LLM) Analysis:

[0982] The preprocessed speech data is then analyzed using a large-scale language model, which extracts features from the speech data and identifies the speech of the conversation partner. The model uses a generative AI model (e.g., GPT-based).

[0983] 4. Emotion Recognition with Emotion Engine:

[0984] The server uses an emotion engine to recognize the delivery person's emotions from the voice data, which includes algorithms that analyze voice characteristics such as tone, pitch, and speed to identify emotional states.

[0985] 5. Audio filtering and conditioning:

[0986] The system removes environmental noise and unwanted sounds from the identified audio data to generate clear audio data, and also adjusts the audio data (e.g., volume) based on the results of emotion recognition.

[0987] 6. Sending filtered audio data and emotional feedback:

[0988] The server re-encodes the filtered audio data and transmits it to the user's information device in real time, and simultaneously transmits the emotional feedback and displays it on the information device.

[0989] Device Role

[0990] The terminal (i.e., the delivery person's smartphone) acquires, sends, and receives voice data, plays back filtered voice, and displays emotional feedback.

[0991] 1. Acquire audio data:

[0992] The device captures audio data from the built-in microphone and uses the device's microphone driver for audio input.

[0993] 2. Sending audio data:

[0994] The acquired voice data is divided into packets and sent to a server using a network protocol.

[0995] 3. Receiving and playing filtered audio:

[0996] It receives the filtered audio data sent from the server, reconstructs it, and plays it over speakers or headphones.

[0997] 4. Displaying Emotional Feedback:

[0998] Emotional feedback is displayed on the delivery person's smartphone, allowing them to check their own emotional state in real time.

[0999] User operations

[1000] The user (i.e., delivery person) activates the system, adjusts settings, and navigates the system through a process of capturing and verifying voice data.

[1001] 1. Boot the system:

[1002] The user launches the noise canceling application on the smartphone.

[1003] 2. Adjust the settings:

[1004] Users can adjust noise cancellation levels, enhance specific sounds, and other customization settings in the app's settings screen.

[1005] 3. Start a conversation:

[1006] Users can use the phone and voice guidance during delivery tasks, and the system provides real-time noise cancellation and emotional feedback during conversations, ensuring clear communication even in noisy environments.

[1007] Examples of specific examples and prompts

[1008] For example, when a delivery person is delivering a pizza, the system can provide clear voice communication even in a noisy traffic environment.The system also monitors the delivery person's emotional state in real time, and if the delivery person is under high stress, the support center will provide appropriate feedback.

[1009] Example prompt sentence:

[1010] Describe a smartphone app that maintains clear voice communication even when the user is in a noisy environment, and recognizes and provides feedback on the user's emotional state. The app performs voice data acquisition, noise reduction, emotion recognition, voice filtering, and emotional feedback display.

[1011] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1012] Step 1:

[1013] The user launches the application on their smartphone, which displays the initial application screen and allows them to adjust the noise cancellation level and other settings.

[1014] Step 2:

[1015] The device (smartphone) acquires voice data using a built-in microphone. Specifically, it captures the user's voice and surrounding environmental sounds (noise) and saves them as digital voice data. The acquired voice data is temporarily stored in the device's memory.

[1016] Step 3:

[1017] The terminal sends the acquired voice data to the server. The voice data is divided into packets and sent to the server using a network protocol (e.g., TCP / IP). The input is the raw voice data, and the output is the voice packets that arrive at the server.

[1018] Step 4:

[1019] The server receives the audio data sent from the device and preprocesses it by applying a noise reduction filter, which removes environmental noise and unwanted sounds in the audio signal. The input is raw audio data, and the output is preprocessed audio data with noise removed.

[1020] Step 5:

[1021] The server analyzes the preprocessed speech data using a large-scale language model (LLM) to identify the speech of the conversation partner. This analysis involves extracting and classifying speech features. The input is the preprocessed speech data, and the output is the identified speech segments of the conversation partner.

[1022] Step 6:

[1023] The server filters out environmental noise and unwanted sounds from the identified audio data by applying a filtering algorithm to highlight only the audio components of interest. This process results in clear audio data. The input is the identified audio data, and the output is the filtered, clear audio data.

[1024] Step 7:

[1025] The server uses an emotion engine to recognize the user's emotion from the voice data. It analyzes acoustic characteristics such as tone, pitch, and rate of the voice to identify the user's emotional state. The input is the filtered voice data, and the output is data indicating the user's emotional state.

[1026] Step 8:

[1027] The server adjusts the audio data according to the user's emotional state and generates appropriate audio feedback. For example, if the user is feeling stressed, it adjusts the volume or applies a specific filter. The input is the emotional state data and the filtered audio data, and the output is the adjusted audio data.

[1028] Step 9:

[1029] The server encodes the filtered and adjusted voice data and sends it to the user's device, and simultaneously sends emotional feedback to be displayed on the device. The input is the adjusted voice data and emotional feedback data, and the output is the data that arrives at the user's device.

[1030] Step 10:

[1031] The terminal reconstructs the received filtered and adjusted audio data and plays it over speakers or headphones. The input is the audio packets received from the server, and the output is the clear audio being played.

[1032] Step 11:

[1033] The device displays emotional feedback to the user, which is displayed on the smartphone screen as an icon or message indicating the user's emotional state. The input is the user's emotional feedback data, and the output is the displayed emotional feedback.

[1034] Through specific processing steps, users can maintain clear voice communication even in noisy environments and understand their own emotional state in real time. This system is very useful in food delivery situations.

[1035] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1036] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1037] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[1038] [Third embodiment]

[1039] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1040] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[1041] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1042] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[1043] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1044] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1045] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1046] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1047] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1048] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1049] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1050] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[1051] The present invention is a system for realizing clear voice communication in online conferences and in noisy environments. This system operates through cooperation between a server, a terminal, and a user.

[1052] Server Roles

[1053] The server plays a central role in receiving voice data sent from the user's device and analyzing it using a large-scale language model (LLM). The server has the following functions:

[1054] 1. Receiving audio data

[1055] The server receives the voice data from the user's terminal and temporarily stores the received voice data in the server's memory.

[1056] 2. Preprocessing of audio data

[1057] The server pre-processes the received audio data, which includes applying noise reduction filters and normalizing the sampling rate.

[1058] 3. Analysis using large-scale language models

[1059] The server analyzes the pre-processed speech data using a large-scale language model, which identifies the speech of the interlocutor and distinguishes it from background noise and unwanted speech.

[1060] 4. Audio Filtering

[1061] The server filters the identified audio data to emphasize important audio components, and the filtered audio data is re-encoded and converted into a format that can be transmitted efficiently.

[1062] 5. Sending filtered audio data

[1063] The server sends the filtered audio data to the user's device, returning it in real time so the user receives clear audio.

[1064] Device Role

[1065] The device acquires audio data, sends it to the server, receives filtered audio from the server, and plays it back. Specific functions include:

[1066] 1. Acquiring audio data

[1067] The device acquires audio data from a built-in or external microphone, capturing the user's voice and environmental sounds via a microphone driver.

[1068] 2. Sending audio data

[1069] The acquired voice data is sent from the device to the server, which divides the voice data into packets and sends them to the server using a network protocol.

[1070] 3. Receiving filtered audio

[1071] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[1072] 4. Playing back received audio

[1073] The reconstructed audio data is played back by the device through speakers or headphones, and the volume and quality of the audio can be adjusted to provide the user with clear audio.

[1074] User operations

[1075] Users interact with the system and adjust settings to take advantage of noise cancellation and specific voice enhancement, including:

[1076] 1. Booting the system

[1077] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[1078] 2. Adjust your settings

[1079] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[1080] 3. Start a conversation

[1081] Users can start an online meeting or call and enjoy the benefits of Noise Cancellation 2.0 in real time while continuing the conversation.

[1082] Specific examples

[1083] For example, if a user is holding a remote meeting at a coffee shop, the following happens:

[1084] 1. The user launches the application and enables noise cancellation.

[1085] 2. The device captures the noise in the cafe and the user's voice and sends them to the server.

[1086] 3. The server receives the voice data and uses LLM to identify the user's voice and remove noise.

[1087] 4. The server sends the filtered audio to the device.

[1088] 5. The device plays the received clear audio to the user in real time.

[1089] This significantly reduces noise in the cafe, allowing conversations to proceed smoothly, enabling users to hold meetings more efficiently.In this way, the system proposed by this invention improves the quality of communication in online meetings and noisy environments.

[1090] The processing flow will be explained below.

[1091] Server Processing Steps

[1092] Step 1:

[1093] The server receives the audio stream from the user's terminal, and the received audio data is temporarily stored in the server's memory.

[1094] Step 2:

[1095] The server pre-processes the received audio data, which may include applying noise reduction filters and normalizing the sampling rate.

[1096] Step 3:

[1097] The server inputs the preprocessed speech data into a large-scale language model (LLM), which analyzes the speech data and extracts features to identify the speech of the conversation partner.

[1098] Step 4:

[1099] The server filters the identified audio data, a process that emphasizes important audio components and reduces ambient noise and unwanted voices.

[1100] Step 5:

[1101] The server encodes the filtered audio data into a format that can be transmitted efficiently.

[1102] Step 6:

[1103] The server transmits the encoded and filtered audio data to the user's terminal in real time via the Internet.

[1104] Terminal processing steps

[1105] Step 1:

[1106] The device acquires audio data from a built-in or external microphone. The device's microphone driver captures the user's voice and environmental sounds.

[1107] Step 2:

[1108] The terminal divides the acquired voice data into packets and transmits them to the server using a network protocol.

[1109] Step 3:

[1110] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[1111] Step 4:

[1112] The device plays the reconstructed audio data through speakers or headphones, adjusting the volume and quality of the audio to provide the user with clear audio.

[1113] User operation steps

[1114] Step 1:

[1115] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[1116] Step 2:

[1117] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[1118] Step 3:

[1119] The user starts an online meeting or call and continues the conversation while enjoying the real-time benefits of Noise Cancellation 2.0.

[1120] Specific examples

[1121] Step 1:

[1122] The user starts an application to hold a remote conference at a cafe and enables noise canceling.

[1123] Step 2:

[1124] The device captures the noise of the cafe and the user's voice and sends it to the server.

[1125] Step 3:

[1126] The server receives the audio data, identifies the user's voice with a large language model, and removes noise.

[1127] Step 4:

[1128] The server transmits the filtered audio to the terminal.

[1129] Step 5:

[1130] The terminal then plays the received clear audio to the user in real time.

[1131] This significantly reduces noise in the cafe, allowing conversations to proceed smoothly, allowing users to hold meetings efficiently.

[1132] Example 1

[1133] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1134] In conventional online meetings and noisy environments, voice communication is often hindered by ambient noise and unwanted sounds. This makes it difficult to achieve clear, crisp voice communication. In particular, in remote meetings and conference calls, if the speaker's voice cannot be clearly heard, the quality of communication declines and work efficiency also deteriorates. The present invention aims to solve these problems and provide a system that effectively clears voices even in noisy environments.

[1135] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1136] In this invention, the server includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying a conversation partner's voice using a generative AI model for analyzing the preprocessed voice data, means for filtering ambient noise and unwanted voices from the identified voice data, and means for transmitting the filtered voice data to the user terminal, thereby enabling the provision of clear, noise-free voice in real time.

[1137] "Audio data" refers to data that represents the user's voice and environmental sounds in digital form.

[1138] "Terminal" refers to a device that acquires voice data and communicates with the server. Examples include PCs, smartphones, and tablets.

[1139] "Server" is a central system for receiving, pre-processing, analyzing, filtering, and transmitting audio data.

[1140] "Preprocessing" refers to processing for applying a noise reduction filter to the received audio data and standardizing the sampling rate.

[1141] A "generative AI model" is a type of artificial intelligence model that uses a large-scale language model to analyze voice data, identify the voice of the conversation partner, and distinguish between unwanted voices and noise.

[1142] "Filtering" refers to processing for removing ambient noise and unwanted sounds from identified audio data and for emphasizing important audio components.

[1143] "Playback" refers to the process of making the filtered audio data audible to the user through the device's speakers or headphones.

[1144] "Settings" are options that allow users to customize the system's behavior, such as noise cancellation or emphasis on specific sounds.

[1145] The present invention is a system for realizing clear voice communication in online conferences and in noisy environments. This system operates through cooperation between a server, a terminal, and a user.

[1146] The server plays a central role in receiving voice data sent from the user's device and analyzing it using a large-scale language model (generative AI model). The server has the following functions:

[1147] 1. Receiving audio data

[1148] The server receives the voice data from the user's device. The received voice data is temporarily stored in the server's memory. For example, the data can be received using open source voice communication software.

[1149] 2. Preprocessing of audio data

[1150] The server preprocesses the received audio data, applying noise reduction filters and standardizing the sampling rate. This process uses tools such as LibROSA, a Python audio processing library, to reduce noise.

[1151] 3. Analysis using large-scale language models

[1152] The server then analyzes the preprocessed audio data using a generative AI model, which identifies the voice of the person speaking and distinguishes it from background noise and unwanted sounds. At this stage, the LLM can be, for example, GPT-4.

[1153] 4. Audio Filtering

[1154] The server filters the identified audio data to emphasize important audio components, and then re-encodes the filtered audio data into a format that can be transmitted efficiently. Specifically, the audio data can be compressed using the OPUS codec.

[1155] 5. Sending filtered audio data

[1156] The server sends the filtered audio data to the user's device, returning it in real time so the user receives clear audio.

[1157] The device's role is to acquire audio data, send it to the server, receive filtered audio from the server, and play it back. Specific functions include:

[1158] 1. Acquiring audio data

[1159] The device acquires audio data from a built-in or external microphone, for example, a USB microphone or a built-in microphone to capture the user's voice and environmental sounds.

[1160] 2. Sending audio data

[1161] The acquired voice data is sent from the terminal to the server, where it is divided into packets and sent to the server using the TCP / IP protocol.

[1162] 3. Receiving filtered audio

[1163] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[1164] 4. Playing back received audio

[1165] The reconstructed audio data is played back by the device through speakers or headphones, specifically via the device's audio driver.

[1166] Users can interact with the system and adjust settings to take advantage of noise cancellation and specific voice enhancement, including:

[1167] 1. Booting the system

[1168] The user launches the noise cancelling application on their device and clicks on the application icon to open the software.

[1169] 2. Adjust your settings

[1170] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[1171] 3. Start a conversation

[1172] Users can start an online meeting or call and enjoy the noise-canceling effect in real time while continuing the conversation.

[1173] Specific examples

[1174] For example, if users are having a remote meeting at a coffee shop, the process would be as follows:

[1175] 1. The user launches the application and enables noise cancellation.

[1176] 2. The device captures the noise of the cafe and the user's voice from the microphone and sends them to the server.

[1177] 3. The server receives the audio data and applies a noise reduction filter as preprocessing.

[1178] 4. The server uses a generative AI model (e.g., GPT-4) to identify the user's voice and remove noise.

[1179] 5. The server sends the filtered audio to the device.

[1180] 6. The device reconstructs the received audio data and plays it over speakers or headphones.

[1181] This significantly reduces the noise in the cafe, allowing for clearer audio during remote meetings. An example prompt is, "Design a system to efficiently cancel noise and make the user's voice clear in a situation where a remote meeting is being held in a cafe. Explain how the user's device captures audio data and communicates it with the server. Also, explain how the server processes the audio data and sends the filtered data back to the device."

[1182] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1183] Step 1: Acquire audio data (device)

[1184] The device uses a built-in or external microphone to capture voice data. This process captures the user's voice and environmental sounds. For example, when a user speaks in a cafe, the device's microphone picks up the user's voice and background noise at the same time. The input is an analog voice signal, and the output is digital voice data.

[1185] Step 2: Sending audio data (device)

[1186] The device sends the captured audio data to the server via the network. During this process, the audio data is divided into packets and transmitted using the TCP / IP protocol. The specific operations include packetizing the data and transmitting it over the network. The input is the digital audio data, and the output is the transmitted data packets.

[1187] Step 3: Receiving audio data (server)

[1188] The server receives voice data sent from the terminal via the network. The received data is temporarily stored in the server's memory. Specific operations include receiving data via the network interface and storing it in memory. The input is the transmitted data packet, and the output is the digital voice data stored in memory.

[1189] Step 4: Preprocessing the audio data (server)

[1190] The server applies a noise reduction filter to the received audio data and standardizes the sampling rate. Specifically, noise reduction is performed using LibROSA, a Python audio processing library. The input is digital audio data stored in memory, and the output is audio data with reduced noise and a standardized sampling rate.

[1191] Step 5: Analysis using large-scale language models (server)

[1192] The server analyzes the preprocessed audio data using a generative AI model. LLM distinguishes between the voice of the conversation partner and background noise from the audio data. In this process, GPT-4 can be used as the model. The input is the preprocessed audio data, and the output is the identified audio components.

[1193] Step 6: Audio Filtering (Server)

[1194] The server uses LLM to filter the identified audio data and emphasize important audio components. This filtering process applies multiple filters to emphasize specific frequency bands. The input is the identified audio components, and the output is clear filtered audio data.

[1195] Step 7: Sending the filtered audio data (server)

[1196] The server sends the filtered audio data to the device. The audio data is efficiently compressed and transferred in real time. Specifically, the OPUS codec is used to compress the data and send it over the network. The input is the filtered audio data, and the output is compressed audio data packets.

[1197] Step 8: Receive filtered audio (terminal)

[1198] The terminal receives the filtered voice data sent from the server and reconstructs the received data packets into a single voice data packet. Specific operations include receiving packets from the network and reconstructing them. The input is the compressed voice data packet, and the output is the reconstructed digital voice data.

[1199] Step 9: Playing back received audio (device)

[1200] The reconstructed audio data is then played back through the device's audio driver on speakers or headphones, with the volume and sound quality adjusted appropriately to provide clear audio to the user. The input is the reconstructed digital audio data, and the output is audio that the user can hear.

[1201] Step 10: User configuration operations (User)

[1202] A user uses a noise-canceling application to adjust settings, including whether to emphasize specific audio components and adjusting the level of noise cancellation. The input is the user's actions and settings, and the output is the customized system settings.

[1203] (Application example 1)

[1204] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1205] With conventional factory robots, voice communication between workers and robots is often hindered by loud noise in the factory, posing problems that affect work efficiency and safety. Furthermore, the lack of appropriate noise filtering methods increases the risk of workers misunderstanding instructions and causing operational errors. Therefore, there is a need for a method that enables clear voice communication even in noisy environments, allowing robots to work safely and efficiently.

[1206] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1207] In this invention, the server includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying the voice of a conversation partner using a large-scale language model for analyzing the preprocessed voice data, means for filtering ambient noise and unwanted voices from the identified voice data, means for transmitting the filtered voice data to the user terminal, and means for realizing voice communication between a worker and a robot in a noisy factory environment, thereby enabling clear voice communication between the worker and the robot even in a noisy factory environment.

[1208] "Voice data" means electronically recorded sound information, particularly data intended for communication by human voice.

[1209] A "user terminal" is a device for transmitting and receiving voice data, and is equipped with a microphone and a speaker.

[1210] "Preprocessing" refers to the initial data processing operations performed on received audio data, specifically noise reduction and standardization of the sampling rate.

[1211] A "large-scale language model" is a machine learning model that learns from large amounts of data for a specific task and analyzes or generates natural language.

[1212] "Interviewer voice identification" refers to the process of detecting and recognizing a specific person's voice from audio data.

[1213] "Noise" refers to background sounds and interfering sounds other than the target voice, and is a factor that reduces the quality of voice communication.

[1214] "Filtering" refers to the process of removing unnecessary components from audio data and extracting and emphasizing necessary parts.

[1215] "Noise environment within a factory" refers to a workplace environment where there is a lot of machinery and other work noise, and where noise levels are generally high.

[1216] "Workers" refers to personnel who operate machinery and equipment within a factory.

[1217] A "robot" is a programmable mechanical device capable of performing specific tasks automatically.

[1218] "Voice communication" refers to the transmission of information between people or machines via voice.

[1219] "Misoperation" refers to an inappropriate response by a robot to unintended operations or instructions, which has a negative impact on production efficiency and safety.

[1220] This invention is a system that realizes clear voice communication between a worker and a robot in a noisy factory environment. This system operates through cooperation between a server, a terminal, and a robot.

[1221] Server Roles

[1222] The server is responsible for the central processing of voice data. Specifically, it has the following functions:

[1223] 1. Receiving audio data

[1224] The server receives the voice data transmitted from the worker's terminal and temporarily stores the received voice data in the server's memory.

[1225] 2. Preprocessing of audio data

[1226] The server applies a noise reduction filter to the received audio data and normalizes the sampling rate, improving the quality of the data.

[1227] 3. Analysis using large-scale language models

[1228] The server uses the pre-processed speech data to analyze it with a large-scale language model to identify the worker's voice, and specific voice instructions are extracted from the analysis.

[1229] 4. Audio Filtering

[1230] The server filters out ambient noise from the identified audio data and emphasizes the important audio components. The filtered audio data is then re-encoded and converted into a format that can be transmitted efficiently.

[1231] 5. Sending filtered audio data

[1232] The server sends the filtered audio data to the robot, which then initiates actions based on it.

[1233] Device Role

[1234] The terminal is responsible for receiving and transmitting audio data. Specifically, it has the following functions:

[1235] 1. Acquiring audio data

[1236] The device acquires audio data from a built-in or external microphone, which is captured via a microphone driver.

[1237] 2. Sending audio data

[1238] The acquired voice data is sent from the device to the server, where it is divided into packets and transmitted using a network protocol.

[1239] 3. Receiving and playing filtered audio

[1240] The terminal receives the filtered voice data sent from the server and plays it through a speaker or headphones, allowing the worker to hear clear audio.

[1241] The role of robots

[1242] The robot operates based on filtered voice data from the server, and has the following functions:

[1243] 1. Analysis of audio data

[1244] The robot analyzes the filtered voice data sent from the server and begins to act as instructed.

[1245] 2. Real-time response

[1246] The robot responds to the worker's instructions in real time and carries out tasks efficiently.

[1247] Specific examples

[1248] For example, when a worker in a factory commands a robot to "start the next task," the flow is as follows:

[1249] 1. The worker captures the voice with the microphone on the terminal and sends it to the server.

[1250] 2. The server analyzes the received voice data and identifies the specified instructions.

[1251] 3. The server sends the filtered audio data to the robot.

[1252] 4. The robot follows the instructions and begins the next task.

[1253] Prompt Sentence Examples

[1254] "Design a voice instruction analysis system for a noisy factory environment. The system must filter out the noise in the factory and clearly convey the operator's instructions to the robot."

[1255] In this way, the system of the present invention enables clear voice communication between workers and robots even in noisy factory environments.

[1256] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1257] Step 1:

[1258] Acquiring audio data

[1259] The terminal uses a built-in or external microphone to acquire the worker's voice data. The input is a voice signal, and the output is digital voice data. This data is captured via a microphone driver.

[1260] Step 2:

[1261] Sending audio data

[1262] The device captures audio data and sends it to the server. The input is digital audio data, and the output is packet data sent to the server. The audio data is divided into packets and sent to the server using a network protocol.

[1263] Step 3:

[1264] Receiving audio data

[1265] The server receives the voice data sent from the terminal. The input is packet data, and the output is digital voice data temporarily stored in memory.

[1266] Step 4:

[1267] Audio data preprocessing

[1268] The server applies a noise reduction filter to the received audio data and normalizes the sampling rate. The input is digital audio data and the output is pre-processed digital audio data. Software filtering techniques are used for noise reduction and the sampling rate is changed to a normalized value.

[1269] Step 5:

[1270] Analysis using large-scale language models

[1271] The server analyzes the preprocessed voice data using a large-scale language model. The input is the preprocessed digital voice data, and the output is the identified voice instruction data. The large-scale language model analyzes the voice characteristics and extracts specific voice instructions.

[1272] Step 6:

[1273] Audio Filtering

[1274] The server removes ambient noise from the identified voice data and emphasizes the important voice components. The input is the identified voice instruction data, and the output is the filtered digital voice data. The filtering uses automatic noise reduction technology to emphasize only the necessary parts.

[1275] Step 7:

[1276] Sending filtered audio data

[1277] The server sends the filtered audio data to the robot. The input is filtered digital audio data, and the output is packets of data sent to the robot. This data is encoded and converted into a format that can be transmitted efficiently.

[1278] Step 8:

[1279] Analyzes voice data and starts operation

[1280] The robot analyzes the filtered voice data sent from the server and starts to act based on the instructions. The input is filtered digital voice data, and the output is the robot's specific action. The robot processes the received voice data and performs the appropriate action.

[1281] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1282] This invention is a system that realizes clear voice communication in online conferences and noisy environments, and also recognizes and provides feedback on the user's emotions. In addition to cooperation between the server, terminals, and users, this system also includes an emotion engine, providing more advanced functionality.

[1283] Server Roles

[1284] The server plays a central role in receiving voice data sent from the user's device and analyzing it using a large-scale language model (LLM). It also uses an emotion engine to recognize the user's emotional state and reflect it in voice filtering. Specific functions include:

[1285] 1. Receiving audio data

[1286] The server receives the voice data from the user's terminal and temporarily stores the received voice data in the server's memory.

[1287] 2. Preprocessing of audio data

[1288] The server pre-processes the received audio data, which includes applying noise reduction filters and normalizing the sampling rate.

[1289] 3. Analysis using large-scale language models

[1290] The server then analyzes the preprocessed speech data using a large-scale language model, which extracts features to identify the speech of the conversation partner and distinguish it from background noise and unwanted speech.

[1291] 4. Emotion Recognition by Emotion Engine

[1292] The server uses an emotion engine to recognize the user's emotion from the voice data. The emotion engine analyzes voice characteristics such as tone, pitch, and rate to identify the emotional state.

[1293] 5. Audio filtering and conditioning

[1294] The server filters the identified audio data, emphasizes important audio components, and adjusts the filtering and volume based on the emotion engine to provide optimal audio feedback according to the user's emotional state.

[1295] 6. Sending filtered audio data

[1296] The server encodes the filtered audio data and converts it into a format that can be transmitted efficiently, then transmits it in real time over the Internet to the user's device.

[1297] Device Role

[1298] The device is responsible for capturing audio data, sending it to the server, and receiving and playing back the filtered audio from the server. Specific functions include:

[1299] 1. Acquiring audio data

[1300] The device acquires audio data from a built-in or external microphone. The device's microphone driver captures the user's voice and environmental sounds.

[1301] 2. Sending audio data

[1302] The acquired voice data is sent from the device to the server, which divides the voice data into packets and sends them to the server using a network protocol.

[1303] 3. Receiving filtered audio

[1304] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[1305] 4. Playing back received audio

[1306] The reconstructed audio data is played back by the device through speakers or headphones, and the device adjusts the volume and quality of the audio to provide clear audio to the user.

[1307] 5. Displaying Emotional Feedback

[1308] The device receives the output of the emotion engine sent from the server and provides emotion feedback to the user, allowing the user to check their own emotional state in real time.

[1309] User operations

[1310] Users interact with the system and adjust settings to enable noise cancellation, emphasis on specific voices, and emotional feedback, including:

[1311] 1. Booting the system

[1312] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[1313] 2. Adjust your settings

[1314] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[1315] 3. Start a conversation

[1316] The user starts an online meeting or call and, during the meeting, proceeds with the conversation while receiving real-time emotional feedback and the effects of Noise Cancellation 2.0.

[1317] Specific examples

[1318] For example, if a user is holding a remote meeting at a coffee shop, the following happens:

[1319] 1. The user launches the application and enables noise cancellation.

[1320] 2. The device captures the noise in the cafe and the user's voice and sends them to the server.

[1321] 3. The server receives the voice data and uses LLM to identify the user's voice and remove noise.

[1322] 4. The server uses an emotion engine to recognize the user's emotions and reflect them in filtering.

[1323] 5. The server sends the filtered audio and emotional feedback to the device.

[1324] 6. The device receives clear audio and provides emotional feedback to the user in real time.

[1325] This not only significantly reduces noise in the cafe and allows conversations to proceed smoothly, but also enables optimal voice communication that is tailored to the user's emotional state.

[1326] The processing flow will be explained below.

[1327] Server Processing Steps

[1328] Step 1:

[1329] The server receives the audio stream from the user's terminal, and the received audio data is temporarily stored in the server's memory.

[1330] Step 2:

[1331] The server pre-processes the received audio data, which may include applying noise reduction filters and normalizing the sampling rate.

[1332] Step 3:

[1333] The server inputs the preprocessed speech data into a large-scale language model (LLM), which analyzes the speech data and extracts features to identify the speech of the conversation partner.

[1334] Step 4:

[1335] The server uses the LLM analysis results to identify and filter out ambient noise and unwanted sounds, highlighting important voice components.

[1336] Step 5:

[1337] The server inputs the filtered voice data into an emotion engine, which analyzes voice characteristics such as tone, pitch, and rate to identify the user's emotional state.

[1338] Step 6:

[1339] The server fine-tunes the voice filtering settings based on the emotional information obtained from the emotion engine. For example, if the user is feeling stressed, the server further reduces ambient noise.

[1340] Step 7:

[1341] The server encodes the filtered audio data and the output of the emotion engine into a format that can be transmitted efficiently.

[1342] Step 8:

[1343] The server transmits the encoded filtered voice data and emotional feedback to the user's device in real time via the Internet.

[1344] Terminal processing steps

[1345] Step 1:

[1346] The device acquires audio data from a built-in or external microphone. The device's microphone driver captures the user's voice and environmental sounds.

[1347] Step 2:

[1348] The terminal divides the acquired voice data into packets and transmits them to the server using a network protocol.

[1349] Step 3:

[1350] The device receives the filtered voice data and emotion feedback sent from the server, and reconstructs the received packets into a single voice data.

[1351] Step 4:

[1352] The device plays the reconstructed audio data in real time over speakers or headphones, adjusting the volume and quality of the audio to provide the user with clear audio.

[1353] Step 5:

[1354] The device displays the received emotional feedback to the user, allowing the user to check their own emotional state in real time.

[1355] User operation steps

[1356] Step 1:

[1357] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[1358] Step 2:

[1359] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[1360] Step 3:

[1361] The user starts an online meeting or call and, during the meeting, proceeds with the conversation while receiving real-time emotional feedback and the effects of Noise Cancellation 2.0.

[1362] Specific examples

[1363] Step 1:

[1364] The user starts an application to hold a remote conference at a cafe and enables noise canceling.

[1365] Step 2:

[1366] The device captures the noise of the cafe and the user's voice and sends it to the server.

[1367] Step 3:

[1368] The server receives the voice data, identifies the user's voice using LLM, and removes noise.

[1369] Step 4:

[1370] The server runs the filtered voice data through an emotion engine to analyze the user's emotional state.

[1371] Step 5:

[1372] The server fine-tunes the voice filtering based on the results of the emotion engine and generates optimal voice data.

[1373] Step 6:

[1374] The server transmits the filtered voice data and the emotional feedback to the terminal.

[1375] Step 7:

[1376] The device provides the user with clear received audio and emotional feedback in real time.

[1377] This not only significantly reduces noise in the cafe and allows conversations to proceed smoothly, but also enables optimal voice communication that is tailored to the user's emotional state.

[1378] Example 2

[1379] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1380] In conventional online conference and call systems, clear communication is difficult due to poor audio quality in noisy environments. Furthermore, the lack of a function to recognize and provide feedback on the user's emotional state in real time hinders smooth communication.

[1381] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1382] In this invention, the server includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying the voice of a conversation partner using a large-scale language model for analyzing the preprocessed voice data, means for using an emotion engine for recognizing emotions from the voice data, means for adjusting the voice data based on the emotional state, and means for transmitting the filtered and adjusted voice data to the user terminal, thereby enabling clear voice communication even in noisy environments and further enabling smooth communication by providing feedback on the user's emotional state in real time.

[1383] "Audio data" refers to audio information acquired from and transmitted by a user's terminal.

[1384] "Preprocessing" refers to processing to improve the quality of audio data, such as applying a noise reduction filter or standardizing the sampling rate.

[1385] A "large-scale language model" is an advanced machine learning model that analyzes text and audio data to understand the context and meaning of conversations.

[1386] An "emotion engine" is a system that analyzes voice characteristics such as tone, pitch, and speed to identify a speaker's emotional state.

[1387] "Filtering" refers to the process of removing unnecessary noise and interference sounds from audio data and emphasizing important audio components.

[1388] "Terminal" refers to a device used to capture, send, receive, and play audio data.

[1389] "Encoding" refers to the process of converting audio data into a format that can be transmitted efficiently.

[1390] "Real-time" refers to immediate processing and feedback.

[1391] "Emotional feedback" refers to informing users of their emotional state and providing information to improve the quality of communication.

[1392] "Settings" are adjustments that a user makes to adjust the system's behavior or options.

[1393] This invention is a system that realizes clear voice communication in online conferences and noisy environments, and also recognizes and provides feedback on the user's emotions. In addition to cooperation between the server, terminals, and users, this system also includes an emotion engine, providing more advanced functionality.

[1394] Server Roles

[1395] The server plays a central role in receiving, analyzing, and providing feedback on voice data sent from the user's device. Specific hardware used includes a high-performance CPU, memory, and network interface. Software includes a noise reduction filter (e.g., RNNoise), a large-scale language model (e.g., OpenAI GPT-3), and an emotion engine (e.g., IBM Watson Tone Analyzer). These software and hardware components are used to perform the following processes:

[1396] 1. Receiving audio data

[1397] The server receives the voice data sent from the user's terminal and temporarily stores it in memory.

[1398] 2. Preprocessing of audio data

[1399] The server applies noise reduction filters to clean up the audio data and also standardizes the sampling rate to equalize the quality of the data.

[1400] 3. Analysis using large-scale language models

[1401] The server analyzes the preprocessed speech data using a large-scale language model to extract features to identify the speech of the conversation partner and distinguish it from ambient noise and unwanted speech.

[1402] 4. Emotion Recognition by Emotion Engine

[1403] The server uses an emotion engine to recognize the user's emotion from the voice data, which analyzes voice characteristics such as tone, pitch, and rate to identify the emotional state.

[1404] 5. Filtering and conditioning of audio data

[1405] The server emphasizes the identified important audio components and adjusts the volume and texture based on the output of the emotion engine.

[1406] 6. Encoding and transmitting filtered audio data

[1407] The server converts the filtered audio data into a format that can be transmitted efficiently and transmits it to the terminal via the network.

[1408] Device Role

[1409] The device is responsible for capturing audio data, sending it to the server, and receiving and playing back the filtered audio from the server. The hardware used includes an internal or external microphone, speakers or headphones, and a network interface. The software includes an audio capture driver and playback software. It performs the following tasks:

[1410] 1. Acquiring audio data

[1411] The device captures audio data using a microphone, and the captured data is temporarily stored in the device's memory.

[1412] 2. Sending audio data

[1413] The terminal transmits the acquired voice data to the server using a network protocol.

[1414] 3. Receiving filtered audio

[1415] The terminal receives the filtered audio data sent from the server and reconstructs it.

[1416] 4. Playing back received audio

[1417] The device then plays the reconstructed audio data over speakers or headphones, providing the user with clear audio.

[1418] 5. Displaying Emotional Feedback

[1419] The terminal receives the output of the emotion engine sent from the server and displays emotion feedback to the user in real time.

[1420] User operations

[1421] The user operates the system and configures the appropriate settings, including noise cancellation, voice emphasis, and whether or not emotional feedback is enabled. The user then launches the Noise Cancellation 2.0 application on the device, adjusts the settings, and starts an online meeting or call.

[1422] Specific examples

[1423] For example, if a user is holding a remote meeting at a coffee shop, the following happens:

[1424] 1. The user launches the Noise Cancelling 2.0 app and enables noise cancellation and voice enhancement.

[1425] 2. The device captures the noise of the cafe and the user's voice through its built-in microphone and sends the audio data to the server.

[1426] 3. The server receives the audio data and preprocesses it with a noise reduction filter.

[1427] 4. The server uses a large-scale language model to distinguish between the user's voice and noise.

[1428] 5. The server uses an emotion engine to identify the user's emotion from the voice data and perform filtering and adjustment.

[1429] 6. The server sends the filtered audio to the device.

[1430] 7. The device receives and plays the filtered audio data.

[1431] 8. The device displays emotional feedback to the user in real time.

[1432] Prompt Sentence Examples

[1433] Explain how Noise Cancellation 2.0 can be used to ensure clear voice communication and provide real-time feedback on user emotions during a remote meeting in a coffee shop.

[1434] As described above, the system of the present invention enables clear voice communication even in noisy environments, and furthermore, by recognizing and feeding back the user's emotions in real time, smooth communication can be realized.

[1435] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1436] Step 1:

[1437] Acquiring audio data

[1438] The device acquires audio data using a built-in or external microphone. The input is an audio signal from the microphone, and the output is digital audio data. The microphone driver starts up and captures the user's voice and environmental sounds in real time. This data is temporarily stored in the device's memory.

[1439] Step 2:

[1440] Sending audio data

[1441] The device sends the captured audio data to the server using a network protocol (e.g. TCP / IP). The input is digital audio data, which is converted into packets that are sent to the server as output. The audio data is divided into small packets and sent over the internet to the server.

[1442] Step 3:

[1443] Receiving voice data and storing it in memory

[1444] The server receives the voice data sent from the terminal and temporarily stores it in memory. The input is data packets, and the output is continuous digital voice data. The receiving module combines the data packets and reconstructs them into a single voice data.

[1445] Step 4:

[1446] Audio data preprocessing

[1447] The server applies a noise reduction filter (e.g., RNNoise) to clean up the audio data. It takes continuous digital audio data as input and produces denoised audio data as output. Noise is removed sample by sample and the sampling rate is normalized.

[1448] Step 5:

[1449] Analysis using large-scale language models

[1450] The server analyzes the preprocessed voice data using a large-scale language model (e.g., OpenAI GPT-3). The input is noise-removed voice data, and the output is the voice identity of the conversation partner. The language model performs contextual analysis and extracts features to distinguish the conversation partner's voice from noise and unwanted sounds.

[1451] Step 6:

[1452] Emotion recognition by emotion engine

[1453] The server uses an emotion engine (e.g., IBM Watson Tone Analyzer) to recognize the user's emotion from the voice data. Feature data is input, and emotional state information is obtained as output. The tone, pitch, and rate of the voice are analyzed to identify the emotional state.

[1454] Step 7:

[1455] Filtering and conditioning audio data

[1456] The server emphasizes the identified voice components and adjusts the volume and texture based on the output of the emotion engine. The inputs are emotional state information and voice identification information, and the output is adjusted voice data. The voice filter emphasizes important voice components and optimizes the volume and tone according to the emotion.

[1457] Step 8:

[1458] Encoding and transmitting filtered audio data

[1459] The server converts the filtered audio data into a format that can be transmitted efficiently (e.g., Opus codec) and sends it over the network to the device. The input is the conditioned audio data, and the output is the encoded audio data. The data is split into packets and transmitted again.

[1460] Step 9:

[1461] Receive and play filtered audio

[1462] The device receives the filtered audio data sent from the server and reconstructs it. The input is the encoded audio data, and the output is the reconstructed audio data. The reconstructed audio data is then played back through speakers or headphones, providing the user with clear audio.

[1463] Step 10:

[1464] Displaying Emotional Feedback

[1465] The device receives the output of the emotion engine sent from the server and displays emotional feedback to the user in real time. Emotional state information is input, and emotional feedback display is obtained as output. This allows the user to check their own emotional state in real time.

[1466] (Application example 2)

[1467] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1468] In recent years, with the increase in online meetings and remote work, the quality of voice communication in noisy environments has become an issue. Maintaining clear voice communication in noisy environments is also difficult during field work such as delivery work. Furthermore, there are currently limited methods for understanding delivery workers' emotions and stress levels in real time and providing appropriate feedback.

[1469] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving voice data from a user's information device, means for preprocessing the voice data, means for identifying the voice of the conversation partner using a large-scale language model for analyzing the preprocessed voice data, means for filtering environmental noise and unnecessary voice from the identified voice data, means for transmitting the filtered voice data to the user's information device, means for using an emotion engine for recognizing the user's emotion from the voice data, means for adjusting the voice data according to the user's emotional state, and means for displaying the adjusted voice data and emotion feedback on the user's information device. This not only enables clear voice communication even in noisy environments, but also makes it possible to grasp the user's emotional state in real time and provide appropriate feedback.

[1470] "Audio Data" means a digital recording of sound, typically captured using an audio input device such as a microphone.

[1471] "User's information equipment" refers to electronic devices used by users, such as smartphones, tablets, and laptops.

[1472] "Preprocessing" refers to performing processes such as noise reduction and standardization of the sampling rate on the acquired audio data.

[1473] A "large-scale language model" refers to a natural language processing model trained on a large amount of data, which enables the analysis and semantic understanding of speech data.

[1474] "Means for identifying the voice of a conversation partner" refers to a method or algorithm for extracting the voice of a specific speaker from multiple voice data.

[1475] "Environmental noise" refers to background noise or unwanted sounds that can be present during a conversation, such as traffic or wind noise.

[1476] "Filtering" refers to a process for emphasizing specific audio components and removing unwanted components.

[1477] "Emotion engine" refers to algorithms and technologies that analyze characteristics of voice data, such as tone, pitch, and speed, to identify a user's emotional state.

[1478] "Emotional feedback" refers to information that analyzes a user's emotional state and provides the results to the user in real time.

[1479] This invention is a system that improves the quality of voice communication in food delivery and simultaneously grasps the emotional state of delivery personnel in real time. The specific configuration and processing method of the system are described below.

[1480] Server Roles

[1481] The server plays a central role in receiving voice data from the user's information device (e.g., a smartphone) and recognizing emotions using an emotion engine. Specific processes include preprocessing the voice data, analyzing it using a large-scale language model, filtering environmental noise, and transmitting the voice data and emotional feedback.

[1482] 1. Receiving audio data:

[1483] The server receives the voice data sent from the delivery person's smartphone. At this stage, the voice data is received in packet format using a network protocol.

[1484] 2. Preprocessing the audio data:

[1485] The server processes the received audio data with a noise reduction filter and normalizes the sampling rate. For data processing, it uses an audio processing library (e.g., librosa).

[1486] 3. Large-scale Language Model (LLM) Analysis:

[1487] The preprocessed speech data is then analyzed using a large-scale language model, which extracts features from the speech data and identifies the speech of the conversation partner. The model uses a generative AI model (e.g., GPT-based).

[1488] 4. Emotion Recognition with Emotion Engine:

[1489] The server uses an emotion engine to recognize the delivery person's emotions from the voice data, which includes algorithms that analyze voice characteristics such as tone, pitch, and speed to identify emotional states.

[1490] 5. Audio filtering and conditioning:

[1491] The system removes environmental noise and unwanted sounds from the identified audio data to generate clear audio data, and also adjusts the audio data (e.g., volume) based on the results of emotion recognition.

[1492] 6. Sending filtered audio data and emotional feedback:

[1493] The server re-encodes the filtered audio data and transmits it to the user's information device in real time, and simultaneously transmits the emotional feedback and displays it on the information device.

[1494] Device Role

[1495] The terminal (i.e., the delivery person's smartphone) acquires, sends, and receives voice data, plays back filtered voice, and displays emotional feedback.

[1496] 1. Acquire audio data:

[1497] The device captures audio data from the built-in microphone and uses the device's microphone driver for audio input.

[1498] 2. Sending audio data:

[1499] The acquired voice data is divided into packets and sent to a server using a network protocol.

[1500] 3. Receiving and playing filtered audio:

[1501] It receives the filtered audio data sent from the server, reconstructs it, and plays it over speakers or headphones.

[1502] 4. Displaying Emotional Feedback:

[1503] Emotional feedback is displayed on the delivery person's smartphone, allowing them to check their own emotional state in real time.

[1504] User operations

[1505] The user (i.e., delivery person) activates the system, adjusts settings, and navigates the system through a process of capturing and verifying voice data.

[1506] 1. Boot the system:

[1507] The user launches the noise canceling application on the smartphone.

[1508] 2. Adjust the settings:

[1509] Users can adjust noise cancellation levels, enhance specific sounds, and other customization settings in the app's settings screen.

[1510] 3. Start a conversation:

[1511] Users can use the phone and voice guidance during delivery tasks, and the system provides real-time noise cancellation and emotional feedback during conversations, ensuring clear communication even in noisy environments.

[1512] Examples of specific examples and prompts

[1513] For example, when a delivery person is delivering a pizza, the system can provide clear voice communication even in a noisy traffic environment.The system also monitors the delivery person's emotional state in real time, and if the delivery person is under high stress, the support center will provide appropriate feedback.

[1514] Example prompt sentence:

[1515] Describe a smartphone app that maintains clear voice communication even when the user is in a noisy environment, and recognizes and provides feedback on the user's emotional state. The app performs voice data acquisition, noise reduction, emotion recognition, voice filtering, and emotional feedback display.

[1516] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1517] Step 1:

[1518] The user launches the application on their smartphone, which displays the initial application screen and allows them to adjust the noise cancellation level and other settings.

[1519] Step 2:

[1520] The device (smartphone) acquires voice data using a built-in microphone. Specifically, it captures the user's voice and surrounding environmental sounds (noise) and saves them as digital voice data. The acquired voice data is temporarily stored in the device's memory.

[1521] Step 3:

[1522] The terminal sends the acquired voice data to the server. The voice data is divided into packets and sent to the server using a network protocol (e.g., TCP / IP). The input is the raw voice data, and the output is the voice packets that arrive at the server.

[1523] Step 4:

[1524] The server receives the audio data sent from the device and preprocesses it by applying a noise reduction filter, which removes environmental noise and unwanted sounds in the audio signal. The input is raw audio data, and the output is preprocessed audio data with noise removed.

[1525] Step 5:

[1526] The server analyzes the preprocessed speech data using a large-scale language model (LLM) to identify the speech of the conversation partner. This analysis involves extracting and classifying speech features. The input is the preprocessed speech data, and the output is the identified speech segments of the conversation partner.

[1527] Step 6:

[1528] The server filters out environmental noise and unwanted sounds from the identified audio data by applying a filtering algorithm to highlight only the audio components of interest. This process results in clear audio data. The input is the identified audio data, and the output is the filtered, clear audio data.

[1529] Step 7:

[1530] The server uses an emotion engine to recognize the user's emotion from the voice data. It analyzes acoustic characteristics such as tone, pitch, and rate of the voice to identify the user's emotional state. The input is the filtered voice data, and the output is data indicating the user's emotional state.

[1531] Step 8:

[1532] The server adjusts the audio data according to the user's emotional state and generates appropriate audio feedback. For example, if the user is feeling stressed, it adjusts the volume or applies a specific filter. The input is the emotional state data and the filtered audio data, and the output is the adjusted audio data.

[1533] Step 9:

[1534] The server encodes the filtered and adjusted voice data and sends it to the user's device, and simultaneously sends emotional feedback to be displayed on the device. The input is the adjusted voice data and emotional feedback data, and the output is the data that arrives at the user's device.

[1535] Step 10:

[1536] The terminal reconstructs the received filtered and adjusted audio data and plays it over speakers or headphones. The input is the audio packets received from the server, and the output is the clear audio being played.

[1537] Step 11:

[1538] The device displays emotional feedback to the user, which is displayed on the smartphone screen as an icon or message indicating the user's emotional state. The input is the user's emotional feedback data, and the output is the displayed emotional feedback.

[1539] Through specific processing steps, users can maintain clear voice communication even in noisy environments and understand their own emotional state in real time. This system is very useful in food delivery situations.

[1540] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1541] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1542] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1543] [Fourth embodiment]

[1544] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1545] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1546] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1547] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1548] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1549] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1550] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1551] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1552] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1553] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1554] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1555] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1556] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1557] The present invention is a system for realizing clear voice communication in online conferences and in noisy environments. This system operates through cooperation between a server, a terminal, and a user.

[1558] Server Roles

[1559] The server plays a central role in receiving voice data sent from the user's device and analyzing it using a large-scale language model (LLM). The server has the following functions:

[1560] 1. Receiving audio data

[1561] The server receives the voice data from the user's terminal and temporarily stores the received voice data in the server's memory.

[1562] 2. Preprocessing of audio data

[1563] The server pre-processes the received audio data, which includes applying noise reduction filters and normalizing the sampling rate.

[1564] 3. Analysis using large-scale language models

[1565] The server analyzes the pre-processed speech data using a large-scale language model, which identifies the speech of the interlocutor and distinguishes it from background noise and unwanted speech.

[1566] 4. Audio Filtering

[1567] The server filters the identified audio data to emphasize important audio components, and the filtered audio data is re-encoded and converted into a format that can be transmitted efficiently.

[1568] 5. Sending filtered audio data

[1569] The server sends the filtered audio data to the user's device, returning it in real time so the user receives clear audio.

[1570] Device Role

[1571] The device acquires audio data, sends it to the server, receives filtered audio from the server, and plays it back. Specific functions include:

[1572] 1. Acquiring audio data

[1573] The device acquires audio data from a built-in or external microphone, capturing the user's voice and environmental sounds via a microphone driver.

[1574] 2. Sending audio data

[1575] The acquired voice data is sent from the device to the server, which divides the voice data into packets and sends them to the server using a network protocol.

[1576] 3. Receiving filtered audio

[1577] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[1578] 4. Playing back received audio

[1579] The reconstructed audio data is played back by the device through speakers or headphones, and the volume and quality of the audio can be adjusted to provide the user with clear audio.

[1580] User operations

[1581] Users interact with the system and adjust settings to take advantage of noise cancellation and specific voice enhancement, including:

[1582] 1. Booting the system

[1583] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[1584] 2. Adjust your settings

[1585] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[1586] 3. Start a conversation

[1587] Users can start an online meeting or call and enjoy the benefits of Noise Cancellation 2.0 in real time while continuing the conversation.

[1588] Specific examples

[1589] For example, if a user is holding a remote meeting at a coffee shop, the following happens:

[1590] 1. The user launches the application and enables noise cancellation.

[1591] 2. The device captures the noise in the cafe and the user's voice and sends them to the server.

[1592] 3. The server receives the voice data and uses LLM to identify the user's voice and remove noise.

[1593] 4. The server sends the filtered audio to the device.

[1594] 5. The device plays the received clear audio to the user in real time.

[1595] This significantly reduces noise in the cafe, allowing conversations to proceed smoothly, enabling users to hold meetings more efficiently.In this way, the system proposed by this invention improves the quality of communication in online meetings and noisy environments.

[1596] The processing flow will be explained below.

[1597] Server Processing Steps

[1598] Step 1:

[1599] The server receives the audio stream from the user's terminal, and the received audio data is temporarily stored in the server's memory.

[1600] Step 2:

[1601] The server pre-processes the received audio data, which may include applying noise reduction filters and normalizing the sampling rate.

[1602] Step 3:

[1603] The server inputs the preprocessed speech data into a large-scale language model (LLM), which analyzes the speech data and extracts features to identify the speech of the conversation partner.

[1604] Step 4:

[1605] The server filters the identified audio data, a process that emphasizes important audio components and reduces ambient noise and unwanted voices.

[1606] Step 5:

[1607] The server encodes the filtered audio data into a format that can be transmitted efficiently.

[1608] Step 6:

[1609] The server transmits the encoded and filtered audio data to the user's terminal in real time via the Internet.

[1610] Terminal processing steps

[1611] Step 1:

[1612] The device acquires audio data from a built-in or external microphone. The device's microphone driver captures the user's voice and environmental sounds.

[1613] Step 2:

[1614] The terminal divides the acquired voice data into packets and transmits them to the server using a network protocol.

[1615] Step 3:

[1616] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[1617] Step 4:

[1618] The device plays the reconstructed audio data through speakers or headphones, adjusting the volume and quality of the audio to provide the user with clear audio.

[1619] User operation steps

[1620] Step 1:

[1621] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[1622] Step 2:

[1623] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[1624] Step 3:

[1625] The user starts an online meeting or call and continues the conversation while enjoying the real-time benefits of Noise Cancellation 2.0.

[1626] Specific examples

[1627] Step 1:

[1628] The user starts an application to hold a remote conference at a cafe and enables noise canceling.

[1629] Step 2:

[1630] The device captures the noise of the cafe and the user's voice and sends it to the server.

[1631] Step 3:

[1632] The server receives the audio data, identifies the user's voice with a large language model, and removes noise.

[1633] Step 4:

[1634] The server transmits the filtered audio to the terminal.

[1635] Step 5:

[1636] The terminal then plays the received clear audio to the user in real time.

[1637] This significantly reduces noise in the cafe, allowing conversations to proceed smoothly, allowing users to hold meetings efficiently.

[1638] Example 1

[1639] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1640] In conventional online meetings and noisy environments, voice communication is often hindered by ambient noise and unwanted sounds. This makes it difficult to achieve clear, crisp voice communication. In particular, in remote meetings and conference calls, if the speaker's voice cannot be clearly heard, the quality of communication declines and work efficiency also deteriorates. The present invention aims to solve these problems and provide a system that effectively clears voices even in noisy environments.

[1641] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1642] In this invention, the server includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying a conversation partner's voice using a generative AI model for analyzing the preprocessed voice data, means for filtering ambient noise and unwanted voices from the identified voice data, and means for transmitting the filtered voice data to the user terminal, thereby enabling the provision of clear, noise-free voice in real time.

[1643] "Audio data" refers to data that represents the user's voice and environmental sounds in digital form.

[1644] "Terminal" refers to a device that acquires voice data and communicates with the server. Examples include PCs, smartphones, and tablets.

[1645] "Server" is a central system for receiving, pre-processing, analyzing, filtering, and transmitting audio data.

[1646] "Preprocessing" refers to processing for applying a noise reduction filter to the received audio data and standardizing the sampling rate.

[1647] A "generative AI model" is a type of artificial intelligence model that uses a large-scale language model to analyze voice data, identify the voice of the conversation partner, and distinguish between unwanted voices and noise.

[1648] "Filtering" refers to processing for removing ambient noise and unwanted sounds from identified audio data and for emphasizing important audio components.

[1649] "Playback" refers to the process of making the filtered audio data audible to the user through the device's speakers or headphones.

[1650] "Settings" are options that allow users to customize the system's behavior, such as noise cancellation or emphasis on specific sounds.

[1651] The present invention is a system for realizing clear voice communication in online conferences and in noisy environments. This system operates through cooperation between a server, a terminal, and a user.

[1652] The server plays a central role in receiving voice data sent from the user's device and analyzing it using a large-scale language model (generative AI model). The server has the following functions:

[1653] 1. Receiving audio data

[1654] The server receives the voice data from the user's device. The received voice data is temporarily stored in the server's memory. For example, the data can be received using open source voice communication software.

[1655] 2. Preprocessing of audio data

[1656] The server preprocesses the received audio data, applying noise reduction filters and standardizing the sampling rate. This process uses tools such as LibROSA, a Python audio processing library, to reduce noise.

[1657] 3. Analysis using large-scale language models

[1658] The server then analyzes the preprocessed audio data using a generative AI model, which identifies the voice of the person speaking and distinguishes it from background noise and unwanted sounds. At this stage, the LLM can be, for example, GPT-4.

[1659] 4. Audio Filtering

[1660] The server filters the identified audio data to emphasize important audio components, and then re-encodes the filtered audio data into a format that can be transmitted efficiently. Specifically, the audio data can be compressed using the OPUS codec.

[1661] 5. Sending filtered audio data

[1662] The server sends the filtered audio data to the user's device, returning it in real time so the user receives clear audio.

[1663] The device's role is to acquire audio data, send it to the server, receive filtered audio from the server, and play it back. Specific functions include:

[1664] 1. Acquiring audio data

[1665] The device acquires audio data from a built-in or external microphone, for example, a USB microphone or a built-in microphone to capture the user's voice and environmental sounds.

[1666] 2. Sending audio data

[1667] The acquired voice data is sent from the terminal to the server, where it is divided into packets and sent to the server using the TCP / IP protocol.

[1668] 3. Receiving filtered audio

[1669] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[1670] 4. Playing back received audio

[1671] The reconstructed audio data is played back by the device through speakers or headphones, specifically via the device's audio driver.

[1672] Users can interact with the system and adjust settings to take advantage of noise cancellation and specific voice enhancement, including:

[1673] 1. Booting the system

[1674] The user launches the noise cancelling application on their device and clicks on the application icon to open the software.

[1675] 2. Adjust your settings

[1676] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[1677] 3. Start a conversation

[1678] Users can start an online meeting or call and enjoy the noise-canceling effect in real time while continuing the conversation.

[1679] Specific examples

[1680] For example, if users are having a remote meeting at a coffee shop, the process would be as follows:

[1681] 1. The user launches the application and enables noise cancellation.

[1682] 2. The device captures the noise of the cafe and the user's voice from the microphone and sends them to the server.

[1683] 3. The server receives the audio data and applies a noise reduction filter as preprocessing.

[1684] 4. The server uses a generative AI model (e.g., GPT-4) to identify the user's voice and remove noise.

[1685] 5. The server sends the filtered audio to the device.

[1686] 6. The device reconstructs the received audio data and plays it over speakers or headphones.

[1687] This significantly reduces the noise in the cafe, allowing for clearer audio during remote meetings. An example prompt is, "Design a system to efficiently cancel noise and make the user's voice clear in a situation where a remote meeting is being held in a cafe. Explain how the user's device captures audio data and communicates it with the server. Also, explain how the server processes the audio data and sends the filtered data back to the device."

[1688] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1689] Step 1: Acquire audio data (device)

[1690] The device uses a built-in or external microphone to capture voice data. This process captures the user's voice and environmental sounds. For example, when a user speaks in a cafe, the device's microphone picks up the user's voice and background noise at the same time. The input is an analog voice signal, and the output is digital voice data.

[1691] Step 2: Sending audio data (device)

[1692] The device sends the captured audio data to the server via the network. During this process, the audio data is divided into packets and transmitted using the TCP / IP protocol. The specific operations include packetizing the data and transmitting it over the network. The input is the digital audio data, and the output is the transmitted data packets.

[1693] Step 3: Receiving audio data (server)

[1694] The server receives voice data sent from the terminal via the network. The received data is temporarily stored in the server's memory. Specific operations include receiving data via the network interface and storing it in memory. The input is the transmitted data packet, and the output is the digital voice data stored in memory.

[1695] Step 4: Preprocessing the audio data (server)

[1696] The server applies a noise reduction filter to the received audio data and standardizes the sampling rate. Specifically, noise reduction is performed using LibROSA, a Python audio processing library. The input is digital audio data stored in memory, and the output is audio data with reduced noise and a standardized sampling rate.

[1697] Step 5: Analysis using large-scale language models (server)

[1698] The server analyzes the preprocessed audio data using a generative AI model. LLM distinguishes between the voice of the conversation partner and background noise from the audio data. In this process, GPT-4 can be used as the model. The input is the preprocessed audio data, and the output is the identified audio components.

[1699] Step 6: Audio Filtering (Server)

[1700] The server uses LLM to filter the identified audio data and emphasize important audio components. This filtering process applies multiple filters to emphasize specific frequency bands. The input is the identified audio components, and the output is clear filtered audio data.

[1701] Step 7: Sending the filtered audio data (server)

[1702] The server sends the filtered audio data to the device. The audio data is efficiently compressed and transferred in real time. Specifically, the OPUS codec is used to compress the data and send it over the network. The input is the filtered audio data, and the output is compressed audio data packets.

[1703] Step 8: Receive filtered audio (terminal)

[1704] The terminal receives the filtered voice data sent from the server and reconstructs the received data packets into a single voice data packet. Specific operations include receiving packets from the network and reconstructing them. The input is the compressed voice data packet, and the output is the reconstructed digital voice data.

[1705] Step 9: Playing back received audio (device)

[1706] The reconstructed audio data is then played back through the device's audio driver on speakers or headphones, with the volume and sound quality adjusted appropriately to provide clear audio to the user. The input is the reconstructed digital audio data, and the output is audio that the user can hear.

[1707] Step 10: User configuration operations (User)

[1708] A user uses a noise-canceling application to adjust settings, including whether to emphasize specific audio components and adjusting the level of noise cancellation. The input is the user's actions and settings, and the output is the customized system settings.

[1709] (Application example 1)

[1710] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1711] With conventional factory robots, voice communication between workers and robots is often hindered by loud noise in the factory, posing problems that affect work efficiency and safety. Furthermore, the lack of appropriate noise filtering methods increases the risk of workers misunderstanding instructions and causing operational errors. Therefore, there is a need for a method that enables clear voice communication even in noisy environments, allowing robots to work safely and efficiently.

[1712] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1713] In this invention, the server includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying the voice of a conversation partner using a large-scale language model for analyzing the preprocessed voice data, means for filtering ambient noise and unwanted voices from the identified voice data, means for transmitting the filtered voice data to the user terminal, and means for realizing voice communication between a worker and a robot in a noisy factory environment, thereby enabling clear voice communication between the worker and the robot even in a noisy factory environment.

[1714] "Voice data" means electronically recorded sound information, particularly data intended for communication by human voice.

[1715] A "user terminal" is a device for transmitting and receiving voice data, and is equipped with a microphone and a speaker.

[1716] "Preprocessing" refers to the initial data processing operations performed on received audio data, specifically noise reduction and standardization of the sampling rate.

[1717] A "large-scale language model" is a machine learning model that learns from large amounts of data for a specific task and analyzes or generates natural language.

[1718] "Interviewer voice identification" refers to the process of detecting and recognizing a specific person's voice from audio data.

[1719] "Noise" refers to background sounds and interfering sounds other than the target voice, and is a factor that reduces the quality of voice communication.

[1720] "Filtering" refers to the process of removing unnecessary components from audio data and extracting and emphasizing necessary parts.

[1721] "Noise environment within a factory" refers to a workplace environment where there is a lot of machinery and other work noise, and where noise levels are generally high.

[1722] "Workers" refers to personnel who operate machinery and equipment within a factory.

[1723] A "robot" is a programmable mechanical device capable of performing specific tasks automatically.

[1724] "Voice communication" refers to the transmission of information between people or machines via voice.

[1725] "Misoperation" refers to an inappropriate response by a robot to unintended operations or instructions, which has a negative impact on production efficiency and safety.

[1726] This invention is a system that realizes clear voice communication between a worker and a robot in a noisy factory environment. This system operates through cooperation between a server, a terminal, and a robot.

[1727] Server Roles

[1728] The server is responsible for the central processing of voice data. Specifically, it has the following functions:

[1729] 1. Receiving audio data

[1730] The server receives the voice data transmitted from the worker's terminal and temporarily stores the received voice data in the server's memory.

[1731] 2. Preprocessing of audio data

[1732] The server applies a noise reduction filter to the received audio data and normalizes the sampling rate, improving the quality of the data.

[1733] 3. Analysis using large-scale language models

[1734] The server uses the pre-processed speech data to analyze it with a large-scale language model to identify the worker's voice, and specific voice instructions are extracted from the analysis.

[1735] 4. Audio Filtering

[1736] The server filters out ambient noise from the identified audio data and emphasizes the important audio components. The filtered audio data is then re-encoded and converted into a format that can be transmitted efficiently.

[1737] 5. Sending filtered audio data

[1738] The server sends the filtered audio data to the robot, which then initiates actions based on it.

[1739] Device Role

[1740] The terminal is responsible for receiving and transmitting audio data. Specifically, it has the following functions:

[1741] 1. Acquiring audio data

[1742] The device acquires audio data from a built-in or external microphone, which is captured via a microphone driver.

[1743] 2. Sending audio data

[1744] The acquired voice data is sent from the device to the server, where it is divided into packets and transmitted using a network protocol.

[1745] 3. Receiving and playing filtered audio

[1746] The terminal receives the filtered voice data sent from the server and plays it through a speaker or headphones, allowing the worker to hear clear audio.

[1747] The role of robots

[1748] The robot operates based on filtered voice data from the server, and has the following functions:

[1749] 1. Analysis of audio data

[1750] The robot analyzes the filtered voice data sent from the server and begins to act as instructed.

[1751] 2. Real-time response

[1752] The robot responds to the worker's instructions in real time and carries out tasks efficiently.

[1753] Specific examples

[1754] For example, when a worker in a factory commands a robot to "start the next task," the flow is as follows:

[1755] 1. The worker captures the voice with the microphone on the terminal and sends it to the server.

[1756] 2. The server analyzes the received voice data and identifies the specified instructions.

[1757] 3. The server sends the filtered audio data to the robot.

[1758] 4. The robot follows the instructions and begins the next task.

[1759] Prompt Sentence Examples

[1760] "Design a voice instruction analysis system for a noisy factory environment. The system must filter out the noise in the factory and clearly convey the operator's instructions to the robot."

[1761] In this way, the system of the present invention enables clear voice communication between workers and robots even in noisy factory environments.

[1762] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1763] Step 1:

[1764] Acquiring audio data

[1765] The terminal uses a built-in or external microphone to acquire the worker's voice data. The input is a voice signal, and the output is digital voice data. This data is captured via a microphone driver.

[1766] Step 2:

[1767] Sending audio data

[1768] The device captures audio data and sends it to the server. The input is digital audio data, and the output is packet data sent to the server. The audio data is divided into packets and sent to the server using a network protocol.

[1769] Step 3:

[1770] Receiving audio data

[1771] The server receives the voice data sent from the terminal. The input is packet data, and the output is digital voice data temporarily stored in memory.

[1772] Step 4:

[1773] Audio data preprocessing

[1774] The server applies a noise reduction filter to the received audio data and normalizes the sampling rate. The input is digital audio data and the output is pre-processed digital audio data. Software filtering techniques are used for noise reduction and the sampling rate is changed to a normalized value.

[1775] Step 5:

[1776] Analysis using large-scale language models

[1777] The server analyzes the preprocessed voice data using a large-scale language model. The input is the preprocessed digital voice data, and the output is the identified voice instruction data. The large-scale language model analyzes the voice characteristics and extracts specific voice instructions.

[1778] Step 6:

[1779] Audio Filtering

[1780] The server removes ambient noise from the identified voice data and emphasizes the important voice components. The input is the identified voice instruction data, and the output is the filtered digital voice data. The filtering uses automatic noise reduction technology to emphasize only the necessary parts.

[1781] Step 7:

[1782] Sending filtered audio data

[1783] The server sends the filtered audio data to the robot. The input is filtered digital audio data, and the output is packets of data sent to the robot. This data is encoded and converted into a format that can be transmitted efficiently.

[1784] Step 8:

[1785] Analyzes voice data and starts operation

[1786] The robot analyzes the filtered voice data sent from the server and starts to act based on the instructions. The input is filtered digital voice data, and the output is the robot's specific action. The robot processes the received voice data and performs the appropriate action.

[1787] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1788] This invention is a system that realizes clear voice communication in online conferences and noisy environments, and also recognizes and provides feedback on the user's emotions. In addition to cooperation between the server, terminals, and users, this system also includes an emotion engine, providing more advanced functionality.

[1789] Server Roles

[1790] The server plays a central role in receiving voice data sent from the user's device and analyzing it using a large-scale language model (LLM). It also uses an emotion engine to recognize the user's emotional state and reflect it in voice filtering. Specific functions include:

[1791] 1. Receiving audio data

[1792] The server receives the voice data from the user's terminal and temporarily stores the received voice data in the server's memory.

[1793] 2. Preprocessing of audio data

[1794] The server pre-processes the received audio data, which includes applying noise reduction filters and normalizing the sampling rate.

[1795] 3. Analysis using large-scale language models

[1796] The server then analyzes the preprocessed speech data using a large-scale language model, which extracts features to identify the speech of the conversation partner and distinguish it from background noise and unwanted speech.

[1797] 4. Emotion Recognition by Emotion Engine

[1798] The server uses an emotion engine to recognize the user's emotion from the voice data. The emotion engine analyzes voice characteristics such as tone, pitch, and rate to identify the emotional state.

[1799] 5. Audio filtering and conditioning

[1800] The server filters the identified audio data, emphasizes important audio components, and adjusts the filtering and volume based on the emotion engine to provide optimal audio feedback according to the user's emotional state.

[1801] 6. Sending filtered audio data

[1802] The server encodes the filtered audio data and converts it into a format that can be transmitted efficiently, then transmits it in real time over the Internet to the user's device.

[1803] Device Role

[1804] The device is responsible for capturing audio data, sending it to the server, and receiving and playing back the filtered audio from the server. Specific functions include:

[1805] 1. Acquiring audio data

[1806] The device acquires audio data from a built-in or external microphone. The device's microphone driver captures the user's voice and environmental sounds.

[1807] 2. Sending audio data

[1808] The acquired voice data is sent from the device to the server, which divides the voice data into packets and sends them to the server using a network protocol.

[1809] 3. Receiving filtered audio

[1810] The terminal receives the filtered voice data sent from the server and reconstructs the received packets into a single voice data.

[1811] 4. Playing back received audio

[1812] The reconstructed audio data is played back by the device through speakers or headphones, and the device adjusts the volume and quality of the audio to provide clear audio to the user.

[1813] 5. Displaying Emotional Feedback

[1814] The device receives the output of the emotion engine sent from the server and provides emotion feedback to the user, allowing the user to check their own emotional state in real time.

[1815] User operations

[1816] Users interact with the system and adjust settings to enable noise cancellation, emphasis on specific voices, and emotional feedback, including:

[1817] 1. Booting the system

[1818] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[1819] 2. Adjust your settings

[1820] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[1821] 3. Start a conversation

[1822] The user starts an online meeting or call and, during the meeting, proceeds with the conversation while receiving real-time emotional feedback and the effects of Noise Cancellation 2.0.

[1823] Specific examples

[1824] For example, if a user is holding a remote meeting at a coffee shop, the following happens:

[1825] 1. The user launches the application and enables noise cancellation.

[1826] 2. The device captures the noise in the cafe and the user's voice and sends them to the server.

[1827] 3. The server receives the voice data and uses LLM to identify the user's voice and remove noise.

[1828] 4. The server uses an emotion engine to recognize the user's emotions and reflect them in filtering.

[1829] 5. The server sends the filtered audio and emotional feedback to the device.

[1830] 6. The device receives clear audio and provides emotional feedback to the user in real time.

[1831] This not only significantly reduces noise in the cafe and allows conversations to proceed smoothly, but also enables optimal voice communication that is tailored to the user's emotional state.

[1832] The processing flow will be explained below.

[1833] Server Processing Steps

[1834] Step 1:

[1835] The server receives the audio stream from the user's terminal, and the received audio data is temporarily stored in the server's memory.

[1836] Step 2:

[1837] The server pre-processes the received audio data, which may include applying noise reduction filters and normalizing the sampling rate.

[1838] Step 3:

[1839] The server inputs the preprocessed speech data into a large-scale language model (LLM), which analyzes the speech data and extracts features to identify the speech of the conversation partner.

[1840] Step 4:

[1841] The server uses the LLM analysis results to identify and filter out ambient noise and unwanted sounds, highlighting important voice components.

[1842] Step 5:

[1843] The server inputs the filtered voice data into an emotion engine, which analyzes voice characteristics such as tone, pitch, and rate to identify the user's emotional state.

[1844] Step 6:

[1845] The server fine-tunes the voice filtering settings based on the emotional information obtained from the emotion engine. For example, if the user is feeling stressed, the server further reduces ambient noise.

[1846] Step 7:

[1847] The server encodes the filtered audio data and the output of the emotion engine into a format that can be transmitted efficiently.

[1848] Step 8:

[1849] The server transmits the encoded filtered voice data and emotional feedback to the user's device in real time via the Internet.

[1850] Terminal processing steps

[1851] Step 1:

[1852] The device acquires audio data from a built-in or external microphone. The device's microphone driver captures the user's voice and environmental sounds.

[1853] Step 2:

[1854] The terminal divides the acquired voice data into packets and transmits them to the server using a network protocol.

[1855] Step 3:

[1856] The device receives the filtered voice data and emotion feedback sent from the server, and reconstructs the received packets into a single voice data.

[1857] Step 4:

[1858] The device plays the reconstructed audio data in real time over speakers or headphones, adjusting the volume and quality of the audio to provide the user with clear audio.

[1859] Step 5:

[1860] The device displays the received emotional feedback to the user, allowing the user to check their own emotional state in real time.

[1861] User operation steps

[1862] Step 1:

[1863] The user launches the Noise Cancelling 2.0 application on their device and clicks on the application icon to open the software.

[1864] Step 2:

[1865] Users adjust options within the app's settings screen, choosing whether to emphasize specific voices, selecting noise cancellation levels and other customization settings.

[1866] Step 3:

[1867] The user starts an online meeting or call and, during the meeting, proceeds with the conversation while receiving real-time emotional feedback and the effects of Noise Cancellation 2.0.

[1868] Specific examples

[1869] Step 1:

[1870] The user starts an application to hold a remote conference at a cafe and enables noise canceling.

[1871] Step 2:

[1872] The device captures the noise of the cafe and the user's voice and sends it to the server.

[1873] Step 3:

[1874] The server receives the voice data, identifies the user's voice using LLM, and removes noise.

[1875] Step 4:

[1876] The server runs the filtered voice data through an emotion engine to analyze the user's emotional state.

[1877] Step 5:

[1878] The server fine-tunes the voice filtering based on the results of the emotion engine and generates optimal voice data.

[1879] Step 6:

[1880] The server transmits the filtered voice data and the emotional feedback to the terminal.

[1881] Step 7:

[1882] The device provides the user with clear received audio and emotional feedback in real time.

[1883] This not only significantly reduces noise in the cafe and allows conversations to proceed smoothly, but also enables optimal voice communication that is tailored to the user's emotional state.

[1884] Example 2

[1885] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1886] In conventional online conference and call systems, clear communication is difficult due to poor audio quality in noisy environments. Furthermore, the lack of a function to recognize and provide feedback on the user's emotional state in real time hinders smooth communication.

[1887] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1888] In this invention, the server includes means for receiving voice data from a user terminal, means for preprocessing the voice data, means for identifying the voice of a conversation partner using a large-scale language model for analyzing the preprocessed voice data, means for using an emotion engine for recognizing emotions from the voice data, means for adjusting the voice data based on the emotional state, and means for transmitting the filtered and adjusted voice data to the user terminal, thereby enabling clear voice communication even in noisy environments and further enabling smooth communication by providing feedback on the user's emotional state in real time.

[1889] "Audio data" refers to audio information acquired from and transmitted by a user's terminal.

[1890] "Preprocessing" refers to processing to improve the quality of audio data, such as applying a noise reduction filter or standardizing the sampling rate.

[1891] A "large-scale language model" is an advanced machine learning model that analyzes text and audio data to understand the context and meaning of conversations.

[1892] An "emotion engine" is a system that analyzes voice characteristics such as tone, pitch, and speed to identify a speaker's emotional state.

[1893] "Filtering" refers to the process of removing unnecessary noise and interference sounds from audio data and emphasizing important audio components.

[1894] "Terminal" refers to a device used to capture, send, receive, and play audio data.

[1895] "Encoding" refers to the process of converting audio data into a format that can be transmitted efficiently.

[1896] "Real-time" refers to immediate processing and feedback.

[1897] "Emotional feedback" refers to informing users of their emotional state and providing information to improve the quality of communication.

[1898] "Settings" are adjustments that a user makes to adjust the system's behavior or options.

[1899] This invention is a system that realizes clear voice communication in online conferences and noisy environments, and also recognizes and provides feedback on the user's emotions. In addition to cooperation between the server, terminals, and users, this system also includes an emotion engine, providing more advanced functionality.

[1900] Server Roles

[1901] The server plays a central role in receiving, analyzing, and providing feedback on voice data sent from the user's device. Specific hardware used includes a high-performance CPU, memory, and network interface. Software includes a noise reduction filter (e.g., RNNoise), a large-scale language model (e.g., OpenAI GPT-3), and an emotion engine (e.g., IBM Watson Tone Analyzer). These software and hardware components are used to perform the following processes:

[1902] 1. Receiving audio data

[1903] The server receives the voice data sent from the user's terminal and temporarily stores it in memory.

[1904] 2. Preprocessing of audio data

[1905] The server applies noise reduction filters to clean up the audio data and also standardizes the sampling rate to equalize the quality of the data.

[1906] 3. Analysis using large-scale language models

[1907] The server analyzes the preprocessed speech data using a large-scale language model to extract features to identify the speech of the conversation partner and distinguish it from ambient noise and unwanted speech.

[1908] 4. Emotion Recognition by Emotion Engine

[1909] The server uses an emotion engine to recognize the user's emotion from the voice data, which analyzes voice characteristics such as tone, pitch, and rate to identify the emotional state.

[1910] 5. Filtering and conditioning of audio data

[1911] The server emphasizes the identified important audio components and adjusts the volume and texture based on the output of the emotion engine.

[1912] 6. Encoding and transmitting filtered audio data

[1913] The server converts the filtered audio data into a format that can be transmitted efficiently and transmits it to the terminal via the network.

[1914] Device Role

[1915] The device is responsible for capturing audio data, sending it to the server, and receiving and playing back the filtered audio from the server. The hardware used includes an internal or external microphone, speakers or headphones, and a network interface. The software includes an audio capture driver and playback software. It performs the following tasks:

[1916] 1. Acquiring audio data

[1917] The device captures audio data using a microphone, and the captured data is temporarily stored in the device's memory.

[1918] 2. Sending audio data

[1919] The terminal transmits the acquired voice data to the server using a network protocol.

[1920] 3. Receiving filtered audio

[1921] The terminal receives the filtered audio data sent from the server and reconstructs it.

[1922] 4. Playing back received audio

[1923] The device then plays the reconstructed audio data over speakers or headphones, providing the user with clear audio.

[1924] 5. Displaying Emotional Feedback

[1925] The terminal receives the output of the emotion engine sent from the server and displays emotion feedback to the user in real time.

[1926] User operations

[1927] The user operates the system and configures the appropriate settings, including noise cancellation, voice emphasis, and whether or not emotional feedback is enabled. The user then launches the Noise Cancellation 2.0 application on the device, adjusts the settings, and starts an online meeting or call.

[1928] Specific examples

[1929] For example, if a user is holding a remote meeting at a coffee shop, the following happens:

[1930] 1. The user launches the Noise Cancelling 2.0 app and enables noise cancellation and voice enhancement.

[1931] 2. The device captures the noise of the cafe and the user's voice through its built-in microphone and sends the audio data to the server.

[1932] 3. The server receives the audio data and preprocesses it with a noise reduction filter.

[1933] 4. The server uses a large-scale language model to distinguish between the user's voice and noise.

[1934] 5. The server uses an emotion engine to identify the user's emotion from the voice data and perform filtering and adjustment.

[1935] 6. The server sends the filtered audio to the device.

[1936] 7. The device receives and plays the filtered audio data.

[1937] 8. The device displays emotional feedback to the user in real time.

[1938] Prompt Sentence Examples

[1939] Explain how Noise Cancellation 2.0 can be used to ensure clear voice communication and provide real-time feedback on user emotions during a remote meeting in a coffee shop.

[1940] As described above, the system of the present invention enables clear voice communication even in noisy environments, and furthermore, by recognizing and feeding back the user's emotions in real time, smooth communication can be realized.

[1941] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1942] Step 1:

[1943] Acquiring audio data

[1944] The device acquires audio data using a built-in or external microphone. The input is an audio signal from the microphone, and the output is digital audio data. The microphone driver starts up and captures the user's voice and environmental sounds in real time. This data is temporarily stored in the device's memory.

[1945] Step 2:

[1946] Sending audio data

[1947] The device sends the captured audio data to the server using a network protocol (e.g. TCP / IP). The input is digital audio data, which is converted into packets that are sent to the server as output. The audio data is divided into small packets and sent over the internet to the server.

[1948] Step 3:

[1949] Receiving voice data and storing it in memory

[1950] The server receives the voice data sent from the terminal and temporarily stores it in memory. The input is data packets, and the output is continuous digital voice data. The receiving module combines the data packets and reconstructs them into a single voice data.

[1951] Step 4:

[1952] Audio data preprocessing

[1953] The server applies a noise reduction filter (e.g., RNNoise) to clean up the audio data. It takes continuous digital audio data as input and produces denoised audio data as output. Noise is removed sample by sample and the sampling rate is normalized.

[1954] Step 5:

[1955] Analysis using large-scale language models

[1956] The server analyzes the preprocessed voice data using a large-scale language model (e.g., OpenAI GPT-3). The input is noise-removed voice data, and the output is the voice identity of the conversation partner. The language model performs contextual analysis and extracts features to distinguish the conversation partner's voice from noise and unwanted sounds.

[1957] Step 6:

[1958] Emotion recognition by emotion engine

[1959] The server uses an emotion engine (e.g., IBM Watson Tone Analyzer) to recognize the user's emotion from the voice data. Feature data is input, and emotional state information is obtained as output. The tone, pitch, and rate of the voice are analyzed to identify the emotional state.

[1960] Step 7:

[1961] Filtering and conditioning audio data

[1962] The server emphasizes the identified voice components and adjusts the volume and texture based on the output of the emotion engine. The inputs are emotional state information and voice identification information, and the output is adjusted voice data. The voice filter emphasizes important voice components and optimizes the volume and tone according to the emotion.

[1963] Step 8:

[1964] Encoding and transmitting filtered audio data

[1965] The server converts the filtered audio data into a format that can be transmitted efficiently (e.g., Opus codec) and sends it over the network to the device. The input is the conditioned audio data, and the output is the encoded audio data. The data is split into packets and transmitted again.

[1966] Step 9:

[1967] Receive and play filtered audio

[1968] The device receives the filtered audio data sent from the server and reconstructs it. The input is the encoded audio data, and the output is the reconstructed audio data. The reconstructed audio data is then played back through speakers or headphones, providing the user with clear audio.

[1969] Step 10:

[1970] Displaying Emotional Feedback

[1971] The device receives the output of the emotion engine sent from the server and displays emotional feedback to the user in real time. Emotional state information is input, and emotional feedback display is obtained as output. This allows the user to check their own emotional state in real time.

[1972] (Application example 2)

[1973] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1974] In recent years, with the increase in online meetings and remote work, the quality of voice communication in noisy environments has become an issue. Maintaining clear voice communication in noisy environments is also difficult during field work such as delivery work. Furthermore, there are currently limited methods for understanding delivery workers' emotions and stress levels in real time and providing appropriate feedback.

[1975] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for receiving voice data from a user's information device, means for preprocessing the voice data, means for identifying the voice of the conversation partner using a large-scale language model for analyzing the preprocessed voice data, means for filtering environmental noise and unnecessary voice from the identified voice data, means for transmitting the filtered voice data to the user's information device, means for using an emotion engine for recognizing the user's emotion from the voice data, means for adjusting the voice data according to the user's emotional state, and means for displaying the adjusted voice data and emotion feedback on the user's information device. This not only enables clear voice communication even in noisy environments, but also makes it possible to grasp the user's emotional state in real time and provide appropriate feedback.

[1976] "Audio Data" means a digital recording of sound, typically captured using an audio input device such as a microphone.

[1977] "User's information equipment" refers to electronic devices used by users, such as smartphones, tablets, and laptops.

[1978] "Preprocessing" refers to performing processes such as noise reduction and standardization of the sampling rate on the acquired audio data.

[1979] A "large-scale language model" refers to a natural language processing model trained on a large amount of data, which enables the analysis and semantic understanding of speech data.

[1980] "Means for identifying the voice of a conversation partner" refers to a method or algorithm for extracting the voice of a specific speaker from multiple voice data.

[1981] "Environmental noise" refers to background noise or unwanted sounds that can be present during a conversation, such as traffic or wind noise.

[1982] "Filtering" refers to a process for emphasizing specific audio components and removing unwanted components.

[1983] "Emotion engine" refers to algorithms and technologies that analyze characteristics of voice data, such as tone, pitch, and speed, to identify a user's emotional state.

[1984] "Emotional feedback" refers to information that analyzes a user's emotional state and provides the results to the user in real time.

[1985] This invention is a system that improves the quality of voice communication in food delivery and simultaneously grasps the emotional state of delivery personnel in real time. The specific configuration and processing method of the system are described below.

[1986] Server Roles

[1987] The server plays a central role in receiving voice data from the user's information device (e.g., a smartphone) and recognizing emotions using an emotion engine. Specific processes include preprocessing the voice data, analyzing it using a large-scale language model, filtering environmental noise, and transmitting the voice data and emotional feedback.

[1988] 1. Receiving audio data:

[1989] The server receives the voice data sent from the delivery person's smartphone. At this stage, the voice data is received in packet format using a network protocol.

[1990] 2. Preprocessing the audio data:

[1991] The server processes the received audio data with a noise reduction filter and normalizes the sampling rate. For data processing, it uses an audio processing library (e.g., librosa).

[1992] 3. Large-scale Language Model (LLM) Analysis:

[1993] The preprocessed speech data is then analyzed using a large-scale language model, which extracts features from the speech data and identifies the speech of the conversation partner. The model uses a generative AI model (e.g., GPT-based).

[1994] 4. Emotion Recognition with Emotion Engine:

[1995] The server uses an emotion engine to recognize the delivery person's emotions from the voice data, which includes algorithms that analyze voice characteristics such as tone, pitch, and speed to identify emotional states.

[1996] 5. Audio filtering and conditioning:

[1997] The system removes environmental noise and unwanted sounds from the identified audio data to generate clear audio data, and also adjusts the audio data (e.g., volume) based on the results of emotion recognition.

[1998] 6. Sending filtered audio data and emotional feedback:

[1999] The server re-encodes the filtered audio data and transmits it to the user's information device in real time, and simultaneously transmits the emotional feedback and displays it on the information device.

[2000] Device Role

[2001] The terminal (i.e., the delivery person's smartphone) acquires, sends, and receives voice data, plays back filtered voice, and displays emotional feedback.

[2002] 1. Acquire audio data:

[2003] The device captures audio data from the built-in microphone and uses the device's microphone driver for audio input.

[2004] 2. Sending audio data:

[2005] The acquired voice data is divided into packets and sent to a server using a network protocol.

[2006] 3. Receiving and playing filtered audio:

[2007] It receives the filtered audio data sent from the server, reconstructs it, and plays it over speakers or headphones.

[2008] 4. Displaying Emotional Feedback:

[2009] Emotional feedback is displayed on the delivery person's smartphone, allowing them to check their own emotional state in real time.

[2010] User operations

[2011] The user (i.e., delivery person) activates the system, adjusts settings, and navigates the system through a process of capturing and verifying voice data.

[2012] 1. Boot the system:

[2013] The user launches the noise canceling application on the smartphone.

[2014] 2. Adjust the settings:

[2015] Users can adjust noise cancellation levels, enhance specific sounds, and other customization settings in the app's settings screen.

[2016] 3. Start a conversation:

[2017] Users can use the phone and voice guidance during delivery tasks, and the system provides real-time noise cancellation and emotional feedback during conversations, ensuring clear communication even in noisy environments.

[2018] Examples of specific examples and prompts

[2019] For example, when a delivery person is delivering a pizza, the system can provide clear voice communication even in a noisy traffic environment.The system also monitors the delivery person's emotional state in real time, and if the delivery person is under high stress, the support center will provide appropriate feedback.

[2020] Example prompt sentence:

[2021] Describe a smartphone app that maintains clear voice communication even when the user is in a noisy environment, and recognizes and provides feedback on the user's emotional state. The app performs voice data acquisition, noise reduction, emotion recognition, voice filtering, and emotional feedback display.

[2022] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[2023] Step 1:

[2024] The user launches the application on their smartphone, which displays the initial application screen and allows them to adjust the noise cancellation level and other settings.

[2025] Step 2:

[2026] The device (smartphone) acquires voice data using a built-in microphone. Specifically, it captures the user's voice and surrounding environmental sounds (noise) and saves them as digital voice data. The acquired voice data is temporarily stored in the device's memory.

[2027] Step 3:

[2028] The terminal sends the acquired voice data to the server. The voice data is divided into packets and sent to the server using a network protocol (e.g., TCP / IP). The input is the raw voice data, and the output is the voice packets that arrive at the server.

[2029] Step 4:

[2030] The server receives the audio data sent from the device and preprocesses it by applying a noise reduction filter, which removes environmental noise and unwanted sounds in the audio signal. The input is raw audio data, and the output is preprocessed audio data with noise removed.

[2031] Step 5:

[2032] The server analyzes the preprocessed speech data using a large-scale language model (LLM) to identify the speech of the conversation partner. This analysis involves extracting and classifying speech features. The input is the preprocessed speech data, and the output is the identified speech segments of the conversation partner.

[2033] Step 6:

[2034] The server filters out environmental noise and unwanted sounds from the identified audio data by applying a filtering algorithm to highlight only the audio components of interest. This process results in clear audio data. The input is the identified audio data, and the output is the filtered, clear audio data.

[2035] Step 7:

[2036] The server uses an emotion engine to recognize the user's emotion from the voice data. It analyzes acoustic characteristics such as tone, pitch, and rate of the voice to identify the user's emotional state. The input is the filtered voice data, and the output is data indicating the user's emotional state.

[2037] Step 8:

[2038] The server adjusts the audio data according to the user's emotional state and generates appropriate audio feedback. For example, if the user is feeling stressed, it adjusts the volume or applies a specific filter. The input is the emotional state data and the filtered audio data, and the output is the adjusted audio data.

[2039] Step 9:

[2040] The server encodes the filtered and adjusted voice data and sends it to the user's device, and simultaneously sends emotional feedback to be displayed on the device. The input is the adjusted voice data and emotional feedback data, and the output is the data that arrives at the user's device.

[2041] Step 10:

[2042] The terminal reconstructs the received filtered and adjusted audio data and plays it over speakers or headphones. The input is the audio packets received from the server, and the output is the clear audio being played.

[2043] Step 11:

[2044] The device displays emotional feedback to the user, which is displayed on the smartphone screen as an icon or message indicating the user's emotional state. The input is the user's emotional feedback data, and the output is the displayed emotional feedback.

[2045] Through specific processing steps, users can maintain clear voice communication even in noisy environments and understand their own emotional state in real time. This system is very useful in food delivery situations.

[2046] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[2047] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[2048] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[2049] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[2050] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[2051] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[2052] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[2053] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[2054] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[2055] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[2056] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[2057] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[2058] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[2059] 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.

[2060] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[2061] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[2062] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[2063] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[2064] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[2065] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[2066] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[2067] The following is further disclosed regarding the above embodiment.

[2068] (Claim 1)

[2069] means for receiving voice data from a user terminal;

[2070] means for preprocessing the audio data;

[2071] means for identifying a conversation partner's voice using a large-scale language model for analyzing the preprocessed voice data;

[2072] means for filtering ambient noise and unwanted sounds from the identified audio data;

[2073] means for transmitting the filtered audio data to a user terminal;

[2074] A system including:

[2075] (Claim 2)

[2076] means for acquiring audio data from a microphone;

[2077] means for transmitting the acquired voice data to a server;

[2078] means for playing back the filtered audio data received from the server;

[2079] 10. The system of claim 1, further comprising:

[2080] (Claim 3)

[2081] a means for the user to adjust certain settings;

[2082] means for adjusting the filtering of audio data based on said settings;

[2083] 10. The system of claim 1, further comprising:

[2084] "Example 1"

[2085] (Claim 1)

[2086] means for receiving voice data from a user terminal;

[2087] means for preprocessing the audio data;

[2088] means for identifying a conversation partner's voice using a generative AI model for analyzing the preprocessed voice data;

[2089] means for filtering ambient noise and unwanted sounds from the identified audio data;

[2090] means for transmitting the filtered audio data to a user terminal;

[2091] A system including:

[2092] (Claim 2)

[2093] means for acquiring audio data from a microphone;

[2094] means for transmitting the acquired voice data to a server;

[2095] means for playing back the filtered audio data received from the server;

[2096] 10. The system of claim 1, further comprising:

[2097] (Claim 3)

[2098] a means for the user to adjust certain settings;

[2099] means for adjusting the filtering of audio data based on said settings;

[2100] 10. The system of claim 1, further comprising:

[2101] "Application Example 1"

[2102] (Claim 1)

[2103] means for receiving voice data from a user terminal;

[2104] means for preprocessing the audio data;

[2105] means for identifying a conversation partner's voice using a large-scale language model for analyzing the preprocessed voice data;

[2106] means for filtering ambient noise and unwanted sounds from the identified audio data;

[2107] means for transmitting the filtered audio data to a user terminal;

[2108] A means for realizing voice communication between a worker and a robot in a noisy environment in a factory;

[2109] A system including:

[2110] (Claim 2)

[2111] means for acquiring audio data from a microphone;

[2112] means for transmitting the acquired voice data to a server;

[2113] means for playing back the filtered audio data received from the server;

[2114] means for analyzing voice instructions from a worker and controlling the operation of the robot;

[2115] 10. The system of claim 1, further comprising:

[2116] (Claim 3)

[2117] a means for the user to adjust certain settings;

[2118] means for adjusting the filtering of audio data based on said settings;

[2119] means for adaptively changing filtering parameters according to the work environment in the factory;

[2120] 10. The system of claim 1, further comprising:

[2121] "Example 2: Combining Emotion Engines"

[2122] (Claim 1)

[2123] means for receiving voice data from a user terminal;

[2124] means for preprocessing the audio data;

[2125] means for identifying a conversation partner's voice using a large-scale language model for analyzing the preprocessed voice data;

[2126] means for filtering ambient noise and unwanted sounds from the identified audio data;

[2127] means for using an emotion engine to recognize emotion from the audio data;

[2128] means for adjusting the audio data based on an emotional state;

[2129] means for transmitting the filtered and conditioned audio data to a user terminal;

[2130] A system including:

[2131] (Claim 2)

[2132] means for acquiring audio data from a microphone;

[2133] means for transmitting the acquired voice data to a server;

[2134] means for playing back the filtered audio data received from the server;

[2135] means for providing emotional feedback to a user;

[2136] 10. The system of claim 1, further comprising:

[2137] (Claim 3)

[2138] a means for the user to adjust certain settings;

[2139] means for filtering and adjusting audio data based on said settings;

[2140] 10. The system of claim 1, further comprising:

[2141] "Application example 2 when combining emotion engines"

[2142] (Claim 1)

[2143] means for receiving voice data from a user's information device;

[2144] means for preprocessing the audio data;

[2145] means for identifying a conversation partner's voice using a large-scale language model for analyzing the preprocessed voice data;

[2146] means for filtering environmental noise and unwanted sounds from the identified audio data;

[2147] means for transmitting the filtered audio data to a user's information device;

[2148] means for using an emotion engine to recognize a user's emotion from the voice data;

[2149] means for adjusting the voice data in response to the emotional state of the user;

[2150] means for displaying the adjusted voice data and emotional feedback on the user's information device;

[2151] A system including:

[2152] (Claim 2)

[2153] means for acquiring audio data from a microphone;

[2154] means for transmitting the acquired voice data to a server;

[2155] means for playing back the filtered audio data received from the server;

[2156] means for displaying user's emotional feedback on an information device;

[2157] 10. The system of claim 1, further comprising:

[2158] (Claim 3)

[2159] a means for the user to adjust certain settings;

[2160] means for adjusting the filtering of audio data based on said settings;

[2161] means for monitoring the user's emotional state and providing feedback according to stress level;

[2162] 10. The system of claim 1, further comprising: [Explanation of symbols]

[2163] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for receiving voice data from a user terminal; means for preprocessing the audio data; means for identifying a conversation partner's voice using a large-scale language model for analyzing the preprocessed voice data; means for filtering ambient noise and unwanted sounds from the identified audio data; means for transmitting the filtered audio data to a user terminal; A system including:

2. means for acquiring audio data from a microphone; means for transmitting the acquired voice data to a server; means for playing back the filtered audio data received from the server; The system of claim 1 further comprising:

3. a means for the user to adjust certain settings; means for adjusting the filtering of audio data based on said settings; The system of claim 1 further comprising:

Images