System

Abstract

A system is provided.SOLUTION: This system includes a means for acquiring voice through a microphone, a means for analyzing the acquired voice and detecting inappropriate utterance, a means for automatically converting the detected inappropriate utterance into gentle words, and a means for distributing the converted voice to the other party.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

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

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

[0004] In recent years, slander and defamation on social media and distribution platforms has become a social problem. In particular, there is a growing risk of inappropriate comments spreading when users speak in the heat of the moment. Such comments place a psychological burden on many people and cause a decline in the quality of communication. Therefore, there is a need for a system that can convert inappropriate comments into kinder words in real time and ensure healthy communication. [Means for solving the problem]

[0005] The present invention provides a system including a means for acquiring voice via a microphone, a means for analyzing the acquired voice and detecting inappropriate remarks, a means for automatically converting the detected inappropriate remarks into kinder words, and a means for delivering the converted voice to the other party. The system also includes a means for converting voice into text data, a means for analyzing the text data and identifying inappropriate words, a means for replacing the identified inappropriate words with kinder words, and a means for reconverting the replaced text data into voice. The system further includes a means for processing the acquired voice data in real time and a means for updating the voice delivered to the other party in real time. This allows inappropriate remarks to be instantly converted into kinder words, maintaining healthy communication.

[0006] A "microphone" is a device that converts sound into an electrical signal and inputs it into a digital device.

[0007] "Speech" is the sound wave signal that contains the human voice and is what is captured by a microphone.

[0008] "Analysis" is the process of processing captured audio as digital data to identify language and specific phrases.

[0009] "Profanity" means words or phrases that are deemed to be offensive, insulting, or objectionable.

[0010] "Detection" is the process of finding specific patterns or phrases in the analyzed audio data.

[0011] "Kind words" are more positive, gentle words or phrases chosen to replace inappropriate remarks.

[0012] "Conversion" is the process of replacing detected inappropriate language with more benign language.

[0013] "Delivery" is the process of delivering the converted audio to the recipient.

[0014] "Text data" is character string information converted from voice data.

[0015] "Speech synthesis" is the process of converting text data back into speech data.

[0016] "Real-time processing" is the process of processing data almost instantly and generating and delivering results without delay. [Brief explanation of the drawings]

[0017] [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

[0018] 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.

[0019] First, the terms used in the following description will be explained.

[0020] 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).

[0021] 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.

[0022] 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.

[0023] 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.

[0024] 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."

[0025] [First embodiment]

[0026] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0027] 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.

[0028] 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).

[0029] 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.

[0030] 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.

[0031] 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.

[0032] 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.

[0033] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0034] 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.

[0035] 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.

[0036] 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.

[0037] 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."

[0038] This invention relates to a system that detects inappropriate comments on social media and distribution platforms in real time and converts them into more gentle comments. This system is realized through a series of processes: voice input, voice analysis, detection of inappropriate comments, comment correction, and voice distribution.

[0039] Basic system configuration

[0040] 1. Voice input

[0041] The user uses a microphone to input speech. For example, consider the user saying "You're an idiot!"

[0042] The device captures this audio as a digital signal and obtains the data.

[0043] 2. Audio analysis

[0044] The terminal transmits the acquired voice data to the server.

[0045] The server uses a voice analysis engine to convert the voice data into text data.

[0046] The server analyzes this text data and identifies inappropriate words and phrases, for example, detecting "idiot" in the text "You're an idiot!" as an inappropriate word.

[0047] 3. Detecting inappropriate comments

[0048] The server flags inappropriate words and phrases identified by the speech analysis engine.

[0049] The flagged text is treated as data indicating inappropriate language.

[0050] 4. Correction of statements

[0051] The server receives the flagged text and searches a database for a more benign equivalent for the inappropriate phrase, for example, changing "stupid" to "I think that's a bit off."

[0052] The server generates the corrected text and records it as new text data.

[0053] 5. Audio Distribution

[0054] The server passes the corrected text data to a speech synthesis engine, which converts it back into speech data.

[0055] The device receives new audio data and distributes it to listeners in real time.

[0056] For example, if the user says, "You're an idiot!", the listener will hear the gentler, "I think you're a bit different!"

[0057] Specific examples

[0058] If a user says, "You're an idiot!"

[0059] 1. Voice input

[0060] The user speaks into the microphone, "You're an idiot!"

[0061] The device captures this audio and transmits it as a digital signal to the server.

[0062] 2. Audio analysis

[0063] The server uses a voice analysis engine to convert "You're an idiot!" into text data.

[0064] The server parses this text and identifies the word "idiot" as inappropriate.

[0065] 3. Detecting inappropriate comments

[0066] The server flags "idiots."

[0067] 4. Correction of statements

[0068] The server converts "idiot" to "I think that's a bit off."

[0069] The finished text is "I think you're a little different!"

[0070] 5. Audio Distribution

[0071] The server passes the corrected text to a speech synthesis engine to convert it into voice data.

[0072] The device receives new audio and distributes it to listeners.

[0073] Usage scene

[0074] This system can be applied to a wide range of situations, including live streaming and interactive social networking sites. By using this system, inappropriate comments that users may make unconsciously can be instantly converted into gentler words, which is expected to improve the quality of communication and prevent conflicts between users.

[0075] The processing flow will be explained below.

[0076] Program processing

[0077] Step 1:

[0078] The user speaks into the microphone.

[0079] Example: "You're an idiot!"

[0080] Step 2:

[0081] The device captures the user's voice with a microphone.

[0082] Step 3:

[0083] The device converts the captured audio data into a digital format and stores it in a buffer.

[0084] Step 4:

[0085] The terminal transmits the audio data stored in the buffer to the server.

[0086] Step 5:

[0087] The server passes the received voice data to the voice analysis engine.

[0088] Step 6:

[0089] The voice analysis engine converts the voice data into text (voice recognition).

[0090] Step 7:

[0091] A speech analysis engine analyzes text data and identifies inappropriate words and phrases.

[0092] Step 8:

[0093] A speech analysis engine identifies and flags inappropriate words and expressions based on specific criteria.

[0094] Example: Identifying "idiot" as an inappropriate word in the text "You're an idiot!"

[0095] Step 9:

[0096] The server receives the flagged text data.

[0097] Step 10:

[0098] The server replaces flagged inappropriate words with kinder words from a database.

[0099] Example: "Stupid" changed to "I think that's a bit different."

[0100] Step 11:

[0101] The server generates the modified text data.

[0102] For example: "You're an idiot!" turns into "I think you're a bit different!"

[0103] Step 12:

[0104] The server passes the corrected text data to the speech synthesis engine.

[0105] Step 13:

[0106] The speech synthesis engine converts the text data into speech data (speech synthesis).

[0107] Step 14:

[0108] The server transmits the generated voice data to the terminal.

[0109] Step 15:

[0110] The audio data received by the terminal is played on the speaker.

[0111] For example, if a user says, "You're an idiot!", the server converts this speech into "I think you're a little bit different!" The listener hears a gentler version of "I think you're a little bit different!" Through this series of processes, inappropriate remarks are converted into gentler words in real time, maintaining healthy communication.

[0112] Example 1

[0113] 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."

[0114] Today, many people communicate in real time on platforms such as social media and live streaming. However, inappropriate comments can be made during these interactions, causing trouble between users. A system that can avoid such situations and improve the quality of communication is needed.

[0115] 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.

[0116] In this invention, the server includes means for converting voice data into text data, means for analyzing the text data to detect inappropriate comments, and means for replacing the detected inappropriate comments with kinder words selected from a database, thereby making it possible to prevent trouble between users and improve the quality of communication.

[0117] A "microphone" is a device that converts sound into an electrical signal.

[0118] A "means for converting audio into a digital signal" is a process or device for converting analog audio into digital data.

[0119] A "means for transmitting to a server over the Internet" is a method or device for transmitting a digital signal to a remote server using a network.

[0120] A "server" is a computer system for processing and storing data.

[0121] "Means for converting into text data" refers to a process or system for converting voice data into a string of characters.

[0122] A "means for analyzing text data" is a method or device for analyzing text data and understanding or classifying its content.

[0123] An "inappropriate language detection means" is a method or apparatus for identifying inappropriate words or phrases from the analyzed text data.

[0124] A "database" is a system for storing a collection of data in an organized manner and for retrieving and updating it as needed.

[0125] A "means for replacing with kinder words" is a method or device for converting detected inappropriate expressions into appropriate expressions.

[0126] A "means for reconverting to audio data" is a process or system for converting text data into audio format.

[0127] A "terminal" is a computer device operated by a user, such as a PC or smartphone.

[0128] A "natural language processing engine" is software or a system for understanding and processing human language.

[0129] A "speech synthesis engine" is software or a system that converts input text into voice data.

[0130] The present invention relates to a system that detects inappropriate comments in real time on social media and distribution platforms and converts them into more polite language. Specific ways in which the present invention can be implemented are described below.

[0131] Audio input and digital signal conversion

[0132] Users input their voice using a microphone. This can be the built-in microphone of a PC or smartphone, or an external microphone. The device captures this voice and converts it into a digital signal. Specifically, audio capture software running on the device converts analog voice into digital data.

[0133] Sending and analyzing voice data

[0134] The device sends the captured digital voice data to a server via the Internet. An application running on the device sends the voice data to the server as a POST request via a dedicated API. The server then passes the received voice data to a voice analysis engine (e.g., Google Cloud Speech-to-Text) and converts the voice into text data.

[0135] The converted text data is then analyzed by a natural language processing (NLP) engine (e.g., spaCy or TensorFlow), which identifies inappropriate words and phrases within the text. For example, the statement "You're an idiot!" is converted to the text "You're an idiot!", with "idiot" identified as an inappropriate word.

[0136] Detecting and correcting inappropriate language

[0137] After detecting an inappropriate remark, the server searches a database (e.g., MySQL or PostgreSQL) for appropriate corrections. The detected inappropriate phrases are replaced with corresponding more benign words. For example, "idiot" can be converted to "I think you're a bit off" to generate the new text "I think you're a bit off!".

[0138] Recoding and audio distribution

[0139] The server passes the corrected text data to a speech synthesis engine (for example, Google Text-to-Speech or Amazon Polly), which reconverts it into new voice data. This new voice data is then sent back to the device and delivered to the listener. This allows, for example, a statement like "You're an idiot!" to be converted into a more gentle statement like "I think you're a little different!" and delivered.

[0140] Specific examples

[0141] Specifically, when a user says "You're an idiot!", the following process occurs:

[0142] 1. The user says into the microphone, "You're an idiot!"

[0143] 2. The device captures this audio and converts it into a digital signal.

[0144] 3. The digital signal is sent over the internet to a server.

[0145] 4. The server uses a speech analysis engine to convert the speech into text, generating the text "You're an idiot!"

[0146] 5. The server analyzes the text with an NLP engine and identifies the word "idiot" as inappropriate.

[0147] 6. The server converts "idiot" to "I think you're a little bit off" and generates the text "I think you're a little bit off!"

[0148] 7. The corrected text is converted back into voice data by the speech synthesis engine.

[0149] 8. This new audio data is sent to the device and distributed to the listener.

[0150] The above is a specific embodiment for carrying out the present invention. By using this system, inappropriate comments made by users can be instantly converted into kinder words, thereby improving the quality of communication.

[0151] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0152] Step 1:

[0153] The user uses the microphone to input speech, for example, saying "You're an idiot!"

[0154] The device converts this audio into a digital signal. The input is an analog audio signal, and the output is digital audio data. A process is performed to convert the analog audio into digital audio data. Specifically, the device's audio capture software converts the audio into a digital signal.

[0155] Step 2:

[0156] The device sends the captured digital audio data to a server over the Internet. The input is the digital audio data, and the output is the data sent to the server over the network. Specifically, an application on the device uses an API to send the audio data to the server as a POST request.

[0157] Step 3:

[0158] The server passes the received voice data to a voice analysis engine, which converts the voice into text data. The input is digital voice data, and the output is text data. A process is performed to convert the voice data into text data. Specifically, a voice analysis engine on the server (for example, Google Cloud Speech-to-Text) converts the voice data into text.

[0159] Step 4:

[0160] The server analyzes the converted text data using a natural language processing (NLP) engine to identify inappropriate words and phrases. The input is text data, and the output is text data with inappropriate words and phrases identified. The text data is analyzed and processed to identify inappropriate words and phrases. Specifically, an NLP engine (for example, spaCy) analyzes the text "You're an idiot!" and identifies "idiot" as an inappropriate word.

[0161] Step 5:

[0162] The server replaces the identified inappropriate words with kinder words from a database. The input is text data with inappropriate words flagged, and the output is the text data replaced with kinder words. Data processing is performed to correct the text data. Specifically, "idiot" is converted to "I think that's a bit off" to generate the text "I think you're a bit off!"

[0163] Step 6:

[0164] The server passes the corrected text data to a speech synthesis engine, which converts it into new speech data. The input is the corrected text data, and the output is the new speech data. A process is performed to reconvert the text data into speech data. Specifically, the speech synthesis engine (for example, Google Text-to-Speech) converts the text "I think you're a little off!" into speech data.

[0165] Step 7:

[0166] The server sends the generated new audio data to the device and distributes it to the listener in real time. The input is the new audio data, and the output is the audio received by the listener. Processing to distribute the audio data is performed. Specifically, the audio playback application on the device plays and distributes the new audio data to the listener.

[0167] (Application example 1)

[0168] 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."

[0169] In conventional social networking sites and streaming platforms, inappropriate comments can be made, causing trouble between users. However, it is difficult to detect and correct inappropriate comments in real time, which can potentially ruin the experience for streamers and viewers. Therefore, this invention aims to promote healthy communication and improve the user experience by detecting inappropriate comments in real time and converting them into appropriate words.

[0170] 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.

[0171] In this invention, the server includes means for acquiring voice through a microphone, means for analyzing the acquired voice using a voice recognition module to detect inappropriate remarks, means for automatically converting the detected inappropriate remarks into kinder words using a generative AI model, means for converting the converted text data back into voice using a voice synthesis module, means for delivering the converted voice to the other party in real time, and database means for providing prompt sentences to the generative AI model, thereby enabling the detection and correction of inappropriate remarks in real time.

[0172] A "microphone" is an input device for converting sound into an electrical signal.

[0173] A "voice recognition module" is a software or hardware component for converting captured speech into text data.

[0174] A "generative AI model" is a trained model that includes an algorithm that uses natural language processing technology to convert inappropriate remarks into kinder ones.

[0175] A "speech synthesis module" is a system component for converting text data into speech data.

[0176] A "prompt sentence" is an input sentence provided to a generative AI model to generate an appropriate output.

[0177] A "database" is a collection of data that systematically organizes information so that it can be quickly searched and used.

[0178] "Real-time" means that the processing occurs immediately, without delay.

[0179] This invention is implemented using a system including a microphone, a speech recognition module, a generative AI model, a speech synthesis module, and a database. Specifically, the invention can be implemented using the following configuration and process.

[0180] The user first inputs their voice through the microphone. This voice is picked up by the device and captured as a digital signal. The captured voice data is converted into text data using a speech recognition module and sent to the server. This speech recognition module can be commercial software such as Google's speech recognition API.

[0181] The server uses a speech analysis engine to analyze the text data generated by the speech recognition module and detect inappropriate remarks. This is done using a generative AI model. The generative AI model has the ability to detect inappropriate remarks and replace them with appropriate expressions. A specific example of this model is the "twitter-roberta-large-offensive" model based on Huggingface's transformers library. The generative AI model is linked to a database that provides prompt sentences, and this database contains pre-registered inappropriate phrases and their replacement candidates.

[0182] The text data of the detected inappropriate remarks is automatically converted into kinder words by a generative AI model. For example, the word "idiot" is converted to "I think that's a bit off." This updates the text data to new text data that uses appropriate words. This new text data is saved by the server and converted back into voice data using a voice synthesis module. A voice synthesis library such as pyttsx3 can be used as the voice synthesis module.

[0183] The server sends the voice data generated by the voice synthesis module to the terminal and distributes it to the other party in real time, allowing the listener to hear the voice with inappropriate remarks corrected.

[0184] As a specific example of its use, if a user says "You're an idiot!" during a live broadcast or game broadcast, this system will convert it into "I think you're a little different!" and broadcast it immediately. Through this process, healthy communication between the broadcaster and the viewers is maintained.

[0185] An example of input to a generative AI model is:

[0186] Input voice: You're an idiot!

[0187] Output Voice: I think you're a little different!

[0188] In this way, a system using the present invention can instantly detect and correct inappropriate comments during live broadcasts, realizing healthy communication.

[0189] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0190] Step 1:

[0191] The user inputs voice through the microphone, which is then picked up by the device and captured as a digital signal. At this stage, the voice is stored on the device as concrete digital data.

[0192] Step 2:

[0193] The device converts the captured voice data into text data using a voice recognition module. The input is digital voice data and the output is text data. In this voice recognition process, speech-to-text conversion is performed using a speech recognition API (e.g., Google's speech recognition API).

[0194] Step 3:

[0195] The server receives the text data sent from the speech recognition module and analyzes it using a speech analysis engine. In this step, the input text data is analyzed to determine whether certain words or phrases are inappropriate.

[0196] Step 4:

[0197] A generative AI model is used to convert text data in which inappropriate remarks have been detected into kinder words. The server searches a database for kinder words that correspond to the inappropriate words and updates the text data. The input here is the analyzed text data, and the output is new text data converted into kinder words. For example, Huggingface's twitter-roberta-large-offensive model is used as the generative AI model.

[0198] Step 5:

[0199] The server passes the new text data to a speech synthesis module, which converts it back into speech data. The input is the converted text data, and the output is speech data. This speech synthesis is performed using a speech synthesis library such as pyttsx3.

[0200] Step 6:

[0201] The server sends the newly generated voice data to the terminal and distributes it to the other party in real time. The input here is the voice data from the speech synthesis module, and the output is the voice data distributed to the recipient. This allows the listener to hear the voice in real time with inappropriate remarks corrected.

[0202] Through these steps, inappropriate comments made by users are converted into kinder words in real time and delivered.

[0203] 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.

[0204] This invention relates to a system that detects inappropriate comments in real time on social networking sites and distribution platforms and converts them into more appropriate and gentle language. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, the system converts them into more appropriate and gentle language, improving the quality of communication.

[0205] Basic system configuration

[0206] 1. Voice input

[0207] The user speaks into the microphone. For example, the user says, "You're an idiot!"

[0208] The device captures this audio and obtains the data as a digital signal.

[0209] 2. Audio analysis

[0210] The terminal transmits the acquired voice data to the server.

[0211] The server uses a voice analysis engine to convert the voice data into text data.

[0212] The server analyzes this text data and identifies inappropriate words and phrases, for example, detecting "idiot" in the text "You're an idiot!" as an inappropriate word.

[0213] 3. Detecting inappropriate comments

[0214] The server flags inappropriate words and phrases identified by the speech analysis engine.

[0215] The flagged text is treated as data indicating inappropriate language.

[0216] 4. Emotion recognition

[0217] The device sends the voice data to the emotion engine.

[0218] The emotion engine analyzes the user's emotions from the voice data and generates emotion data.

[0219] The emotion data analyzed by the emotion engine is used to select subsequent kind words. For example, if a user says "idiot" with an angry emotion, the emotion data includes "anger."

[0220] 5. Correction of statements

[0221] The server receives the flagged text and searches a database for benign words that correspond to the inappropriate phrases.

[0222] The server adjusts its choice of kind words based on the emotional data, for example, converting "stupid" into "I think that's a bit off," while converting "let's calm down and think about it" if the emotional data indicates "anger."

[0223] The server generates the corrected text data and records it as new text data.

[0224] 6. Audio Distribution

[0225] The server passes the corrected text data to a speech synthesis engine, which converts it back into speech data.

[0226] The device receives new audio data and distributes it to listeners in real time.

[0227] For example, if a user says "You're an idiot!" and the emotion is analyzed as "anger," the listener will hear a gentle and appropriate response: "Calm down and think about it!"

[0228] Specific examples

[0229] If the user emotionally says, "You're an idiot!"

[0230] 1. Voice input

[0231] The user speaks into the microphone, "You're an idiot!"

[0232] The device captures this audio and transmits it as a digital signal to the server.

[0233] 2. Audio analysis

[0234] The server uses a voice analysis engine to convert "You're an idiot!" into text data.

[0235] The server parses this text and identifies the word "idiot" as inappropriate.

[0236] 3. Detecting inappropriate comments

[0237] The server flags "idiots."

[0238] 4. Emotion recognition

[0239] The device sends the voice data to the emotion engine.

[0240] The emotion engine extracts the emotion of "anger" from the voice data.

[0241] 5. Correction of statements

[0242] The server converts "idiot" into "calm down and think about it."

[0243] The finished text will be "You need to calm down and think about it!"

[0244] 6. Audio Distribution

[0245] The server passes the corrected text to a speech synthesis engine to convert it into voice data.

[0246] The device receives new audio and distributes it to listeners.

[0247] Usage scene

[0248] This system can be applied to a wide range of situations, such as live streaming and interactive social networking sites. By combining it with an emotion engine, it is possible to instantly convert inappropriate comments that users may unconsciously make into appropriate and kind words, which is expected to improve the quality of communication and prevent conflicts between users.

[0249] The processing flow will be explained below.

[0250] Program processing

[0251] Step 1:

[0252] The user speaks into the microphone.

[0253] Example: "You're an idiot!"

[0254] Step 2:

[0255] The device captures the user's voice with a microphone.

[0256] Step 3:

[0257] The device converts the captured audio data into a digital format and stores it in a buffer.

[0258] Step 4:

[0259] The terminal transmits the audio data stored in the buffer to the server.

[0260] Step 5:

[0261] The server passes the received voice data to the voice analysis engine.

[0262] Step 6:

[0263] The voice analysis engine converts the voice data into text (voice recognition).

[0264] Step 7:

[0265] A speech analysis engine analyzes text data and identifies inappropriate words and phrases.

[0266] Step 8:

[0267] A speech analysis engine identifies and flags inappropriate words and expressions based on specific criteria.

[0268] Example: Identifying "idiot" as an inappropriate word in the text "You're an idiot!"

[0269] Step 9:

[0270] The server receives the flagged text data.

[0271] Step 10:

[0272] The device sends the voice data to the emotion engine.

[0273] Step 11:

[0274] The emotion engine analyzes the user's emotions from the voice data and generates emotion data.

[0275] Step 12:

[0276] The emotion engine sends the emotion analysis results to the server.

[0277] Example: The emotion engine detects "anger."

[0278] Step 13:

[0279] The server refers to the emotion data when replacing inappropriate words with kinder words.

[0280] Example: "Stupid" can be added to "I think that's a bit wrong," and if the emotion is "anger," it can be changed to "Let's calm down and think about it."

[0281] Step 14:

[0282] The server generates the modified text data.

[0283] For example: "You're an idiot!" becomes "You need to calm down and think!"

[0284] Step 15:

[0285] The server passes the corrected text data to the speech synthesis engine.

[0286] Step 16:

[0287] The speech synthesis engine converts the text data into speech data (speech synthesis).

[0288] Step 17:

[0289] The server transmits the generated voice data to the terminal.

[0290] Step 18:

[0291] The audio data received by the terminal is played on the speaker.

[0292] For example, if a user says, "You're an idiot!" and the emotion engine analyzes this as "anger," the final voice the listener hears will be a gentle and appropriate "You need to calm down and think about it!" This series of processes instantly converts inappropriate remarks into gentler words, enabling healthy communication in real time.

[0293] Example 2

[0294] 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."

[0295] On conventional social networking sites and distribution platforms, problems could arise due to inappropriate comments made by users. Simply detecting and filtering inappropriate comments is difficult to completely prevent, making it impossible to improve the quality of communication. Furthermore, simply converting words without considering the user's feelings can lead to the problem of not accurately conveying the intent of the comment.

[0296] 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.

[0297] In this invention, the server includes means for acquiring voice through a microphone, means for analyzing the acquired voice and detecting inappropriate remarks, means for automatically converting the detected inappropriate remarks into kinder words, means for analyzing the user's emotions and generating emotion data, means for adjusting the selection of kinder words based on the emotion data, and means for delivering the converted voice to the other party. This makes it possible to appropriately detect and correct the user's remarks in real time and flexibly convert words taking the user's emotions into consideration. This can prevent problems caused by inappropriate remarks and improve the quality of communication.

[0298] The "means for acquiring voice through a microphone" refers to a hardware and software device that captures the voice spoken by the user into the microphone as a digital signal and converts it into a form that can be processed within the system.

[0299] The "means for analyzing the acquired voice and detecting inappropriate remarks" refers to a device and program that uses a voice analysis engine to convert the acquired voice data into text data, and then analyzes the text data to identify inappropriate words and phrases.

[0300] "Means for automatically converting detected inappropriate language to kinder language" means devices and programs that have a database and algorithms for replacing words or phrases identified as inappropriate with kinder language.

[0301] The "means for analyzing user emotions and generating emotion data" refers to a device and program that uses an emotion recognition engine to analyze user emotions from voice data and output the analysis results as emotion data.

[0302] The "means for adjusting the selection of kind words based on emotional data" refers to a device and a program that selects appropriate kind words based on the generated emotional data and converts inappropriate words into the most appropriate words in accordance with the emotion.

[0303] The "means for delivering the converted voice to the other party" refers to a device and a program that reconverts the text data converted into friendly words into voice data and delivers the voice data to the listener in real time.

[0304] This invention relates to a system that detects inappropriate comments in real time on social media and distribution platforms and converts them into more appropriate and gentle language. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, the system converts them into more appropriate and gentle language, improving the quality of communication.

[0305] Basic system configuration

[0306] 1. Voice input

[0307] The user speaks into the microphone. For example, the user says, "You're an idiot!"

[0308] The device captures this audio and obtains the data as a digital signal.

[0309] 2. Audio analysis

[0310] The terminal transmits the acquired voice data to the server.

[0311] The server uses a speech analysis engine (e.g., Google Cloud Speech-to-Text) to convert the voice data into text data.

[0312] The server analyzes this text data and identifies inappropriate words and phrases, for example, detecting "idiot" in the text "You're an idiot!" as an inappropriate word.

[0313] 3. Detecting inappropriate comments

[0314] The server flags inappropriate words and phrases identified by the speech analysis engine.

[0315] The flagged text is treated as data indicating inappropriate language.

[0316] 4. Emotion recognition

[0317] The device sends the voice data to an emotion engine (e.g., IBM Watson Tone Analyzer).

[0318] The emotion engine analyzes the user's emotions from the voice data and generates emotion data. For example, if a user says "idiot" in anger, the emotion data will include "anger."

[0319] 5. Correction of statements

[0320] The server receives the flagged text and searches a database for benign words that correspond to the inappropriate phrases.

[0321] The server adjusts its choice of kind words based on the emotional data, for example, converting "stupid" into "I think that's a bit off," while converting "let's calm down and think about it" if the emotional data indicates "anger."

[0322] The server generates the corrected text data and records it as new text data.

[0323] 6. Audio Distribution

[0324] The server passes the corrected text data to a speech synthesis engine (e.g., Amazon Polly) and converts it back into speech data.

[0325] The device receives new audio data and delivers it to the listener in real time. For example, a voice saying "You're an idiot!" can be changed to "Calm down and think!"

[0326] Specific examples

[0327] If the user emotionally says, "You're an idiot!"

[0328] 1. Voice input

[0329] The user speaks into the microphone, "You're an idiot!"

[0330] The device captures this audio and transmits it as a digital signal to the server.

[0331] 2. Audio analysis

[0332] The server uses a speech analysis engine to convert "You're an idiot!" into text data.

[0333] The server parses this text and identifies the word "idiot" as inappropriate.

[0334] 3. Detecting inappropriate comments

[0335] The server flags "idiots."

[0336] 4. Emotion recognition

[0337] The device sends the voice data to the emotion engine.

[0338] The emotion engine extracts the emotion of "anger" from the voice data.

[0339] 5. Correction of statements

[0340] The server converts "idiot" into "calm down and think about it."

[0341] The finished text will be "You need to calm down and think about it!"

[0342] 6. Audio Distribution

[0343] The server passes the corrected text to a speech synthesis engine to convert it into voice data.

[0344] The device receives the new audio and broadcasts it to the listener, who hears the gentle words, "Stay calm and think!"

[0345] Prompt Sentence Examples

[0346] An example of a specific prompt would be:

[0347] If a user says "You're an idiot!", convert the speech data into text, detect the inappropriate phrase "idiot", analyze the emotion of anger, and then convert it into "You need to calm down and think!"

[0348] This prompt allows the system to properly analyze the user's speech content and emotions and make appropriate conversions.

[0349] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0350] Step 1: Capturing Audio Input

[0351] Input: User-generated speech

[0352] Output: Audio data as a digital signal

[0353] The device uses a microphone to capture the user's voice. For example, if the user says "You're an idiot!", the microphone captures this voice in real time. The device converts the captured voice into a digital signal and sends it to the server.

[0354] Step 2: Convert audio data to text

[0355] Input: Audio data as a digital signal

[0356] Output: Text data

[0357] The server processes the received voice data. Specifically, it uses a voice analysis engine (e.g., Google Cloud Speech-to-Text) to convert the voice data into text data. For example, the voice data "You're an idiot!" is converted into text data "You're an idiot!"

[0358] Step 3: Detect inappropriate language

[0359] Input: Text data

[0360] Output: Text data with inappropriate words flagged

[0361] The server analyzes the converted text data. Specifically, it uses a natural language processing engine to identify inappropriate words and phrases in the text. For example, the server flags the inappropriate words and phrases in the text "You're an idiot!" and passes the flagged text on to the next processing step.

[0362] Step 4: Emotion Recognition

[0363] Input: Audio data

[0364] Output: Emotion data

[0365] The device sends the collected voice data to an emotion recognition engine (e.g., IBM Watson Tone Analyzer). The emotion engine analyzes the user's emotions from the voice data and generates emotion data. For example, if a user says "idiot" out of anger, the emotion data will include "anger." The generated emotion data is sent to the server.

[0366] Step 5: Modify your statement

[0367] Input: Text and sentiment data with inappropriate words flagged

[0368] Output: Corrected text data

[0369] The server receives text data with inappropriate words flagged and emotion data. Specifically, it analyzes the text data and searches a database for appropriate kind words. For example, it converts "idiot" into "calm down and think about it." The server adjusts the choice of kind words based on the emotion data. Finally, it generates corrected text data and records it as new text data.

[0370] Step 6: Synthesize and deliver audio

[0371] Input: Modified text data

[0372] Output: New audio data

[0373] The server receives the corrected text data and passes it to a speech synthesis engine (e.g., Amazon Polly). The speech synthesis engine converts the corrected text into speech data. For example, "You're an idiot!" is converted into "Calm down and think about it!" The device receives the newly generated speech data and delivers it to the listener in real time. The listener can hear the gentle words "Calm down and think about it!"

[0374] (Application example 2)

[0375] 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."

[0376] In recent years, with the spread of live streaming and real-time communication, inappropriate comments have become a problem. Inappropriate comments can hurt people, cause trouble, and hinder healthy communication. Furthermore, due to a lack of correction based on the speaker's emotions, mechanical conversion is often insufficient. To address this issue, there is a need for a system that converts comments into appropriate words based on emotions in real time.

[0377] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0378] In this invention, the server includes means for acquiring voice through a microphone, means for analyzing the acquired voice and detecting inappropriate remarks, means for recognizing the user's emotions when an inappropriate remark is detected, means for converting the detected inappropriate remarks into kinder words based on the recognized emotions, and means for delivering the converted voice to the other party. This makes it possible to convert inappropriate remarks into appropriate words in real time, taking the user's emotions into consideration.

[0379] (definition statement)

[0380] A "microphone" is a device that converts sound into an electrical signal and is used to capture what a user says.

[0381] The "means for analyzing voice and detecting inappropriate remarks" refers to a means for analyzing acquired voice data and identifying inappropriate words and phrases therefrom.

[0382] The "means for recognizing emotions" is a means for analyzing the user's emotions from the acquired voice data and generating emotion data.

[0383] The "means for converting to kinder words" is a means for automatically replacing inappropriate remarks with kinder words, and also a means for selecting appropriate expressions based on emotional data.

[0384] The "means for reconverting to audio" is a means for converting the corrected text data into audio data.

[0385] The "means for processing in real time" refers to a means for immediately processing acquired voice data without delay.

[0386] "Means for delivering to the other party in real time" refers to means for instantly transmitting the corrected audio data to the listener or the other party.

[0387] The present invention is a system for detecting inappropriate comments in real time and converting them into friendly comments. This system can be applied when users are live streaming.

[0388] Basic system configuration

[0389] The system is implemented using the following hardware and software:

[0390] Microphone: A device that converts the user's voice into an electrical signal and captures the sound.

[0391] Terminal: A device such as a computer or smartphone that analyzes audio data acquired from a microphone.

[0392] Server: A remote server containing a speech analysis engine, emotion recognition engine, text-to-speech engine, and speech synthesis engine.

[0393] System operation explanation

[0394] 1. Audio capture:

[0395] The user speaks into a microphone and transmits the audio in real time to the device, where an application on the device captures the audio and obtains the data as a digital signal.

[0396] 2. Audio analysis:

[0397] The device sends the captured voice data to the server, which uses a speech analysis engine (e.g., Google Speech Recognition API) on the server to convert the voice data into text data. A text analysis algorithm (e.g., TextBlob) is used to identify inappropriate words and phrases.

[0398] 3. Emotion recognition:

[0399] The emotion recognition engine analyzes the user's emotions from the voice data and generates emotion data. For emotion recognition, a voice emotion recognition library, for example, is used.

[0400] 4. Statement correction:

[0401] The server receives the flagged text and searches a database for a kinder equivalent to the inappropriate phrase. Based on the emotion data, it adjusts the choice of kinder words. For example, it converts "stupid" to "calm down and think about it," but if the emotion data indicates "anger," it converts it to "think again."

[0402] 5. Speech synthesis:

[0403] The converted text data is converted back into audio data using a speech synthesis engine (e.g., Google Text-to-Speech). The device receives the new audio data and delivers it to the listener in real time.

[0404] Specific examples

[0405] For example, if a user says "You're an idiot!" during a stream, the system works as follows: This utterance is converted in real time into "Calm down and think!" and broadcast as audio to the viewer.

[0406] Example prompts to input to the generative AI model

[0407] "I'd like to develop an application that converts speech signals into text data and modifies inappropriate expressions in the text based on emotion recognition. Please generate the code to accomplish this process in real time in a voice-delivery environment using the SpeechRecognition library, Google Speech Recognition API, emotion recognition library, and Google Text-to-Speech library."

[0408] This system can instantly detect inappropriate comments made during broadcasts and appropriately change them based on emotions, enabling healthy communication.

[0409] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0410] (Program processing flow)

[0411] Step 1:

[0412] Audio Capture

[0413] During live streaming, the user speaks into the microphone. The device captures this voice and obtains it as a digital signal. The input is the user's voice signal, and the output is the captured digital voice data. In this step, the device uses the microphone to capture voice in real time.

[0414] Step 2:

[0415] Audio data transmission

[0416] The device sends the captured audio data to the server. The input is the captured digital audio data, and the output is the audio data sent to the server. In this step, the device uploads the audio data to the server.

[0417] Step 3:

[0418] Audio analysis

[0419] The server converts the acquired voice data into text data using a voice analysis engine (e.g., Google Speech Recognition API). The input is the voice data sent to the server, and the output is the converted text data. In this step, the server recognizes the voice and generates the result as text.

[0420] Step 4:

[0421] Inappropriate speech detection

[0422] The server uses a text analysis algorithm (e.g., TextBlob) to detect inappropriate content from the converted text data. The input is text data, and the output is text data with inappropriate words and phrases flagged. In this step, the server analyzes the text data to identify inappropriate content.

[0423] Step 5:

[0424] emotion recognition

[0425] The server uses an emotion recognition engine (e.g., a voice emotion recognition library) to analyze the user's emotion from the voice data. The input is voice data, and the output is emotion data. In this step, the server analyzes the emotion of the voice and determines the emotion the user expressed when speaking.

[0426] Step 6:

[0427] Correction of statement

[0428] The server converts inappropriate comments into appropriate ones based on the emotion data. The input is text data with inappropriate words and emotion data flagged, and the output is the corrected text data. In this step, the server replaces the inappropriate comments with more appropriate words based on the emotion.

[0429] Step 7:

[0430] Speech synthesis

[0431] The server converts the corrected text data back into speech data using a speech synthesis engine (e.g., Google Text-to-Speech API). The input is the corrected text data, and the output is the synthesized speech data. In this step, the server converts the text data back into speech.

[0432] Step 8:

[0433] Audio data distribution

[0434] The device receives new audio data obtained from the server and delivers it to the listener in real time. The input is the synthesized audio data, and the output is the modified audio delivered to the listener. In this step, the device delivers the new audio in real time.

[0435] 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.

[0436] 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.

[0437] 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.

[0438] [Second embodiment]

[0439] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0440] 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.

[0441] 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).

[0442] 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.

[0443] 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.

[0444] 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).

[0445] 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.

[0446] 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.

[0447] 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.

[0448] 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.

[0449] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0450] 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."

[0451] This invention relates to a system that detects inappropriate comments on social media and distribution platforms in real time and converts them into more gentle comments. This system is realized through a series of processes: voice input, voice analysis, detection of inappropriate comments, comment correction, and voice distribution.

[0452] Basic system configuration

[0453] 1. Voice input

[0454] The user uses a microphone to input speech. For example, consider the user saying "You're an idiot!"

[0455] The device captures this audio as a digital signal and obtains the data.

[0456] 2. Audio analysis

[0457] The terminal transmits the acquired voice data to the server.

[0458] The server uses a voice analysis engine to convert the voice data into text data.

[0459] The server analyzes this text data and identifies inappropriate words and phrases, for example, detecting "idiot" in the text "You're an idiot!" as an inappropriate word.

[0460] 3. Detecting inappropriate comments

[0461] The server flags inappropriate words and phrases identified by the speech analysis engine.

[0462] The flagged text is treated as data indicating inappropriate language.

[0463] 4. Correction of statements

[0464] The server receives the flagged text and searches a database for a more benign equivalent for the inappropriate phrase, for example, changing "stupid" to "I think that's a bit off."

[0465] The server generates the corrected text and records it as new text data.

[0466] 5. Audio Distribution

[0467] The server passes the corrected text data to a speech synthesis engine, which converts it back into speech data.

[0468] The device receives new audio data and distributes it to listeners in real time.

[0469] For example, if the user says, "You're an idiot!", the listener will hear the gentler, "I think you're a bit different!"

[0470] Specific examples

[0471] If a user says, "You're an idiot!"

[0472] 1. Voice input

[0473] The user speaks into the microphone, "You're an idiot!"

[0474] The device captures this audio and transmits it as a digital signal to the server.

[0475] 2. Audio analysis

[0476] The server uses a voice analysis engine to convert "You're an idiot!" into text data.

[0477] The server parses this text and identifies the word "idiot" as inappropriate.

[0478] 3. Detecting inappropriate comments

[0479] The server flags "idiots."

[0480] 4. Correction of statements

[0481] The server converts "idiot" to "I think that's a bit off."

[0482] The finished text is "I think you're a little different!"

[0483] 5. Audio Distribution

[0484] The server passes the corrected text to a speech synthesis engine to convert it into voice data.

[0485] The device receives new audio and distributes it to listeners.

[0486] Usage scene

[0487] This system can be applied to a wide range of situations, including live streaming and interactive social networking sites. By using this system, inappropriate comments that users may make unconsciously can be instantly converted into gentler words, which is expected to improve the quality of communication and prevent conflicts between users.

[0488] The processing flow will be explained below.

[0489] Program processing

[0490] Step 1:

[0491] The user speaks into the microphone.

[0492] Example: "You're an idiot!"

[0493] Step 2:

[0494] The device captures the user's voice with a microphone.

[0495] Step 3:

[0496] The device converts the captured audio data into a digital format and stores it in a buffer.

[0497] Step 4:

[0498] The terminal transmits the audio data stored in the buffer to the server.

[0499] Step 5:

[0500] The server passes the received voice data to the voice analysis engine.

[0501] Step 6:

[0502] The voice analysis engine converts the voice data into text (voice recognition).

[0503] Step 7:

[0504] A speech analysis engine analyzes text data and identifies inappropriate words and phrases.

[0505] Step 8:

[0506] A speech analysis engine identifies and flags inappropriate words and expressions based on specific criteria.

[0507] Example: Identifying "idiot" as an inappropriate word in the text "You're an idiot!"

[0508] Step 9:

[0509] The server receives the flagged text data.

[0510] Step 10:

[0511] The server replaces flagged inappropriate words with kinder words from a database.

[0512] Example: "Stupid" changed to "I think that's a bit different."

[0513] Step 11:

[0514] The server generates the modified text data.

[0515] For example: "You're an idiot!" turns into "I think you're a bit different!"

[0516] Step 12:

[0517] The server passes the corrected text data to the speech synthesis engine.

[0518] Step 13:

[0519] The speech synthesis engine converts the text data into speech data (speech synthesis).

[0520] Step 14:

[0521] The server transmits the generated voice data to the terminal.

[0522] Step 15:

[0523] The audio data received by the terminal is played on the speaker.

[0524] For example, if a user says, "You're an idiot!", the server converts this speech into "I think you're a little bit different!" The listener hears a gentler version of "I think you're a little bit different!" Through this series of processes, inappropriate remarks are converted into gentler words in real time, maintaining healthy communication.

[0525] Example 1

[0526] 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."

[0527] Today, many people communicate in real time on platforms such as social media and live streaming. However, inappropriate comments can be made during these interactions, causing trouble between users. A system that can avoid such situations and improve the quality of communication is needed.

[0528] 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.

[0529] In this invention, the server includes means for converting voice data into text data, means for analyzing the text data to detect inappropriate comments, and means for replacing the detected inappropriate comments with kinder words selected from a database, thereby making it possible to prevent trouble between users and improve the quality of communication.

[0530] A "microphone" is a device that converts sound into an electrical signal.

[0531] A "means for converting audio into a digital signal" is a process or device for converting analog audio into digital data.

[0532] A "means for transmitting to a server over the Internet" is a method or device for transmitting a digital signal to a remote server using a network.

[0533] A "server" is a computer system for processing and storing data.

[0534] "Means for converting into text data" refers to a process or system for converting voice data into a string of characters.

[0535] A "means for analyzing text data" is a method or device for analyzing text data and understanding or classifying its content.

[0536] An "inappropriate language detection means" is a method or apparatus for identifying inappropriate words or phrases from the analyzed text data.

[0537] A "database" is a system for storing a collection of data in an organized manner and for retrieving and updating it as needed.

[0538] A "means for replacing with kinder words" is a method or device for converting detected inappropriate expressions into appropriate expressions.

[0539] A "means for reconverting to audio data" is a process or system for converting text data into audio format.

[0540] A "terminal" is a computer device operated by a user, such as a PC or smartphone.

[0541] A "natural language processing engine" is software or a system for understanding and processing human language.

[0542] A "speech synthesis engine" is software or a system that converts input text into voice data.

[0543] The present invention relates to a system that detects inappropriate comments in real time on social media and distribution platforms and converts them into more polite language. Specific ways in which the present invention can be implemented are described below.

[0544] Audio input and digital signal conversion

[0545] Users input their voice using a microphone. This can be the built-in microphone of a PC or smartphone, or an external microphone. The device captures this voice and converts it into a digital signal. Specifically, audio capture software running on the device converts analog voice into digital data.

[0546] Sending and analyzing voice data

[0547] The device sends the captured digital voice data to a server via the Internet. An application running on the device sends the voice data to the server as a POST request via a dedicated API. The server then passes the received voice data to a voice analysis engine (e.g., Google Cloud Speech-to-Text) and converts the voice into text data.

[0548] The converted text data is then analyzed by a natural language processing (NLP) engine (e.g., spaCy or TensorFlow), which identifies inappropriate words and phrases within the text. For example, the statement "You're an idiot!" is converted to the text "You're an idiot!", with "idiot" identified as an inappropriate word.

[0549] Detecting and correcting inappropriate language

[0550] After detecting an inappropriate remark, the server searches a database (e.g., MySQL or PostgreSQL) for appropriate corrections. The detected inappropriate phrases are replaced with corresponding more benign words. For example, "idiot" can be converted to "I think you're a bit off" to generate the new text "I think you're a bit off!".

[0551] Recoding and audio distribution

[0552] The server passes the corrected text data to a speech synthesis engine (for example, Google Text-to-Speech or Amazon Polly), which reconverts it into new voice data. This new voice data is then sent back to the device and delivered to the listener. This allows, for example, a statement like "You're an idiot!" to be converted into a more gentle statement like "I think you're a little different!" and delivered.

[0553] Specific examples

[0554] Specifically, when a user says "You're an idiot!", the following process occurs:

[0555] 1. The user says into the microphone, "You're an idiot!"

[0556] 2. The device captures this audio and converts it into a digital signal.

[0557] 3. The digital signal is sent over the internet to a server.

[0558] 4. The server uses a speech analysis engine to convert the speech into text, generating the text "You're an idiot!"

[0559] 5. The server analyzes the text with an NLP engine and identifies the word "idiot" as inappropriate.

[0560] 6. The server converts "idiot" to "I think you're a little bit off" and generates the text "I think you're a little bit off!"

[0561] 7. The corrected text is converted back into voice data by the speech synthesis engine.

[0562] 8. This new audio data is sent to the device and distributed to the listener.

[0563] The above is a specific embodiment for carrying out the present invention. By using this system, inappropriate comments made by users can be instantly converted into kinder words, thereby improving the quality of communication.

[0564] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0565] Step 1:

[0566] The user uses the microphone to input speech, for example, saying "You're an idiot!"

[0567] The device converts this audio into a digital signal. The input is an analog audio signal, and the output is digital audio data. A process is performed to convert the analog audio into digital audio data. Specifically, the device's audio capture software converts the audio into a digital signal.

[0568] Step 2:

[0569] The device sends the captured digital audio data to a server over the Internet. The input is the digital audio data, and the output is the data sent to the server over the network. Specifically, an application on the device uses an API to send the audio data to the server as a POST request.

[0570] Step 3:

[0571] The server passes the received voice data to a voice analysis engine, which converts the voice into text data. The input is digital voice data, and the output is text data. A process is performed to convert the voice data into text data. Specifically, a voice analysis engine on the server (for example, Google Cloud Speech-to-Text) converts the voice data into text.

[0572] Step 4:

[0573] The server analyzes the converted text data using a natural language processing (NLP) engine to identify inappropriate words and phrases. The input is text data, and the output is text data with inappropriate words and phrases identified. The text data is analyzed and processed to identify inappropriate words and phrases. Specifically, an NLP engine (for example, spaCy) analyzes the text "You're an idiot!" and identifies "idiot" as an inappropriate word.

[0574] Step 5:

[0575] The server replaces the identified inappropriate words with kinder words from a database. The input is text data with inappropriate words flagged, and the output is the text data replaced with kinder words. Data processing is performed to correct the text data. Specifically, "idiot" is converted to "I think that's a bit off" to generate the text "I think you're a bit off!"

[0576] Step 6:

[0577] The server passes the corrected text data to a speech synthesis engine, which converts it into new speech data. The input is the corrected text data, and the output is the new speech data. A process is performed to reconvert the text data into speech data. Specifically, the speech synthesis engine (for example, Google Text-to-Speech) converts the text "I think you're a little off!" into speech data.

[0578] Step 7:

[0579] The server sends the generated new audio data to the device and distributes it to the listener in real time. The input is the new audio data, and the output is the audio received by the listener. Processing to distribute the audio data is performed. Specifically, the audio playback application on the device plays and distributes the new audio data to the listener.

[0580] (Application example 1)

[0581] 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."

[0582] In conventional social networking sites and streaming platforms, inappropriate comments can be made, causing trouble between users. However, it is difficult to detect and correct inappropriate comments in real time, which can potentially ruin the experience for streamers and viewers. Therefore, this invention aims to promote healthy communication and improve the user experience by detecting inappropriate comments in real time and converting them into appropriate words.

[0583] 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.

[0584] In this invention, the server includes means for acquiring voice through a microphone, means for analyzing the acquired voice using a voice recognition module to detect inappropriate remarks, means for automatically converting the detected inappropriate remarks into kinder words using a generative AI model, means for converting the converted text data back into voice using a voice synthesis module, means for delivering the converted voice to the other party in real time, and database means for providing prompt sentences to the generative AI model, thereby enabling the detection and correction of inappropriate remarks in real time.

[0585] A "microphone" is an input device for converting sound into an electrical signal.

[0586] A "voice recognition module" is a software or hardware component for converting captured speech into text data.

[0587] A "generative AI model" is a trained model that includes an algorithm that uses natural language processing technology to convert inappropriate remarks into kinder ones.

[0588] A "speech synthesis module" is a system component for converting text data into speech data.

[0589] A "prompt sentence" is an input sentence provided to a generative AI model to generate an appropriate output.

[0590] A "database" is a collection of data that systematically organizes information so that it can be quickly searched and used.

[0591] "Real-time" means that the processing occurs immediately, without delay.

[0592] This invention is implemented using a system including a microphone, a speech recognition module, a generative AI model, a speech synthesis module, and a database. Specifically, the invention can be implemented using the following configuration and process.

[0593] The user first inputs their voice through the microphone. This voice is picked up by the device and captured as a digital signal. The captured voice data is converted into text data using a speech recognition module and sent to the server. This speech recognition module can be commercial software such as Google's speech recognition API.

[0594] The server uses a speech analysis engine to analyze the text data generated by the speech recognition module and detect inappropriate remarks. This is done using a generative AI model. The generative AI model has the ability to detect inappropriate remarks and replace them with appropriate expressions. A specific example of this model is the "twitter-roberta-large-offensive" model based on Huggingface's transformers library. The generative AI model is linked to a database that provides prompt sentences, and this database contains pre-registered inappropriate phrases and their replacement candidates.

[0595] The text data of the detected inappropriate remarks is automatically converted into kinder words by a generative AI model. For example, the word "idiot" is converted to "I think that's a bit off." This updates the text data to new text data that uses appropriate words. This new text data is saved by the server and converted back into voice data using a voice synthesis module. A voice synthesis library such as pyttsx3 can be used as the voice synthesis module.

[0596] The server sends the voice data generated by the voice synthesis module to the terminal and distributes it to the other party in real time, allowing the listener to hear the voice with inappropriate remarks corrected.

[0597] As a specific example of its use, if a user says "You're an idiot!" during a live broadcast or game broadcast, this system will convert it into "I think you're a little different!" and broadcast it immediately. Through this process, healthy communication between the broadcaster and the viewers is maintained.

[0598] An example of input to a generative AI model is:

[0599] Input voice: You're an idiot!

[0600] Output Voice: I think you're a little different!

[0601] In this way, a system using the present invention can instantly detect and correct inappropriate comments during live broadcasts, realizing healthy communication.

[0602] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0603] Step 1:

[0604] The user inputs voice through the microphone, which is then picked up by the device and captured as a digital signal. At this stage, the voice is stored on the device as concrete digital data.

[0605] Step 2:

[0606] The device converts the captured voice data into text data using a voice recognition module. The input is digital voice data and the output is text data. In this voice recognition process, speech-to-text conversion is performed using a speech recognition API (e.g., Google's speech recognition API).

[0607] Step 3:

[0608] The server receives the text data sent from the speech recognition module and analyzes it using a speech analysis engine. In this step, the input text data is analyzed to determine whether certain words or phrases are inappropriate.

[0609] Step 4:

[0610] A generative AI model is used to convert text data in which inappropriate remarks have been detected into kinder words. The server searches a database for kinder words that correspond to the inappropriate words and updates the text data. The input here is the analyzed text data, and the output is new text data converted into kinder words. For example, Huggingface's twitter-roberta-large-offensive model is used as the generative AI model.

[0611] Step 5:

[0612] The server passes the new text data to a speech synthesis module, which converts it back into speech data. The input is the converted text data, and the output is speech data. This speech synthesis is performed using a speech synthesis library such as pyttsx3.

[0613] Step 6:

[0614] The server sends the newly generated voice data to the terminal and distributes it to the other party in real time. The input here is the voice data from the speech synthesis module, and the output is the voice data distributed to the recipient. This allows the listener to hear the voice in real time with inappropriate remarks corrected.

[0615] Through these steps, inappropriate comments made by users are converted into kinder words in real time and delivered.

[0616] 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.

[0617] This invention relates to a system that detects inappropriate comments in real time on social networking sites and distribution platforms and converts them into more appropriate and gentle language. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, the system converts them into more appropriate and gentle language, improving the quality of communication.

[0618] Basic system configuration

[0619] 1. Voice input

[0620] The user speaks into the microphone. For example, the user says, "You're an idiot!"

[0621] The device captures this audio and obtains the data as a digital signal.

[0622] 2. Audio analysis

[0623] The terminal transmits the acquired voice data to the server.

[0624] The server uses a voice analysis engine to convert the voice data into text data.

[0625] The server analyzes this text data and identifies inappropriate words and phrases, for example, detecting "idiot" in the text "You're an idiot!" as an inappropriate word.

[0626] 3. Detecting inappropriate comments

[0627] The server flags inappropriate words and phrases identified by the speech analysis engine.

[0628] The flagged text is treated as data indicating inappropriate language.

[0629] 4. Emotion recognition

[0630] The device sends the voice data to the emotion engine.

[0631] The emotion engine analyzes the user's emotions from the voice data and generates emotion data.

[0632] The emotion data analyzed by the emotion engine is used to select subsequent kind words. For example, if a user says "idiot" with an angry emotion, the emotion data includes "anger."

[0633] 5. Correction of statements

[0634] The server receives the flagged text and searches a database for benign words that correspond to the inappropriate phrases.

[0635] The server adjusts its choice of kind words based on the emotional data, for example, converting "stupid" into "I think that's a bit off," while converting "let's calm down and think about it" if the emotional data indicates "anger."

[0636] The server generates the corrected text data and records it as new text data.

[0637] 6. Audio Distribution

[0638] The server passes the corrected text data to a speech synthesis engine, which converts it back into speech data.

[0639] The device receives new audio data and distributes it to listeners in real time.

[0640] For example, if a user says "You're an idiot!" and the emotion is analyzed as "anger," the listener will hear a gentle and appropriate response: "Calm down and think about it!"

[0641] Specific examples

[0642] If the user emotionally says, "You're an idiot!"

[0643] 1. Voice input

[0644] The user speaks into the microphone, "You're an idiot!"

[0645] The device captures this audio and transmits it as a digital signal to the server.

[0646] 2. Audio analysis

[0647] The server uses a voice analysis engine to convert "You're an idiot!" into text data.

[0648] The server parses this text and identifies the word "idiot" as inappropriate.

[0649] 3. Detecting inappropriate comments

[0650] The server flags "idiots."

[0651] 4. Emotion recognition

[0652] The device sends the voice data to the emotion engine.

[0653] The emotion engine extracts the emotion of "anger" from the voice data.

[0654] 5. Correction of statements

[0655] The server converts "idiot" into "calm down and think about it."

[0656] The finished text will be "You need to calm down and think about it!"

[0657] 6. Audio Distribution

[0658] The server passes the corrected text to a speech synthesis engine to convert it into voice data.

[0659] The device receives new audio and distributes it to listeners.

[0660] Usage scene

[0661] This system can be applied to a wide range of situations, such as live streaming and interactive social networking sites. By combining it with an emotion engine, it is possible to instantly convert inappropriate comments that users may unconsciously make into appropriate and kind words, which is expected to improve the quality of communication and prevent conflicts between users.

[0662] The processing flow will be explained below.

[0663] Program processing

[0664] Step 1:

[0665] The user speaks into the microphone.

[0666] Example: "You're an idiot!"

[0667] Step 2:

[0668] The device captures the user's voice with a microphone.

[0669] Step 3:

[0670] The device converts the captured audio data into a digital format and stores it in a buffer.

[0671] Step 4:

[0672] The terminal transmits the audio data stored in the buffer to the server.

[0673] Step 5:

[0674] The server passes the received voice data to the voice analysis engine.

[0675] Step 6:

[0676] The voice analysis engine converts the voice data into text (voice recognition).

[0677] Step 7:

[0678] A speech analysis engine analyzes text data and identifies inappropriate words and phrases.

[0679] Step 8:

[0680] A speech analysis engine identifies and flags inappropriate words and expressions based on specific criteria.

[0681] Example: Identifying "idiot" as an inappropriate word in the text "You're an idiot!"

[0682] Step 9:

[0683] The server receives the flagged text data.

[0684] Step 10:

[0685] The device sends the voice data to the emotion engine.

[0686] Step 11:

[0687] The emotion engine analyzes the user's emotions from the voice data and generates emotion data.

[0688] Step 12:

[0689] The emotion engine sends the emotion analysis results to the server.

[0690] Example: The emotion engine detects "anger."

[0691] Step 13:

[0692] The server refers to the emotion data when replacing inappropriate words with kinder words.

[0693] Example: "Stupid" can be added to "I think that's a bit wrong," and if the emotion is "anger," it can be changed to "Let's calm down and think about it."

[0694] Step 14:

[0695] The server generates the modified text data.

[0696] For example: "You're an idiot!" becomes "You need to calm down and think!"

[0697] Step 15:

[0698] The server passes the corrected text data to the speech synthesis engine.

[0699] Step 16:

[0700] The speech synthesis engine converts the text data into speech data (speech synthesis).

[0701] Step 17:

[0702] The server transmits the generated voice data to the terminal.

[0703] Step 18:

[0704] The audio data received by the terminal is played on the speaker.

[0705] For example, if a user says, "You're an idiot!" and the emotion engine analyzes this as "anger," the final voice the listener hears will be a gentle and appropriate "You need to calm down and think about it!" This series of processes instantly converts inappropriate remarks into gentler words, enabling healthy communication in real time.

[0706] Example 2

[0707] 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."

[0708] On conventional social networking sites and distribution platforms, problems could arise due to inappropriate comments made by users. Simply detecting and filtering inappropriate comments is difficult to completely prevent, making it impossible to improve the quality of communication. Furthermore, simply converting words without considering the user's feelings can lead to the problem of not accurately conveying the intent of the comment.

[0709] 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.

[0710] In this invention, the server includes means for acquiring voice through a microphone, means for analyzing the acquired voice and detecting inappropriate remarks, means for automatically converting the detected inappropriate remarks into kinder words, means for analyzing the user's emotions and generating emotion data, means for adjusting the selection of kinder words based on the emotion data, and means for delivering the converted voice to the other party. This makes it possible to appropriately detect and correct the user's remarks in real time and flexibly convert words taking the user's emotions into consideration. This can prevent problems caused by inappropriate remarks and improve the quality of communication.

[0711] The "means for acquiring voice through a microphone" refers to a hardware and software device that captures the voice spoken by the user into the microphone as a digital signal and converts it into a form that can be processed within the system.

[0712] The "means for analyzing the acquired voice and detecting inappropriate remarks" refers to a device and program that uses a voice analysis engine to convert the acquired voice data into text data, and then analyzes the text data to identify inappropriate words and phrases.

[0713] "Means for automatically converting detected inappropriate language to kinder language" means devices and programs that have a database and algorithms for replacing words or phrases identified as inappropriate with kinder language.

[0714] The "means for analyzing user emotions and generating emotion data" refers to a device and program that uses an emotion recognition engine to analyze user emotions from voice data and output the analysis results as emotion data.

[0715] The "means for adjusting the selection of kind words based on emotional data" refers to a device and a program that selects appropriate kind words based on the generated emotional data and converts inappropriate words into the most appropriate words in accordance with the emotion.

[0716] The "means for delivering the converted voice to the other party" refers to a device and a program that reconverts the text data converted into friendly words into voice data and delivers the voice data to the listener in real time.

[0717] This invention relates to a system that detects inappropriate comments in real time on social media and distribution platforms and converts them into more appropriate and gentle language. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, the system converts them into more appropriate and gentle language, improving the quality of communication.

[0718] Basic system configuration

[0719] 1. Voice input

[0720] The user speaks into the microphone. For example, the user says, "You're an idiot!"

[0721] The device captures this audio and obtains the data as a digital signal.

[0722] 2. Audio analysis

[0723] The terminal transmits the acquired voice data to the server.

[0724] The server uses a speech analysis engine (e.g., Google Cloud Speech-to-Text) to convert the voice data into text data.

[0725] The server analyzes this text data and identifies inappropriate words and phrases, for example, detecting "idiot" in the text "You're an idiot!" as an inappropriate word.

[0726] 3. Detecting inappropriate comments

[0727] The server flags inappropriate words and phrases identified by the speech analysis engine.

[0728] The flagged text is treated as data indicating inappropriate language.

[0729] 4. Emotion recognition

[0730] The device sends the voice data to an emotion engine (e.g., IBM Watson Tone Analyzer).

[0731] The emotion engine analyzes the user's emotions from the voice data and generates emotion data. For example, if a user says "idiot" in anger, the emotion data will include "anger."

[0732] 5. Correction of statements

[0733] The server receives the flagged text and searches a database for benign words that correspond to the inappropriate phrases.

[0734] The server adjusts its choice of kind words based on the emotional data, for example, converting "stupid" into "I think that's a bit off," while converting "let's calm down and think about it" if the emotional data indicates "anger."

[0735] The server generates the corrected text data and records it as new text data.

[0736] 6. Audio Distribution

[0737] The server passes the corrected text data to a speech synthesis engine (e.g., Amazon Polly) and converts it back into speech data.

[0738] The device receives new audio data and delivers it to the listener in real time. For example, a voice saying "You're an idiot!" can be changed to "Calm down and think!"

[0739] Specific examples

[0740] If the user emotionally says, "You're an idiot!"

[0741] 1. Voice input

[0742] The user speaks into the microphone, "You're an idiot!"

[0743] The device captures this audio and transmits it as a digital signal to the server.

[0744] 2. Audio analysis

[0745] The server uses a speech analysis engine to convert "You're an idiot!" into text data.

[0746] The server parses this text and identifies the word "idiot" as inappropriate.

[0747] 3. Detecting inappropriate comments

[0748] The server flags "idiots."

[0749] 4. Emotion recognition

[0750] The device sends the voice data to the emotion engine.

[0751] The emotion engine extracts the emotion of "anger" from the voice data.

[0752] 5. Correction of statements

[0753] The server converts "idiot" into "calm down and think about it."

[0754] The finished text will be "You need to calm down and think about it!"

[0755] 6. Audio Distribution

[0756] The server passes the corrected text to a speech synthesis engine to convert it into voice data.

[0757] The device receives the new audio and broadcasts it to the listener, who hears the gentle words, "Stay calm and think!"

[0758] Prompt Sentence Examples

[0759] An example of a specific prompt would be:

[0760] If a user says "You're an idiot!", convert the speech data into text, detect the inappropriate phrase "idiot", analyze the emotion of anger, and then convert it into "You need to calm down and think!"

[0761] This prompt allows the system to properly analyze the user's speech content and emotions and make appropriate conversions.

[0762] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0763] Step 1: Capturing Audio Input

[0764] Input: User-generated speech

[0765] Output: Audio data as a digital signal

[0766] The device uses a microphone to capture the user's voice. For example, if the user says "You're an idiot!", the microphone captures this voice in real time. The device converts the captured voice into a digital signal and sends it to the server.

[0767] Step 2: Convert audio data to text

[0768] Input: Audio data as a digital signal

[0769] Output: Text data

[0770] The server processes the received voice data. Specifically, it uses a voice analysis engine (e.g., Google Cloud Speech-to-Text) to convert the voice data into text data. For example, the voice data "You're an idiot!" is converted into text data "You're an idiot!"

[0771] Step 3: Detect inappropriate language

[0772] Input: Text data

[0773] Output: Text data with inappropriate words flagged

[0774] The server analyzes the converted text data. Specifically, it uses a natural language processing engine to identify inappropriate words and phrases in the text. For example, the server flags the inappropriate words and phrases in the text "You're an idiot!" and passes the flagged text on to the next processing step.

[0775] Step 4: Emotion Recognition

[0776] Input: Audio data

[0777] Output: Emotion data

[0778] The device sends the collected voice data to an emotion recognition engine (e.g., IBM Watson Tone Analyzer). The emotion engine analyzes the user's emotions from the voice data and generates emotion data. For example, if a user says "idiot" out of anger, the emotion data will include "anger." The generated emotion data is sent to the server.

[0779] Step 5: Modify your statement

[0780] Input: Text and sentiment data with inappropriate words flagged

[0781] Output: Corrected text data

[0782] The server receives text data with inappropriate words flagged and emotion data. Specifically, it analyzes the text data and searches a database for appropriate kind words. For example, it converts "idiot" into "calm down and think about it." The server adjusts the choice of kind words based on the emotion data. Finally, it generates corrected text data and records it as new text data.

[0783] Step 6: Synthesize and deliver audio

[0784] Input: Modified text data

[0785] Output: New audio data

[0786] The server receives the corrected text data and passes it to a speech synthesis engine (e.g., Amazon Polly). The speech synthesis engine converts the corrected text into speech data. For example, "You're an idiot!" is converted into "Calm down and think about it!" The device receives the newly generated speech data and delivers it to the listener in real time. The listener can hear the gentle words "Calm down and think about it!"

[0787] (Application example 2)

[0788] 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."

[0789] In recent years, with the spread of live streaming and real-time communication, inappropriate comments have become a problem. Inappropriate comments can hurt people, cause trouble, and hinder healthy communication. Furthermore, due to a lack of correction based on the speaker's emotions, mechanical conversion is often insufficient. To address this issue, there is a need for a system that converts comments into appropriate words based on emotions in real time.

[0790] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0791] In this invention, the server includes means for acquiring voice through a microphone, means for analyzing the acquired voice and detecting inappropriate remarks, means for recognizing the user's emotions when an inappropriate remark is detected, means for converting the detected inappropriate remarks into kinder words based on the recognized emotions, and means for delivering the converted voice to the other party. This makes it possible to convert inappropriate remarks into appropriate words in real time, taking the user's emotions into consideration.

[0792] (definition statement)

[0793] A "microphone" is a device that converts sound into an electrical signal and is used to capture what a user says.

[0794] The "means for analyzing voice and detecting inappropriate remarks" refers to a means for analyzing acquired voice data and identifying inappropriate words and phrases therefrom.

[0795] The "means for recognizing emotions" is a means for analyzing the user's emotions from the acquired voice data and generating emotion data.

[0796] The "means for converting to kinder words" is a means for automatically replacing inappropriate remarks with kinder words, and also a means for selecting appropriate expressions based on emotional data.

[0797] The "means for reconverting to audio" is a means for converting the corrected text data into audio data.

[0798] The "means for processing in real time" refers to a means for immediately processing acquired voice data without delay.

[0799] "Means for delivering to the other party in real time" refers to means for instantly transmitting the corrected audio data to the listener or the other party.

[0800] The present invention is a system for detecting inappropriate comments in real time and converting them into friendly comments. This system can be applied when users are live streaming.

[0801] Basic system configuration

[0802] The system is implemented using the following hardware and software:

[0803] Microphone: A device that converts the user's voice into an electrical signal and captures the sound.

[0804] Terminal: A device such as a computer or smartphone that analyzes audio data acquired from a microphone.

[0805] Server: A remote server containing a speech analysis engine, emotion recognition engine, text-to-speech engine, and speech synthesis engine.

[0806] System operation explanation

[0807] 1. Audio capture:

[0808] The user speaks into a microphone and transmits the audio in real time to the device, where an application on the device captures the audio and obtains the data as a digital signal.

[0809] 2. Audio analysis:

[0810] The device sends the captured voice data to the server, which uses a speech analysis engine (e.g., Google Speech Recognition API) on the server to convert the voice data into text data. A text analysis algorithm (e.g., TextBlob) is used to identify inappropriate words and phrases.

[0811] 3. Emotion recognition:

[0812] The emotion recognition engine analyzes the user's emotions from the voice data and generates emotion data. For emotion recognition, a voice emotion recognition library, for example, is used.

[0813] 4. Statement correction:

[0814] The server receives the flagged text and searches a database for a kinder equivalent to the inappropriate phrase. Based on the emotion data, it adjusts the choice of kinder words. For example, it converts "stupid" to "calm down and think about it," but if the emotion data indicates "anger," it converts it to "think again."

[0815] 5. Speech synthesis:

[0816] The converted text data is converted back into audio data using a speech synthesis engine (e.g., Google Text-to-Speech). The device receives the new audio data and delivers it to the listener in real time.

[0817] Specific examples

[0818] For example, if a user says "You're an idiot!" during a stream, the system works as follows: This utterance is converted in real time into "Calm down and think!" and broadcast as audio to the viewer.

[0819] Example prompts to input to the generative AI model

[0820] "I'd like to develop an application that converts speech signals into text data and modifies inappropriate expressions in the text based on emotion recognition. Please generate the code to accomplish this process in real time in a voice-delivery environment using the SpeechRecognition library, Google Speech Recognition API, emotion recognition library, and Google Text-to-Speech library."

[0821] This system can instantly detect inappropriate comments made during broadcasts and appropriately change them based on emotions, enabling healthy communication.

[0822] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0823] (Program processing flow)

[0824] Step 1:

[0825] Audio Capture

[0826] During live streaming, the user speaks into the microphone. The device captures this voice and obtains it as a digital signal. The input is the user's voice signal, and the output is the captured digital voice data. In this step, the device uses the microphone to capture voice in real time.

[0827] Step 2:

[0828] Audio data transmission

[0829] The device sends the captured audio data to the server. The input is the captured digital audio data, and the output is the audio data sent to the server. In this step, the device uploads the audio data to the server.

[0830] Step 3:

[0831] Audio analysis

[0832] The server converts the acquired voice data into text data using a voice analysis engine (e.g., Google Speech Recognition API). The input is the voice data sent to the server, and the output is the converted text data. In this step, the server recognizes the voice and generates the result as text.

[0833] Step 4:

[0834] Inappropriate speech detection

[0835] The server uses a text analysis algorithm (e.g., TextBlob) to detect inappropriate content from the converted text data. The input is text data, and the output is text data with inappropriate words and phrases flagged. In this step, the server analyzes the text data to identify inappropriate content.

[0836] Step 5:

[0837] emotion recognition

[0838] The server uses an emotion recognition engine (e.g., a voice emotion recognition library) to analyze the user's emotion from the voice data. The input is voice data, and the output is emotion data. In this step, the server analyzes the emotion of the voice and determines the emotion the user expressed when speaking.

[0839] Step 6:

[0840] Correction of statement

[0841] The server converts inappropriate comments into appropriate ones based on the emotion data. The input is text data with inappropriate words and emotion data flagged, and the output is the corrected text data. In this step, the server replaces the inappropriate comments with more appropriate words based on the emotion.

[0842] Step 7:

[0843] Speech synthesis

[0844] The server converts the corrected text data back into speech data using a speech synthesis engine (e.g., Google Text-to-Speech API). The input is the corrected text data, and the output is the synthesized speech data. In this step, the server converts the text data back into speech.

[0845] Step 8:

[0846] Audio data distribution

[0847] The device receives new audio data obtained from the server and delivers it to the listener in real time. The input is the synthesized audio data, and the output is the modified audio delivered to the listener. In this step, the device delivers the new audio in real time.

[0848] 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.

[0849] 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.

[0850] 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.

[0851] [Third embodiment]

[0852] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0853] 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.

[0854] 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).

[0855] 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.

[0856] 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.

[0857] 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).

[0858] 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.

[0859] 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.

[0860] 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.

[0861] 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.

[0862] 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.

[0863] 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."

[0864] This invention relates to a system that detects inappropriate comments on social media and distribution platforms in real time and converts them into more gentle comments. This system is realized through a series of processes: voice input, voice analysis, detection of inappropriate comments, comment correction, and voice distribution.

[0865] Basic system configuration

[0866] 1. Voice input

[0867] The user uses a microphone to input speech. For example, consider the user saying "You're an idiot!"

[0868] The device captures this audio as a digital signal and obtains the data.

[0869] 2. Audio analysis

[0870] The terminal transmits the acquired voice data to the server.

[0871] The server uses a voice analysis engine to convert the voice data into text data.

[0872] The server analyzes this text data and identifies inappropriate words and phrases, for example, detecting "idiot" in the text "You're an idiot!" as an inappropriate word.

[0873] 3. Detecting inappropriate comments

[0874] The server flags inappropriate words and phrases identified by the speech analysis engine.

[0875] The flagged text is treated as data indicating inappropriate language.

[0876] 4. Correction of statements

[0877] The server receives the flagged text and searches a database for a more benign equivalent for the inappropriate phrase, for example, changing "stupid" to "I think that's a bit off."

[0878] The server generates the corrected text and records it as new text data.

[0879] 5. Audio Distribution

[0880] The server passes the corrected text data to a speech synthesis engine, which converts it back into speech data.

[0881] The device receives new audio data and distributes it to listeners in real time.

[0882] For example, if the user says, "You're an idiot!", the listener will hear the gentler, "I think you're a bit different!"

[0883] Specific examples

[0884] If a user says, "You're an idiot!"

[0885] 1. Voice input

[0886] The user speaks into the microphone, "You're an idiot!"

[0887] The device captures this audio and transmits it as a digital signal to the server.

[0888] 2. Audio analysis

[0889] The server uses a voice analysis engine to convert "You're an idiot!" into text data.

[0890] The server parses this text and identifies the word "idiot" as inappropriate.

[0891] 3. Detecting inappropriate comments

[0892] The server flags "idiots."

[0893] 4. Correction of statements

[0894] The server converts "idiot" to "I think that's a bit off."

[0895] The finished text is "I think you're a little different!"

[0896] 5. Audio Distribution

[0897] The server passes the corrected text to a speech synthesis engine to convert it into voice data.

[0898] The device receives new audio and distributes it to listeners.

[0899] Usage scene

[0900] This system can be applied to a wide range of situations, including live streaming and interactive social networking sites. By using this system, inappropriate comments that users may make unconsciously can be instantly converted into gentler words, which is expected to improve the quality of communication and prevent conflicts between users.

[0901] The processing flow will be explained below.

[0902] Program processing

[0903] Step 1:

[0904] The user speaks into the microphone.

[0905] Example: "You're an idiot!"

[0906] Step 2:

[0907] The device captures the user's voice with a microphone.

[0908] Step 3:

[0909] The device converts the captured audio data into a digital format and stores it in a buffer.

[0910] Step 4:

[0911] The terminal transmits the audio data stored in the buffer to the server.

[0912] Step 5:

[0913] The server passes the received voice data to the voice analysis engine.

[0914] Step 6:

[0915] The voice analysis engine converts the voice data into text (voice recognition).

[0916] Step 7:

[0917] A speech analysis engine analyzes text data and identifies inappropriate words and phrases.

[0918] Step 8:

[0919] A speech analysis engine identifies and flags inappropriate words and expressions based on specific criteria.

[0920] Example: Identifying "idiot" as an inappropriate word in the text "You're an idiot!"

[0921] Step 9:

[0922] The server receives the flagged text data.

[0923] Step 10:

[0924] The server replaces flagged inappropriate words with kinder words from a database.

[0925] Example: "Stupid" changed to "I think that's a bit different."

[0926] Step 11:

[0927] The server generates the modified text data.

[0928] For example: "You're an idiot!" turns into "I think you're a bit different!"

[0929] Step 12:

[0930] The server passes the corrected text data to the speech synthesis engine.

[0931] Step 13:

[0932] The speech synthesis engine converts the text data into speech data (speech synthesis).

[0933] Step 14:

[0934] The server transmits the generated voice data to the terminal.

[0935] Step 15:

[0936] The audio data received by the terminal is played on the speaker.

[0937] For example, if a user says, "You're an idiot!", the server converts this speech into "I think you're a little bit different!" The listener hears a gentler version of "I think you're a little bit different!" Through this series of processes, inappropriate remarks are converted into gentler words in real time, maintaining healthy communication.

[0938] Example 1

[0939] 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."

[0940] Today, many people communicate in real time on platforms such as social media and live streaming. However, inappropriate comments can be made during these interactions, causing trouble between users. A system that can avoid such situations and improve the quality of communication is needed.

[0941] 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.

[0942] In this invention, the server includes means for converting voice data into text data, means for analyzing the text data to detect inappropriate comments, and means for replacing the detected inappropriate comments with kinder words selected from a database, thereby making it possible to prevent trouble between users and improve the quality of communication.

[0943] A "microphone" is a device that converts sound into an electrical signal.

[0944] A "means for converting audio into a digital signal" is a process or device for converting analog audio into digital data.

[0945] A "means for transmitting to a server over the Internet" is a method or device for transmitting a digital signal to a remote server using a network.

[0946] A "server" is a computer system for processing and storing data.

[0947] "Means for converting into text data" refers to a process or system for converting voice data into a string of characters.

[0948] A "means for analyzing text data" is a method or device for analyzing text data and understanding or classifying its content.

[0949] An "inappropriate language detection means" is a method or apparatus for identifying inappropriate words or phrases from the analyzed text data.

[0950] A "database" is a system for storing a collection of data in an organized manner and for retrieving and updating it as needed.

[0951] A "means for replacing with kinder words" is a method or device for converting detected inappropriate expressions into appropriate expressions.

[0952] A "means for reconverting to audio data" is a process or system for converting text data into audio format.

[0953] A "terminal" is a computer device operated by a user, such as a PC or smartphone.

[0954] A "natural language processing engine" is software or a system for understanding and processing human language.

[0955] A "speech synthesis engine" is software or a system that converts input text into voice data.

[0956] The present invention relates to a system that detects inappropriate comments in real time on social media and distribution platforms and converts them into more polite language. Specific ways in which the present invention can be implemented are described below.

[0957] Audio input and digital signal conversion

[0958] Users input their voice using a microphone. This can be the built-in microphone of a PC or smartphone, or an external microphone. The device captures this voice and converts it into a digital signal. Specifically, audio capture software running on the device converts analog voice into digital data.

[0959] Sending and analyzing voice data

[0960] The device sends the captured digital voice data to a server via the Internet. An application running on the device sends the voice data to the server as a POST request via a dedicated API. The server then passes the received voice data to a voice analysis engine (e.g., Google Cloud Speech-to-Text) and converts the voice into text data.

[0961] The converted text data is then analyzed by a natural language processing (NLP) engine (e.g., spaCy or TensorFlow), which identifies inappropriate words and phrases within the text. For example, the statement "You're an idiot!" is converted to the text "You're an idiot!", with "idiot" identified as an inappropriate word.

[0962] Detecting and correcting inappropriate language

[0963] After detecting an inappropriate remark, the server searches a database (e.g., MySQL or PostgreSQL) for appropriate corrections. The detected inappropriate phrases are replaced with corresponding more benign words. For example, "idiot" can be converted to "I think you're a bit off" to generate the new text "I think you're a bit off!".

[0964] Recoding and audio distribution

[0965] The server passes the corrected text data to a speech synthesis engine (for example, Google Text-to-Speech or Amazon Polly), which reconverts it into new voice data. This new voice data is then sent back to the device and delivered to the listener. This allows, for example, a statement like "You're an idiot!" to be converted into a more gentle statement like "I think you're a little different!" and delivered.

[0966] Specific examples

[0967] Specifically, when a user says "You're an idiot!", the following process occurs:

[0968] 1. The user says into the microphone, "You're an idiot!"

[0969] 2. The device captures this audio and converts it into a digital signal.

[0970] 3. The digital signal is sent over the internet to a server.

[0971] 4. The server uses a speech analysis engine to convert the speech into text, generating the text "You're an idiot!"

[0972] 5. The server analyzes the text with an NLP engine and identifies the word "idiot" as inappropriate.

[0973] 6. The server converts "idiot" to "I think you're a little bit off" and generates the text "I think you're a little bit off!"

[0974] 7. The corrected text is converted back into voice data by the speech synthesis engine.

[0975] 8. This new audio data is sent to the device and distributed to the listener.

[0976] The above is a specific embodiment for carrying out the present invention. By using this system, inappropriate comments made by users can be instantly converted into kinder words, thereby improving the quality of communication.

[0977] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0978] Step 1:

[0979] The user uses the microphone to input speech, for example, saying "You're an idiot!"

[0980] The device converts this audio into a digital signal. The input is an analog audio signal, and the output is digital audio data. A process is performed to convert the analog audio into digital audio data. Specifically, the device's audio capture software converts the audio into a digital signal.

[0981] Step 2:

[0982] The device sends the captured digital audio data to a server over the Internet. The input is the digital audio data, and the output is the data sent to the server over the network. Specifically, an application on the device uses an API to send the audio data to the server as a POST request.

[0983] Step 3:

[0984] The server passes the received voice data to a voice analysis engine, which converts the voice into text data. The input is digital voice data, and the output is text data. A process is performed to convert the voice data into text data. Specifically, a voice analysis engine on the server (for example, Google Cloud Speech-to-Text) converts the voice data into text.

[0985] Step 4:

[0986] The server analyzes the converted text data using a natural language processing (NLP) engine to identify inappropriate words and phrases. The input is text data, and the output is text data with inappropriate words and phrases identified. The text data is analyzed and processed to identify inappropriate words and phrases. Specifically, an NLP engine (for example, spaCy) analyzes the text "You're an idiot!" and identifies "idiot" as an inappropriate word.

[0987] Step 5:

[0988] The server replaces the identified inappropriate words with kinder words from a database. The input is text data with inappropriate words flagged, and the output is the text data replaced with kinder words. Data processing is performed to correct the text data. Specifically, "idiot" is converted to "I think that's a bit off" to generate the text "I think you're a bit off!"

[0989] Step 6:

[0990] The server passes the corrected text data to a speech synthesis engine, which converts it into new speech data. The input is the corrected text data, and the output is the new speech data. A process is performed to reconvert the text data into speech data. Specifically, the speech synthesis engine (for example, Google Text-to-Speech) converts the text "I think you're a little off!" into speech data.

[0991] Step 7:

[0992] The server sends the generated new audio data to the device and distributes it to the listener in real time. The input is the new audio data, and the output is the audio received by the listener. Processing to distribute the audio data is performed. Specifically, the audio playback application on the device plays and distributes the new audio data to the listener.

[0993] (Application example 1)

[0994] 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."

[0995] In conventional social networking sites and streaming platforms, inappropriate comments can be made, causing trouble between users. However, it is difficult to detect and correct inappropriate comments in real time, which can potentially ruin the experience for streamers and viewers. Therefore, this invention aims to promote healthy communication and improve the user experience by detecting inappropriate comments in real time and converting them into appropriate words.

[0996] 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.

[0997] In this invention, the server includes means for acquiring voice through a microphone, means for analyzing the acquired voice using a voice recognition module to detect inappropriate remarks, means for automatically converting the detected inappropriate remarks into kinder words using a generative AI model, means for converting the converted text data back into voice using a voice synthesis module, means for delivering the converted voice to the other party in real time, and database means for providing prompt sentences to the generative AI model, thereby enabling the detection and correction of inappropriate remarks in real time.

[0998] A "microphone" is an input device for converting sound into an electrical signal.

[0999] A "voice recognition module" is a software or hardware component for converting captured speech into text data.

[1000] A "generative AI model" is a trained model that includes an algorithm that uses natural language processing technology to convert inappropriate remarks into kinder ones.

[1001] A "speech synthesis module" is a system component for converting text data into speech data.

[1002] A "prompt sentence" is an input sentence provided to a generative AI model to generate an appropriate output.

[1003] A "database" is a collection of data that systematically organizes information so that it can be quickly searched and used.

[1004] "Real-time" means that the processing occurs immediately, without delay.

[1005] This invention is implemented using a system including a microphone, a speech recognition module, a generative AI model, a speech synthesis module, and a database. Specifically, the invention can be implemented using the following configuration and process.

[1006] The user first inputs their voice through the microphone. This voice is picked up by the device and captured as a digital signal. The captured voice data is converted into text data using a speech recognition module and sent to the server. This speech recognition module can be commercial software such as Google's speech recognition API.

[1007] The server uses a speech analysis engine to analyze the text data generated by the speech recognition module and detect inappropriate remarks. This is done using a generative AI model. The generative AI model has the ability to detect inappropriate remarks and replace them with appropriate expressions. A specific example of this model is the "twitter-roberta-large-offensive" model based on Huggingface's transformers library. The generative AI model is linked to a database that provides prompt sentences, and this database contains pre-registered inappropriate phrases and their replacement candidates.

[1008] The text data of the detected inappropriate remarks is automatically converted into kinder words by a generative AI model. For example, the word "idiot" is converted to "I think that's a bit off." This updates the text data to new text data that uses appropriate words. This new text data is saved by the server and converted back into voice data using a voice synthesis module. A voice synthesis library such as pyttsx3 can be used as the voice synthesis module.

[1009] The server sends the voice data generated by the voice synthesis module to the terminal and distributes it to the other party in real time, allowing the listener to hear the voice with inappropriate remarks corrected.

[1010] As a specific example of its use, if a user says "You're an idiot!" during a live broadcast or game broadcast, this system will convert it into "I think you're a little different!" and broadcast it immediately. Through this process, healthy communication between the broadcaster and the viewers is maintained.

[1011] An example of input to a generative AI model is:

[1012] Input voice: You're an idiot!

[1013] Output Voice: I think you're a little different!

[1014] In this way, a system using the present invention can instantly detect and correct inappropriate comments during live broadcasts, realizing healthy communication.

[1015] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1016] Step 1:

[1017] The user inputs voice through the microphone, which is then picked up by the device and captured as a digital signal. At this stage, the voice is stored on the device as concrete digital data.

[1018] Step 2:

[1019] The device converts the captured voice data into text data using a voice recognition module. The input is digital voice data and the output is text data. In this voice recognition process, speech-to-text conversion is performed using a speech recognition API (e.g., Google's speech recognition API).

[1020] Step 3:

[1021] The server receives the text data sent from the speech recognition module and analyzes it using a speech analysis engine. In this step, the input text data is analyzed to determine whether certain words or phrases are inappropriate.

[1022] Step 4:

[1023] A generative AI model is used to convert text data in which inappropriate remarks have been detected into kinder words. The server searches a database for kinder words that correspond to the inappropriate words and updates the text data. The input here is the analyzed text data, and the output is new text data converted into kinder words. For example, Huggingface's twitter-roberta-large-offensive model is used as the generative AI model.

[1024] Step 5:

[1025] The server passes the new text data to a speech synthesis module, which converts it back into speech data. The input is the converted text data, and the output is speech data. This speech synthesis is performed using a speech synthesis library such as pyttsx3.

[1026] Step 6:

[1027] The server sends the newly generated voice data to the terminal and distributes it to the other party in real time. The input here is the voice data from the speech synthesis module, and the output is the voice data distributed to the recipient. This allows the listener to hear the voice in real time with inappropriate remarks corrected.

[1028] Through these steps, inappropriate comments made by users are converted into kinder words in real time and delivered.

[1029] 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.

[1030] This invention relates to a system that detects inappropriate comments in real time on social networking sites and distribution platforms and converts them into more appropriate and gentle language. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, the system converts them into more appropriate and gentle language, improving the quality of communication.

[1031] Basic system configuration

[1032] 1. Voice input

[1033] The user speaks into the microphone. For example, the user says, "You're an idiot!"

[1034] The device captures this audio and obtains the data as a digital signal.

[1035] 2. Audio analysis

[1036] The terminal transmits the acquired voice data to the server.

[1037] The server uses a voice analysis engine to convert the voice data into text data.

[1038] The server analyzes this text data and identifies inappropriate words and phrases, for example, detecting "idiot" in the text "You're an idiot!" as an inappropriate word.

[1039] 3. Detecting inappropriate comments

[1040] The server flags inappropriate words and phrases identified by the speech analysis engine.

[1041] The flagged text is treated as data indicating inappropriate language.

[1042] 4. Emotion recognition

[1043] The device sends the voice data to the emotion engine.

[1044] The emotion engine analyzes the user's emotions from the voice data and generates emotion data.

[1045] The emotion data analyzed by the emotion engine is used to select subsequent kind words. For example, if a user says "idiot" with an angry emotion, the emotion data includes "anger."

[1046] 5. Correction of statements

[1047] The server receives the flagged text and searches a database for benign words that correspond to the inappropriate phrases.

[1048] The server adjusts its choice of kind words based on the emotional data, for example, converting "stupid" into "I think that's a bit off," while converting "let's calm down and think about it" if the emotional data indicates "anger."

[1049] The server generates the corrected text data and records it as new text data.

[1050] 6. Audio Distribution

[1051] The server passes the corrected text data to a speech synthesis engine, which converts it back into speech data.

[1052] The device receives new audio data and distributes it to listeners in real time.

[1053] For example, if a user says "You're an idiot!" and the emotion is analyzed as "anger," the listener will hear a gentle and appropriate response: "Calm down and think about it!"

[1054] Specific examples

[1055] If the user emotionally says, "You're an idiot!"

[1056] 1. Voice input

[1057] The user speaks into the microphone, "You're an idiot!"

[1058] The device captures this audio and transmits it as a digital signal to the server.

[1059] 2. Audio analysis

[1060] The server uses a voice analysis engine to convert "You're an idiot!" into text data.

[1061] The server parses this text and identifies the word "idiot" as inappropriate.

[1062] 3. Detecting inappropriate comments

[1063] The server flags "idiots."

[1064] 4. Emotion recognition

[1065] The device sends the voice data to the emotion engine.

[1066] The emotion engine extracts the emotion of "anger" from the voice data.

[1067] 5. Correction of statements

[1068] The server converts "idiot" into "calm down and think about it."

[1069] The finished text will be "You need to calm down and think about it!"

[1070] 6. Audio Distribution

[1071] The server passes the corrected text to a speech synthesis engine to convert it into voice data.

[1072] The device receives new audio and distributes it to listeners.

[1073] Usage scene

[1074] This system can be applied to a wide range of situations, such as live streaming and interactive social networking sites. By combining it with an emotion engine, it is possible to instantly convert inappropriate comments that users may unconsciously make into appropriate and kind words, which is expected to improve the quality of communication and prevent conflicts between users.

[1075] The processing flow will be explained below.

[1076] Program processing

[1077] Step 1:

[1078] The user speaks into the microphone.

[1079] Example: "You're an idiot!"

[1080] Step 2:

[1081] The device captures the user's voice with a microphone.

[1082] Step 3:

[1083] The device converts the captured audio data into a digital format and stores it in a buffer.

[1084] Step 4:

[1085] The terminal transmits the audio data stored in the buffer to the server.

[1086] Step 5:

[1087] The server passes the received voice data to the voice analysis engine.

[1088] Step 6:

[1089] The voice analysis engine converts the voice data into text (voice recognition).

[1090] Step 7:

[1091] A speech analysis engine analyzes text data and identifies inappropriate words and phrases.

[1092] Step 8:

[1093] A speech analysis engine identifies and flags inappropriate words and expressions based on specific criteria.

[1094] Example: Identifying "idiot" as an inappropriate word in the text "You're an idiot!"

[1095] Step 9:

[1096] The server receives the flagged text data.

[1097] Step 10:

[1098] The device sends the voice data to the emotion engine.

[1099] Step 11:

[1100] The emotion engine analyzes the user's emotions from the voice data and generates emotion data.

[1101] Step 12:

[1102] The emotion engine sends the emotion analysis results to the server.

[1103] Example: The emotion engine detects "anger."

[1104] Step 13:

[1105] The server refers to the emotion data when replacing inappropriate words with kinder words.

[1106] Example: "Stupid" can be added to "I think that's a bit wrong," and if the emotion is "anger," it can be changed to "Let's calm down and think about it."

[1107] Step 14:

[1108] The server generates the modified text data.

[1109] For example: "You're an idiot!" becomes "You need to calm down and think!"

[1110] Step 15:

[1111] The server passes the corrected text data to the speech synthesis engine.

[1112] Step 16:

[1113] The speech synthesis engine converts the text data into speech data (speech synthesis).

[1114] Step 17:

[1115] The server transmits the generated voice data to the terminal.

[1116] Step 18:

[1117] The audio data received by the terminal is played on the speaker.

[1118] For example, if a user says, "You're an idiot!" and the emotion engine analyzes this as "anger," the final voice the listener hears will be a gentle and appropriate "You need to calm down and think about it!" This series of processes instantly converts inappropriate remarks into gentler words, enabling healthy communication in real time.

[1119] Example 2

[1120] 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."

[1121] On conventional social networking sites and distribution platforms, problems could arise due to inappropriate comments made by users. Simply detecting and filtering inappropriate comments is difficult to completely prevent, making it impossible to improve the quality of communication. Furthermore, simply converting words without considering the user's feelings can lead to the problem of not accurately conveying the intent of the comment.

[1122] 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.

[1123] In this invention, the server includes means for acquiring voice through a microphone, means for analyzing the acquired voice and detecting inappropriate remarks, means for automatically converting the detected inappropriate remarks into kinder words, means for analyzing the user's emotions and generating emotion data, means for adjusting the selection of kinder words based on the emotion data, and means for delivering the converted voice to the other party. This makes it possible to appropriately detect and correct the user's remarks in real time and flexibly convert words taking the user's emotions into consideration. This can prevent problems caused by inappropriate remarks and improve the quality of communication.

[1124] The "means for acquiring voice through a microphone" refers to a hardware and software device that captures the voice spoken by the user into the microphone as a digital signal and converts it into a form that can be processed within the system.

[1125] The "means for analyzing the acquired voice and detecting inappropriate remarks" refers to a device and program that uses a voice analysis engine to convert the acquired voice data into text data, and then analyzes the text data to identify inappropriate words and phrases.

[1126] "Means for automatically converting detected inappropriate language to kinder language" means devices and programs that have a database and algorithms for replacing words or phrases identified as inappropriate with kinder language.

[1127] The "means for analyzing user emotions and generating emotion data" refers to a device and program that uses an emotion recognition engine to analyze user emotions from voice data and output the analysis results as emotion data.

[1128] The "means for adjusting the selection of kind words based on emotional data" refers to a device and a program that selects appropriate kind words based on the generated emotional data and converts inappropriate words into the most appropriate words in accordance with the emotion.

[1129] The "means for delivering the converted voice to the other party" refers to a device and a program that reconverts the text data converted into friendly words into voice data and delivers the voice data to the listener in real time.

[1130] This invention relates to a system that detects inappropriate comments in real time on social media and distribution platforms and converts them into more appropriate and gentle language. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, the system converts them into more appropriate and gentle language, improving the quality of communication.

[1131] Basic system configuration

[1132] 1. Voice input

[1133] The user speaks into the microphone. For example, the user says, "You're an idiot!"

[1134] The device captures this audio and obtains the data as a digital signal.

[1135] 2. Audio analysis

[1136] The terminal transmits the acquired voice data to the server.

[1137] The server uses a speech analysis engine (e.g., Google Cloud Speech-to-Text) to convert the voice data into text data.

[1138] The server analyzes this text data and identifies inappropriate words and phrases, for example, detecting "idiot" in the text "You're an idiot!" as an inappropriate word.

[1139] 3. Detecting inappropriate comments

[1140] The server flags inappropriate words and phrases identified by the speech analysis engine.

[1141] The flagged text is treated as data indicating inappropriate language.

[1142] 4. Emotion recognition

[1143] The device sends the voice data to an emotion engine (e.g., IBM Watson Tone Analyzer).

[1144] The emotion engine analyzes the user's emotions from the voice data and generates emotion data. For example, if a user says "idiot" in anger, the emotion data will include "anger."

[1145] 5. Correction of statements

[1146] The server receives the flagged text and searches a database for benign words that correspond to the inappropriate phrases.

[1147] The server adjusts its choice of kind words based on the emotional data, for example, converting "stupid" into "I think that's a bit off," while converting "let's calm down and think about it" if the emotional data indicates "anger."

[1148] The server generates the corrected text data and records it as new text data.

[1149] 6. Audio Distribution

[1150] The server passes the corrected text data to a speech synthesis engine (e.g., Amazon Polly) and converts it back into speech data.

[1151] The device receives new audio data and delivers it to the listener in real time. For example, a voice saying "You're an idiot!" can be changed to "Calm down and think!"

[1152] Specific examples

[1153] If the user emotionally says, "You're an idiot!"

[1154] 1. Voice input

[1155] The user speaks into the microphone, "You're an idiot!"

[1156] The device captures this audio and transmits it as a digital signal to the server.

[1157] 2. Audio analysis

[1158] The server uses a speech analysis engine to convert "You're an idiot!" into text data.

[1159] The server parses this text and identifies the word "idiot" as inappropriate.

[1160] 3. Detecting inappropriate comments

[1161] The server flags "idiots."

[1162] 4. Emotion recognition

[1163] The device sends the voice data to the emotion engine.

[1164] The emotion engine extracts the emotion of "anger" from the voice data.

[1165] 5. Correction of statements

[1166] The server converts "idiot" into "calm down and think about it."

[1167] The finished text will be "You need to calm down and think about it!"

[1168] 6. Audio Distribution

[1169] The server passes the corrected text to a speech synthesis engine to convert it into voice data.

[1170] The device receives the new audio and broadcasts it to the listener, who hears the gentle words, "Stay calm and think!"

[1171] Prompt Sentence Examples

[1172] An example of a specific prompt would be:

[1173] If a user says "You're an idiot!", convert the speech data into text, detect the inappropriate phrase "idiot", analyze the emotion of anger, and then convert it into "You need to calm down and think!"

[1174] This prompt allows the system to properly analyze the user's speech content and emotions and make appropriate conversions.

[1175] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1176] Step 1: Capturing Audio Input

[1177] Input: User-generated speech

[1178] Output: Audio data as a digital signal

[1179] The device uses a microphone to capture the user's voice. For example, if the user says "You're an idiot!", the microphone captures this voice in real time. The device converts the captured voice into a digital signal and sends it to the server.

[1180] Step 2: Convert audio data to text

[1181] Input: Audio data as a digital signal

[1182] Output: Text data

[1183] The server processes the received voice data. Specifically, it uses a voice analysis engine (e.g., Google Cloud Speech-to-Text) to convert the voice data into text data. For example, the voice data "You're an idiot!" is converted into text data "You're an idiot!"

[1184] Step 3: Detect inappropriate language

[1185] Input: Text data

[1186] Output: Text data with inappropriate words flagged

[1187] The server analyzes the converted text data. Specifically, it uses a natural language processing engine to identify inappropriate words and phrases in the text. For example, the server flags the inappropriate words and phrases in the text "You're an idiot!" and passes the flagged text on to the next processing step.

[1188] Step 4: Emotion Recognition

[1189] Input: Audio data

[1190] Output: Emotion data

[1191] The device sends the collected voice data to an emotion recognition engine (e.g., IBM Watson Tone Analyzer). The emotion engine analyzes the user's emotions from the voice data and generates emotion data. For example, if a user says "idiot" out of anger, the emotion data will include "anger." The generated emotion data is sent to the server.

[1192] Step 5: Modify your statement

[1193] Input: Text and sentiment data with inappropriate words flagged

[1194] Output: Corrected text data

[1195] The server receives text data with inappropriate words flagged and emotion data. Specifically, it analyzes the text data and searches a database for appropriate kind words. For example, it converts "idiot" into "calm down and think about it." The server adjusts the choice of kind words based on the emotion data. Finally, it generates corrected text data and records it as new text data.

[1196] Step 6: Synthesize and deliver audio

[1197] Input: Modified text data

[1198] Output: New audio data

[1199] The server receives the corrected text data and passes it to a speech synthesis engine (e.g., Amazon Polly). The speech synthesis engine converts the corrected text into speech data. For example, "You're an idiot!" is converted into "Calm down and think about it!" The device receives the newly generated speech data and delivers it to the listener in real time. The listener can hear the gentle words "Calm down and think about it!"

[1200] (Application example 2)

[1201] 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."

[1202] In recent years, with the spread of live streaming and real-time communication, inappropriate comments have become a problem. Inappropriate comments can hurt people, cause trouble, and hinder healthy communication. Furthermore, due to a lack of correction based on the speaker's emotions, mechanical conversion is often insufficient. To address this issue, there is a need for a system that converts comments into appropriate words based on emotions in real time.

[1203] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1204] In this invention, the server includes means for acquiring voice through a microphone, means for analyzing the acquired voice and detecting inappropriate remarks, means for recognizing the user's emotions when an inappropriate remark is detected, means for converting the detected inappropriate remarks into kinder words based on the recognized emotions, and means for delivering the converted voice to the other party. This makes it possible to convert inappropriate remarks into appropriate words in real time, taking the user's emotions into consideration.

[1205] (definition statement)

[1206] A "microphone" is a device that converts sound into an electrical signal and is used to capture what a user says.

[1207] The "means for analyzing voice and detecting inappropriate remarks" refers to a means for analyzing acquired voice data and identifying inappropriate words and phrases therefrom.

[1208] The "means for recognizing emotions" is a means for analyzing the user's emotions from the acquired voice data and generating emotion data.

[1209] The "means for converting to kinder words" is a means for automatically replacing inappropriate remarks with kinder words, and also a means for selecting appropriate expressions based on emotional data.

[1210] The "means for reconverting to audio" is a means for converting the corrected text data into audio data.

[1211] The "means for processing in real time" refers to a means for immediately processing acquired voice data without delay.

[1212] "Means for delivering to the other party in real time" refers to means for instantly transmitting the corrected audio data to the listener or the other party.

[1213] The present invention is a system for detecting inappropriate comments in real time and converting them into friendly comments. This system can be applied when users are live streaming.

[1214] Basic system configuration

[1215] The system is implemented using the following hardware and software:

[1216] Microphone: A device that converts the user's voice into an electrical signal and captures the sound.

[1217] Terminal: A device such as a computer or smartphone that analyzes audio data acquired from a microphone.

[1218] Server: A remote server containing a speech analysis engine, emotion recognition engine, text-to-speech engine, and speech synthesis engine.

[1219] System operation explanation

[1220] 1. Audio capture:

[1221] The user speaks into a microphone and transmits the audio in real time to the device, where an application on the device captures the audio and obtains the data as a digital signal.

[1222] 2. Audio analysis:

[1223] The device sends the captured voice data to the server, which uses a speech analysis engine (e.g., Google Speech Recognition API) on the server to convert the voice data into text data. A text analysis algorithm (e.g., TextBlob) is used to identify inappropriate words and phrases.

[1224] 3. Emotion recognition:

[1225] The emotion recognition engine analyzes the user's emotions from the voice data and generates emotion data. For emotion recognition, a voice emotion recognition library, for example, is used.

[1226] 4. Statement correction:

[1227] The server receives the flagged text and searches a database for a kinder equivalent to the inappropriate phrase. Based on the emotion data, it adjusts the choice of kinder words. For example, it converts "stupid" to "calm down and think about it," but if the emotion data indicates "anger," it converts it to "think again."

[1228] 5. Speech synthesis:

[1229] The converted text data is converted back into audio data using a speech synthesis engine (e.g., Google Text-to-Speech). The device receives the new audio data and delivers it to the listener in real time.

[1230] Specific examples

[1231] For example, if a user says "You're an idiot!" during a stream, the system works as follows: This utterance is converted in real time into "Calm down and think!" and broadcast as audio to the viewer.

[1232] Example prompts to input to the generative AI model

[1233] "I'd like to develop an application that converts speech signals into text data and modifies inappropriate expressions in the text based on emotion recognition. Please generate the code to accomplish this process in real time in a voice-delivery environment using the SpeechRecognition library, Google Speech Recognition API, emotion recognition library, and Google Text-to-Speech library."

[1234] This system can instantly detect inappropriate comments made during broadcasts and appropriately change them based on emotions, enabling healthy communication.

[1235] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1236] (Program processing flow)

[1237] Step 1:

[1238] Audio Capture

[1239] During live streaming, the user speaks into the microphone. The device captures this voice and obtains it as a digital signal. The input is the user's voice signal, and the output is the captured digital voice data. In this step, the device uses the microphone to capture voice in real time.

[1240] Step 2:

[1241] Audio data transmission

[1242] The device sends the captured audio data to the server. The input is the captured digital audio data, and the output is the audio data sent to the server. In this step, the device uploads the audio data to the server.

[1243] Step 3:

[1244] Audio analysis

[1245] The server converts the acquired voice data into text data using a voice analysis engine (e.g., Google Speech Recognition API). The input is the voice data sent to the server, and the output is the converted text data. In this step, the server recognizes the voice and generates the result as text.

[1246] Step 4:

[1247] Inappropriate speech detection

[1248] The server uses a text analysis algorithm (e.g., TextBlob) to detect inappropriate content from the converted text data. The input is text data, and the output is text data with inappropriate words and phrases flagged. In this step, the server analyzes the text data to identify inappropriate content.

[1249] Step 5:

[1250] emotion recognition

[1251] The server uses an emotion recognition engine (e.g., a voice emotion recognition library) to analyze the user's emotion from the voice data. The input is voice data, and the output is emotion data. In this step, the server analyzes the emotion of the voice and determines the emotion the user expressed when speaking.

[1252] Step 6:

[1253] Correction of statement

[1254] The server converts inappropriate comments into appropriate ones based on the emotion data. The input is text data with inappropriate words and emotion data flagged, and the output is the corrected text data. In this step, the server replaces the inappropriate comments with more appropriate words based on the emotion.

[1255] Step 7:

[1256] Speech synthesis

[1257] The server converts the corrected text data back into speech data using a speech synthesis engine (e.g., Google Text-to-Speech API). The input is the corrected text data, and the output is the synthesized speech data. In this step, the server converts the text data back into speech.

[1258] Step 8:

[1259] Audio data distribution

[1260] The device receives new audio data obtained from the server and delivers it to the listener in real time. The input is the synthesized audio data, and the output is the modified audio delivered to the listener. In this step, the device delivers the new audio in real time.

[1261] 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.

[1262] 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.

[1263] 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.

[1264] [Fourth embodiment]

[1265] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1266] 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.

[1267] 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).

[1268] 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.

[1269] 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.

[1270] 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).

[1271] 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.

[1272] 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.

[1273] 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.

[1274] 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.

[1275] 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.

[1276] 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.

[1277] 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."

[1278] This invention relates to a system that detects inappropriate comments on social media and distribution platforms in real time and converts them into more gentle comments. This system is realized through a series of processes: voice input, voice analysis, detection of inappropriate comments, comment correction, and voice distribution.

[1279] Basic system configuration

[1280] 1. Voice input

[1281] The user uses a microphone to input speech. For example, consider the user saying "You're an idiot!"

[1282] The device captures this audio as a digital signal and obtains the data.

[1283] 2. Audio analysis

[1284] The terminal transmits the acquired voice data to the server.

[1285] The server uses a voice analysis engine to convert the voice data into text data.

[1286] The server analyzes this text data and identifies inappropriate words and phrases, for example, detecting "idiot" in the text "You're an idiot!" as an inappropriate word.

[1287] 3. Detecting inappropriate comments

[1288] The server flags inappropriate words and phrases identified by the speech analysis engine.

[1289] The flagged text is treated as data indicating inappropriate language.

[1290] 4. Correction of statements

[1291] The server receives the flagged text and searches a database for a more benign equivalent for the inappropriate phrase, for example, changing "stupid" to "I think that's a bit off."

[1292] The server generates the corrected text and records it as new text data.

[1293] 5. Audio Distribution

[1294] The server passes the corrected text data to a speech synthesis engine, which converts it back into speech data.

[1295] The device receives new audio data and distributes it to listeners in real time.

[1296] For example, if the user says, "You're an idiot!", the listener will hear the gentler, "I think you're a bit different!"

[1297] Specific examples

[1298] If a user says, "You're an idiot!"

[1299] 1. Voice input

[1300] The user speaks into the microphone, "You're an idiot!"

[1301] The device captures this audio and transmits it as a digital signal to the server.

[1302] 2. Audio analysis

[1303] The server uses a voice analysis engine to convert "You're an idiot!" into text data.

[1304] The server parses this text and identifies the word "idiot" as inappropriate.

[1305] 3. Detecting inappropriate comments

[1306] The server flags "idiots."

[1307] 4. Correction of statements

[1308] The server converts "idiot" to "I think that's a bit off."

[1309] The finished text is "I think you're a little different!"

[1310] 5. Audio Distribution

[1311] The server passes the corrected text to a speech synthesis engine to convert it into voice data.

[1312] The device receives new audio and distributes it to listeners.

[1313] Usage scene

[1314] This system can be applied to a wide range of situations, including live streaming and interactive social networking sites. By using this system, inappropriate comments that users may make unconsciously can be instantly converted into gentler words, which is expected to improve the quality of communication and prevent conflicts between users.

[1315] The processing flow will be explained below.

[1316] Program processing

[1317] Step 1:

[1318] The user speaks into the microphone.

[1319] Example: "You're an idiot!"

[1320] Step 2:

[1321] The device captures the user's voice with a microphone.

[1322] Step 3:

[1323] The device converts the captured audio data into a digital format and stores it in a buffer.

[1324] Step 4:

[1325] The terminal transmits the audio data stored in the buffer to the server.

[1326] Step 5:

[1327] The server passes the received voice data to the voice analysis engine.

[1328] Step 6:

[1329] The voice analysis engine converts the voice data into text (voice recognition).

[1330] Step 7:

[1331] A speech analysis engine analyzes text data and identifies inappropriate words and phrases.

[1332] Step 8:

[1333] A speech analysis engine identifies and flags inappropriate words and expressions based on specific criteria.

[1334] Example: Identifying "idiot" as an inappropriate word in the text "You're an idiot!"

[1335] Step 9:

[1336] The server receives the flagged text data.

[1337] Step 10:

[1338] The server replaces flagged inappropriate words with kinder words from a database.

[1339] Example: "Stupid" changed to "I think that's a bit different."

[1340] Step 11:

[1341] The server generates the modified text data.

[1342] For example: "You're an idiot!" turns into "I think you're a bit different!"

[1343] Step 12:

[1344] The server passes the corrected text data to the speech synthesis engine.

[1345] Step 13:

[1346] The speech synthesis engine converts the text data into speech data (speech synthesis).

[1347] Step 14:

[1348] The server transmits the generated voice data to the terminal.

[1349] Step 15:

[1350] The audio data received by the terminal is played on the speaker.

[1351] For example, if a user says, "You're an idiot!", the server converts this speech into "I think you're a little bit different!" The listener hears a gentler version of "I think you're a little bit different!" Through this series of processes, inappropriate remarks are converted into gentler words in real time, maintaining healthy communication.

[1352] Example 1

[1353] 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."

[1354] Today, many people communicate in real time on platforms such as social media and live streaming. However, inappropriate comments can be made during these interactions, causing trouble between users. A system that can avoid such situations and improve the quality of communication is needed.

[1355] 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.

[1356] In this invention, the server includes means for converting voice data into text data, means for analyzing the text data to detect inappropriate comments, and means for replacing the detected inappropriate comments with kinder words selected from a database, thereby making it possible to prevent trouble between users and improve the quality of communication.

[1357] A "microphone" is a device that converts sound into an electrical signal.

[1358] A "means for converting audio into a digital signal" is a process or device for converting analog audio into digital data.

[1359] A "means for transmitting to a server over the Internet" is a method or device for transmitting a digital signal to a remote server using a network.

[1360] A "server" is a computer system for processing and storing data.

[1361] "Means for converting into text data" refers to a process or system for converting voice data into a string of characters.

[1362] A "means for analyzing text data" is a method or device for analyzing text data and understanding or classifying its content.

[1363] An "inappropriate language detection means" is a method or apparatus for identifying inappropriate words or phrases from the analyzed text data.

[1364] A "database" is a system for storing a collection of data in an organized manner and for retrieving and updating it as needed.

[1365] A "means for replacing with kinder words" is a method or device for converting detected inappropriate expressions into appropriate expressions.

[1366] A "means for reconverting to audio data" is a process or system for converting text data into audio format.

[1367] A "terminal" is a computer device operated by a user, such as a PC or smartphone.

[1368] A "natural language processing engine" is software or a system for understanding and processing human language.

[1369] A "speech synthesis engine" is software or a system that converts input text into voice data.

[1370] The present invention relates to a system that detects inappropriate comments in real time on social media and distribution platforms and converts them into more polite language. Specific ways in which the present invention can be implemented are described below.

[1371] Audio input and digital signal conversion

[1372] Users input their voice using a microphone. This can be the built-in microphone of a PC or smartphone, or an external microphone. The device captures this voice and converts it into a digital signal. Specifically, audio capture software running on the device converts analog voice into digital data.

[1373] Sending and analyzing voice data

[1374] The device sends the captured digital voice data to a server via the Internet. An application running on the device sends the voice data to the server as a POST request via a dedicated API. The server then passes the received voice data to a voice analysis engine (e.g., Google Cloud Speech-to-Text) and converts the voice into text data.

[1375] The converted text data is then analyzed by a natural language processing (NLP) engine (e.g., spaCy or TensorFlow), which identifies inappropriate words and phrases within the text. For example, the statement "You're an idiot!" is converted to the text "You're an idiot!", with "idiot" identified as an inappropriate word.

[1376] Detecting and correcting inappropriate language

[1377] After detecting an inappropriate remark, the server searches a database (e.g., MySQL or PostgreSQL) for appropriate corrections. The detected inappropriate phrases are replaced with corresponding more benign words. For example, "idiot" can be converted to "I think you're a bit off" to generate the new text "I think you're a bit off!".

[1378] Recoding and audio distribution

[1379] The server passes the corrected text data to a speech synthesis engine (for example, Google Text-to-Speech or Amazon Polly), which reconverts it into new voice data. This new voice data is then sent back to the device and delivered to the listener. This allows, for example, a statement like "You're an idiot!" to be converted into a more gentle statement like "I think you're a little different!" and delivered.

[1380] Specific examples

[1381] Specifically, when a user says "You're an idiot!", the following process occurs:

[1382] 1. The user says into the microphone, "You're an idiot!"

[1383] 2. The device captures this audio and converts it into a digital signal.

[1384] 3. The digital signal is sent over the internet to a server.

[1385] 4. The server uses a speech analysis engine to convert the speech into text, generating the text "You're an idiot!"

[1386] 5. The server analyzes the text with an NLP engine and identifies the word "idiot" as inappropriate.

[1387] 6. The server converts "idiot" to "I think you're a little bit off" and generates the text "I think you're a little bit off!"

[1388] 7. The corrected text is converted back into voice data by the speech synthesis engine.

[1389] 8. This new audio data is sent to the device and distributed to the listener.

[1390] The above is a specific embodiment for carrying out the present invention. By using this system, inappropriate comments made by users can be instantly converted into kinder words, thereby improving the quality of communication.

[1391] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1392] Step 1:

[1393] The user uses the microphone to input speech, for example, saying "You're an idiot!"

[1394] The device converts this audio into a digital signal. The input is an analog audio signal, and the output is digital audio data. A process is performed to convert the analog audio into digital audio data. Specifically, the device's audio capture software converts the audio into a digital signal.

[1395] Step 2:

[1396] The device sends the captured digital audio data to a server over the Internet. The input is the digital audio data, and the output is the data sent to the server over the network. Specifically, an application on the device uses an API to send the audio data to the server as a POST request.

[1397] Step 3:

[1398] The server passes the received voice data to a voice analysis engine, which converts the voice into text data. The input is digital voice data, and the output is text data. A process is performed to convert the voice data into text data. Specifically, a voice analysis engine on the server (for example, Google Cloud Speech-to-Text) converts the voice data into text.

[1399] Step 4:

[1400] The server analyzes the converted text data using a natural language processing (NLP) engine to identify inappropriate words and phrases. The input is text data, and the output is text data with inappropriate words and phrases identified. The text data is analyzed and processed to identify inappropriate words and phrases. Specifically, an NLP engine (for example, spaCy) analyzes the text "You're an idiot!" and identifies "idiot" as an inappropriate word.

[1401] Step 5:

[1402] The server replaces the identified inappropriate words with kinder words from a database. The input is text data with inappropriate words flagged, and the output is the text data replaced with kinder words. Data processing is performed to correct the text data. Specifically, "idiot" is converted to "I think that's a bit off" to generate the text "I think you're a bit off!"

[1403] Step 6:

[1404] The server passes the corrected text data to a speech synthesis engine, which converts it into new speech data. The input is the corrected text data, and the output is the new speech data. A process is performed to reconvert the text data into speech data. Specifically, the speech synthesis engine (for example, Google Text-to-Speech) converts the text "I think you're a little off!" into speech data.

[1405] Step 7:

[1406] The server sends the generated new audio data to the device and distributes it to the listener in real time. The input is the new audio data, and the output is the audio received by the listener. Processing to distribute the audio data is performed. Specifically, the audio playback application on the device plays and distributes the new audio data to the listener.

[1407] (Application example 1)

[1408] 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."

[1409] In conventional social networking sites and streaming platforms, inappropriate comments can be made, causing trouble between users. However, it is difficult to detect and correct inappropriate comments in real time, which can potentially ruin the experience for streamers and viewers. Therefore, this invention aims to promote healthy communication and improve the user experience by detecting inappropriate comments in real time and converting them into appropriate words.

[1410] 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.

[1411] In this invention, the server includes means for acquiring voice through a microphone, means for analyzing the acquired voice using a voice recognition module to detect inappropriate remarks, means for automatically converting the detected inappropriate remarks into kinder words using a generative AI model, means for converting the converted text data back into voice using a voice synthesis module, means for delivering the converted voice to the other party in real time, and database means for providing prompt sentences to the generative AI model, thereby enabling the detection and correction of inappropriate remarks in real time.

[1412] A "microphone" is an input device for converting sound into an electrical signal.

[1413] A "voice recognition module" is a software or hardware component for converting captured speech into text data.

[1414] A "generative AI model" is a trained model that includes an algorithm that uses natural language processing technology to convert inappropriate remarks into kinder ones.

[1415] A "speech synthesis module" is a system component for converting text data into speech data.

[1416] A "prompt sentence" is an input sentence provided to a generative AI model to generate an appropriate output.

[1417] A "database" is a collection of data that systematically organizes information so that it can be quickly searched and used.

[1418] "Real-time" means that the processing occurs immediately, without delay.

[1419] This invention is implemented using a system including a microphone, a speech recognition module, a generative AI model, a speech synthesis module, and a database. Specifically, the invention can be implemented using the following configuration and process.

[1420] The user first inputs their voice through the microphone. This voice is picked up by the device and captured as a digital signal. The captured voice data is converted into text data using a speech recognition module and sent to the server. This speech recognition module can be commercial software such as Google's speech recognition API.

[1421] The server uses a speech analysis engine to analyze the text data generated by the speech recognition module and detect inappropriate remarks. This is done using a generative AI model. The generative AI model has the ability to detect inappropriate remarks and replace them with appropriate expressions. A specific example of this model is the "twitter-roberta-large-offensive" model based on Huggingface's transformers library. The generative AI model is linked to a database that provides prompt sentences, and this database contains pre-registered inappropriate phrases and their replacement candidates.

[1422] The text data of the detected inappropriate remarks is automatically converted into kinder words by a generative AI model. For example, the word "idiot" is converted to "I think that's a bit off." This updates the text data to new text data that uses appropriate words. This new text data is saved by the server and converted back into voice data using a voice synthesis module. A voice synthesis library such as pyttsx3 can be used as the voice synthesis module.

[1423] The server sends the voice data generated by the voice synthesis module to the terminal and distributes it to the other party in real time, allowing the listener to hear the voice with inappropriate remarks corrected.

[1424] As a specific example of its use, if a user says "You're an idiot!" during a live broadcast or game broadcast, this system will convert it into "I think you're a little different!" and broadcast it immediately. Through this process, healthy communication between the broadcaster and the viewers is maintained.

[1425] An example of input to a generative AI model is:

[1426] Input voice: You're an idiot!

[1427] Output Voice: I think you're a little different!

[1428] In this way, a system using the present invention can instantly detect and correct inappropriate comments during live broadcasts, realizing healthy communication.

[1429] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1430] Step 1:

[1431] The user inputs voice through the microphone, which is then picked up by the device and captured as a digital signal. At this stage, the voice is stored on the device as concrete digital data.

[1432] Step 2:

[1433] The device converts the captured voice data into text data using a voice recognition module. The input is digital voice data and the output is text data. In this voice recognition process, speech-to-text conversion is performed using a speech recognition API (e.g., Google's speech recognition API).

[1434] Step 3:

[1435] The server receives the text data sent from the speech recognition module and analyzes it using a speech analysis engine. In this step, the input text data is analyzed to determine whether certain words or phrases are inappropriate.

[1436] Step 4:

[1437] A generative AI model is used to convert text data in which inappropriate remarks have been detected into kinder words. The server searches a database for kinder words that correspond to the inappropriate words and updates the text data. The input here is the analyzed text data, and the output is new text data converted into kinder words. For example, Huggingface's twitter-roberta-large-offensive model is used as the generative AI model.

[1438] Step 5:

[1439] The server passes the new text data to a speech synthesis module, which converts it back into speech data. The input is the converted text data, and the output is speech data. This speech synthesis is performed using a speech synthesis library such as pyttsx3.

[1440] Step 6:

[1441] The server sends the newly generated voice data to the terminal and distributes it to the other party in real time. The input here is the voice data from the speech synthesis module, and the output is the voice data distributed to the recipient. This allows the listener to hear the voice in real time with inappropriate remarks corrected.

[1442] Through these steps, inappropriate comments made by users are converted into kinder words in real time and delivered.

[1443] 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.

[1444] This invention relates to a system that detects inappropriate comments in real time on social networking sites and distribution platforms and converts them into more appropriate and gentle language. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, the system converts them into more appropriate and gentle language, improving the quality of communication.

[1445] Basic system configuration

[1446] 1. Voice input

[1447] The user speaks into the microphone. For example, the user says, "You're an idiot!"

[1448] The device captures this audio and obtains the data as a digital signal.

[1449] 2. Audio analysis

[1450] The terminal transmits the acquired voice data to the server.

[1451] The server uses a voice analysis engine to convert the voice data into text data.

[1452] The server analyzes this text data and identifies inappropriate words and phrases, for example, detecting "idiot" in the text "You're an idiot!" as an inappropriate word.

[1453] 3. Detecting inappropriate comments

[1454] The server flags inappropriate words and phrases identified by the speech analysis engine.

[1455] The flagged text is treated as data indicating inappropriate language.

[1456] 4. Emotion recognition

[1457] The device sends the voice data to the emotion engine.

[1458] The emotion engine analyzes the user's emotions from the voice data and generates emotion data.

[1459] The emotion data analyzed by the emotion engine is used to select subsequent kind words. For example, if a user says "idiot" with an angry emotion, the emotion data includes "anger."

[1460] 5. Correction of statements

[1461] The server receives the flagged text and searches a database for benign words that correspond to the inappropriate phrases.

[1462] The server adjusts its choice of kind words based on the emotional data, for example, converting "stupid" into "I think that's a bit off," while converting "let's calm down and think about it" if the emotional data indicates "anger."

[1463] The server generates the corrected text data and records it as new text data.

[1464] 6. Audio Distribution

[1465] The server passes the corrected text data to a speech synthesis engine, which converts it back into speech data.

[1466] The device receives new audio data and distributes it to listeners in real time.

[1467] For example, if a user says "You're an idiot!" and the emotion is analyzed as "anger," the listener will hear a gentle and appropriate response: "Calm down and think about it!"

[1468] Specific examples

[1469] If the user emotionally says, "You're an idiot!"

[1470] 1. Voice input

[1471] The user speaks into the microphone, "You're an idiot!"

[1472] The device captures this audio and transmits it as a digital signal to the server.

[1473] 2. Audio analysis

[1474] The server uses a voice analysis engine to convert "You're an idiot!" into text data.

[1475] The server parses this text and identifies the word "idiot" as inappropriate.

[1476] 3. Detecting inappropriate comments

[1477] The server flags "idiots."

[1478] 4. Emotion recognition

[1479] The device sends the voice data to the emotion engine.

[1480] The emotion engine extracts the emotion of "anger" from the voice data.

[1481] 5. Correction of statements

[1482] The server converts "idiot" into "calm down and think about it."

[1483] The finished text will be "You need to calm down and think about it!"

[1484] 6. Audio Distribution

[1485] The server passes the corrected text to a speech synthesis engine to convert it into voice data.

[1486] The device receives new audio and distributes it to listeners.

[1487] Usage scene

[1488] This system can be applied to a wide range of situations, such as live streaming and interactive social networking sites. By combining it with an emotion engine, it is possible to instantly convert inappropriate comments that users may unconsciously make into appropriate and kind words, which is expected to improve the quality of communication and prevent conflicts between users.

[1489] The processing flow will be explained below.

[1490] Program processing

[1491] Step 1:

[1492] The user speaks into the microphone.

[1493] Example: "You're an idiot!"

[1494] Step 2:

[1495] The device captures the user's voice with a microphone.

[1496] Step 3:

[1497] The device converts the captured audio data into a digital format and stores it in a buffer.

[1498] Step 4:

[1499] The terminal transmits the audio data stored in the buffer to the server.

[1500] Step 5:

[1501] The server passes the received voice data to the voice analysis engine.

[1502] Step 6:

[1503] The voice analysis engine converts the voice data into text (voice recognition).

[1504] Step 7:

[1505] A speech analysis engine analyzes text data and identifies inappropriate words and phrases.

[1506] Step 8:

[1507] A speech analysis engine identifies and flags inappropriate words and expressions based on specific criteria.

[1508] Example: Identifying "idiot" as an inappropriate word in the text "You're an idiot!"

[1509] Step 9:

[1510] The server receives the flagged text data.

[1511] Step 10:

[1512] The device sends the voice data to the emotion engine.

[1513] Step 11:

[1514] The emotion engine analyzes the user's emotions from the voice data and generates emotion data.

[1515] Step 12:

[1516] The emotion engine sends the emotion analysis results to the server.

[1517] Example: The emotion engine detects "anger."

[1518] Step 13:

[1519] The server refers to the emotion data when replacing inappropriate words with kinder words.

[1520] Example: "Stupid" can be added to "I think that's a bit wrong," and if the emotion is "anger," it can be changed to "Let's calm down and think about it."

[1521] Step 14:

[1522] The server generates the modified text data.

[1523] For example: "You're an idiot!" becomes "You need to calm down and think!"

[1524] Step 15:

[1525] The server passes the corrected text data to the speech synthesis engine.

[1526] Step 16:

[1527] The speech synthesis engine converts the text data into speech data (speech synthesis).

[1528] Step 17:

[1529] The server transmits the generated voice data to the terminal.

[1530] Step 18:

[1531] The audio data received by the terminal is played on the speaker.

[1532] For example, if a user says, "You're an idiot!" and the emotion engine analyzes this as "anger," the final voice the listener hears will be a gentle and appropriate "You need to calm down and think about it!" This series of processes instantly converts inappropriate remarks into gentler words, enabling healthy communication in real time.

[1533] Example 2

[1534] 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."

[1535] On conventional social networking sites and distribution platforms, problems could arise due to inappropriate comments made by users. Simply detecting and filtering inappropriate comments is difficult to completely prevent, making it impossible to improve the quality of communication. Furthermore, simply converting words without considering the user's feelings can lead to the problem of not accurately conveying the intent of the comment.

[1536] 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.

[1537] In this invention, the server includes means for acquiring voice through a microphone, means for analyzing the acquired voice and detecting inappropriate remarks, means for automatically converting the detected inappropriate remarks into kinder words, means for analyzing the user's emotions and generating emotion data, means for adjusting the selection of kinder words based on the emotion data, and means for delivering the converted voice to the other party. This makes it possible to appropriately detect and correct the user's remarks in real time and flexibly convert words taking the user's emotions into consideration. This can prevent problems caused by inappropriate remarks and improve the quality of communication.

[1538] The "means for acquiring voice through a microphone" refers to a hardware and software device that captures the voice spoken by the user into the microphone as a digital signal and converts it into a form that can be processed within the system.

[1539] The "means for analyzing the acquired voice and detecting inappropriate remarks" refers to a device and program that uses a voice analysis engine to convert the acquired voice data into text data, and then analyzes the text data to identify inappropriate words and phrases.

[1540] "Means for automatically converting detected inappropriate language to kinder language" means devices and programs that have a database and algorithms for replacing words or phrases identified as inappropriate with kinder language.

[1541] The "means for analyzing user emotions and generating emotion data" refers to a device and program that uses an emotion recognition engine to analyze user emotions from voice data and output the analysis results as emotion data.

[1542] The "means for adjusting the selection of kind words based on emotional data" refers to a device and a program that selects appropriate kind words based on the generated emotional data and converts inappropriate words into the most appropriate words in accordance with the emotion.

[1543] The "means for delivering the converted voice to the other party" refers to a device and a program that reconverts the text data converted into friendly words into voice data and delivers the voice data to the listener in real time.

[1544] This invention relates to a system that detects inappropriate comments in real time on social media and distribution platforms and converts them into more appropriate and gentle language. Furthermore, by combining it with an emotion engine that recognizes the user's emotions, the system converts them into more appropriate and gentle language, improving the quality of communication.

[1545] Basic system configuration

[1546] 1. Voice input

[1547] The user speaks into the microphone. For example, the user says, "You're an idiot!"

[1548] The device captures this audio and obtains the data as a digital signal.

[1549] 2. Audio analysis

[1550] The terminal transmits the acquired voice data to the server.

[1551] The server uses a speech analysis engine (e.g., Google Cloud Speech-to-Text) to convert the voice data into text data.

[1552] The server analyzes this text data and identifies inappropriate words and phrases, for example, detecting "idiot" in the text "You're an idiot!" as an inappropriate word.

[1553] 3. Detecting inappropriate comments

[1554] The server flags inappropriate words and phrases identified by the speech analysis engine.

[1555] The flagged text is treated as data indicating inappropriate language.

[1556] 4. Emotion recognition

[1557] The device sends the voice data to an emotion engine (e.g., IBM Watson Tone Analyzer).

[1558] The emotion engine analyzes the user's emotions from the voice data and generates emotion data. For example, if a user says "idiot" in anger, the emotion data will include "anger."

[1559] 5. Correction of statements

[1560] The server receives the flagged text and searches a database for benign words that correspond to the inappropriate phrases.

[1561] The server adjusts its choice of kind words based on the emotional data, for example, converting "stupid" into "I think that's a bit off," while converting "let's calm down and think about it" if the emotional data indicates "anger."

[1562] The server generates the corrected text data and records it as new text data.

[1563] 6. Audio Distribution

[1564] The server passes the corrected text data to a speech synthesis engine (e.g., Amazon Polly) and converts it back into speech data.

[1565] The device receives new audio data and delivers it to the listener in real time. For example, a voice saying "You're an idiot!" can be changed to "Calm down and think!"

[1566] Specific examples

[1567] If the user emotionally says, "You're an idiot!"

[1568] 1. Voice input

[1569] The user speaks into the microphone, "You're an idiot!"

[1570] The device captures this audio and transmits it as a digital signal to the server.

[1571] 2. Audio analysis

[1572] The server uses a speech analysis engine to convert "You're an idiot!" into text data.

[1573] The server parses this text and identifies the word "idiot" as inappropriate.

[1574] 3. Detecting inappropriate comments

[1575] The server flags "idiots."

[1576] 4. Emotion recognition

[1577] The device sends the voice data to the emotion engine.

[1578] The emotion engine extracts the emotion of "anger" from the voice data.

[1579] 5. Correction of statements

[1580] The server converts "idiot" into "calm down and think about it."

[1581] The finished text will be "You need to calm down and think about it!"

[1582] 6. Audio Distribution

[1583] The server passes the corrected text to a speech synthesis engine to convert it into voice data.

[1584] The device receives the new audio and broadcasts it to the listener, who hears the gentle words, "Stay calm and think!"

[1585] Prompt Sentence Examples

[1586] An example of a specific prompt would be:

[1587] If a user says "You're an idiot!", convert the speech data into text, detect the inappropriate phrase "idiot", analyze the emotion of anger, and then convert it into "You need to calm down and think!"

[1588] This prompt allows the system to properly analyze the user's speech content and emotions and make appropriate conversions.

[1589] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1590] Step 1: Capturing Audio Input

[1591] Input: User-generated speech

[1592] Output: Audio data as a digital signal

[1593] The device uses a microphone to capture the user's voice. For example, if the user says "You're an idiot!", the microphone captures this voice in real time. The device converts the captured voice into a digital signal and sends it to the server.

[1594] Step 2: Convert audio data to text

[1595] Input: Audio data as a digital signal

[1596] Output: Text data

[1597] The server processes the received voice data. Specifically, it uses a voice analysis engine (e.g., Google Cloud Speech-to-Text) to convert the voice data into text data. For example, the voice data "You're an idiot!" is converted into text data "You're an idiot!"

[1598] Step 3: Detect inappropriate language

[1599] Input: Text data

[1600] Output: Text data with inappropriate words flagged

[1601] The server analyzes the converted text data. Specifically, it uses a natural language processing engine to identify inappropriate words and phrases in the text. For example, the server flags the inappropriate words and phrases in the text "You're an idiot!" and passes the flagged text on to the next processing step.

[1602] Step 4: Emotion Recognition

[1603] Input: Audio data

[1604] Output: Emotion data

[1605] The device sends the collected voice data to an emotion recognition engine (e.g., IBM Watson Tone Analyzer). The emotion engine analyzes the user's emotions from the voice data and generates emotion data. For example, if a user says "idiot" out of anger, the emotion data will include "anger." The generated emotion data is sent to the server.

[1606] Step 5: Modify your statement

[1607] Input: Text and sentiment data with inappropriate words flagged

[1608] Output: Corrected text data

[1609] The server receives text data with inappropriate words flagged and emotion data. Specifically, it analyzes the text data and searches a database for appropriate kind words. For example, it converts "idiot" into "calm down and think about it." The server adjusts the choice of kind words based on the emotion data. Finally, it generates corrected text data and records it as new text data.

[1610] Step 6: Synthesize and deliver audio

[1611] Input: Modified text data

[1612] Output: New audio data

[1613] The server receives the corrected text data and passes it to a speech synthesis engine (e.g., Amazon Polly). The speech synthesis engine converts the corrected text into speech data. For example, "You're an idiot!" is converted into "Calm down and think about it!" The device receives the newly generated speech data and delivers it to the listener in real time. The listener can hear the gentle words "Calm down and think about it!"

[1614] (Application example 2)

[1615] 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."

[1616] In recent years, with the spread of live streaming and real-time communication, inappropriate comments have become a problem. Inappropriate comments can hurt people, cause trouble, and hinder healthy communication. Furthermore, due to a lack of correction based on the speaker's emotions, mechanical conversion is often insufficient. To address this issue, there is a need for a system that converts comments into appropriate words based on emotions in real time.

[1617] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1618] In this invention, the server includes means for acquiring voice through a microphone, means for analyzing the acquired voice and detecting inappropriate remarks, means for recognizing the user's emotions when an inappropriate remark is detected, means for converting the detected inappropriate remarks into kinder words based on the recognized emotions, and means for delivering the converted voice to the other party. This makes it possible to convert inappropriate remarks into appropriate words in real time, taking the user's emotions into consideration.

[1619] (definition statement)

[1620] A "microphone" is a device that converts sound into an electrical signal and is used to capture what a user says.

[1621] The "means for analyzing voice and detecting inappropriate remarks" refers to a means for analyzing acquired voice data and identifying inappropriate words and phrases therefrom.

[1622] The "means for recognizing emotions" is a means for analyzing the user's emotions from the acquired voice data and generating emotion data.

[1623] The "means for converting to kinder words" is a means for automatically replacing inappropriate remarks with kinder words, and also a means for selecting appropriate expressions based on emotional data.

[1624] The "means for reconverting to audio" is a means for converting the corrected text data into audio data.

[1625] The "means for processing in real time" refers to a means for immediately processing acquired voice data without delay.

[1626] "Means for delivering to the other party in real time" refers to means for instantly transmitting the corrected audio data to the listener or the other party.

[1627] The present invention is a system for detecting inappropriate comments in real time and converting them into friendly comments. This system can be applied when users are live streaming.

[1628] Basic system configuration

[1629] The system is implemented using the following hardware and software:

[1630] Microphone: A device that converts the user's voice into an electrical signal and captures the sound.

[1631] Terminal: A device such as a computer or smartphone that analyzes audio data acquired from a microphone.

[1632] Server: A remote server containing a speech analysis engine, emotion recognition engine, text-to-speech engine, and speech synthesis engine.

[1633] System operation explanation

[1634] 1. Audio capture:

[1635] The user speaks into a microphone and transmits the audio in real time to the device, where an application on the device captures the audio and obtains the data as a digital signal.

[1636] 2. Audio analysis:

[1637] The device sends the captured voice data to the server, which uses a speech analysis engine (e.g., Google Speech Recognition API) on the server to convert the voice data into text data. A text analysis algorithm (e.g., TextBlob) is used to identify inappropriate words and phrases.

[1638] 3. Emotion recognition:

[1639] The emotion recognition engine analyzes the user's emotions from the voice data and generates emotion data. For emotion recognition, a voice emotion recognition library, for example, is used.

[1640] 4. Statement correction:

[1641] The server receives the flagged text and searches a database for a kinder equivalent to the inappropriate phrase. Based on the emotion data, it adjusts the choice of kinder words. For example, it converts "stupid" to "calm down and think about it," but if the emotion data indicates "anger," it converts it to "think again."

[1642] 5. Speech synthesis:

[1643] The converted text data is converted back into audio data using a speech synthesis engine (e.g., Google Text-to-Speech). The device receives the new audio data and delivers it to the listener in real time.

[1644] Specific examples

[1645] For example, if a user says "You're an idiot!" during a stream, the system works as follows: This utterance is converted in real time into "Calm down and think!" and broadcast as audio to the viewer.

[1646] Example prompts to input to the generative AI model

[1647] "I'd like to develop an application that converts speech signals into text data and modifies inappropriate expressions in the text based on emotion recognition. Please generate the code to accomplish this process in real time in a voice-delivery environment using the SpeechRecognition library, Google Speech Recognition API, emotion recognition library, and Google Text-to-Speech library."

[1648] This system can instantly detect inappropriate comments made during broadcasts and appropriately change them based on emotions, enabling healthy communication.

[1649] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1650] (Program processing flow)

[1651] Step 1:

[1652] Audio Capture

[1653] During live streaming, the user speaks into the microphone. The device captures this voice and obtains it as a digital signal. The input is the user's voice signal, and the output is the captured digital voice data. In this step, the device uses the microphone to capture voice in real time.

[1654] Step 2:

[1655] Audio data transmission

[1656] The device sends the captured audio data to the server. The input is the captured digital audio data, and the output is the audio data sent to the server. In this step, the device uploads the audio data to the server.

[1657] Step 3:

[1658] Audio analysis

[1659] The server converts the acquired voice data into text data using a voice analysis engine (e.g., Google Speech Recognition API). The input is the voice data sent to the server, and the output is the converted text data. In this step, the server recognizes the voice and generates the result as text.

[1660] Step 4:

[1661] Inappropriate speech detection

[1662] The server uses a text analysis algorithm (e.g., TextBlob) to detect inappropriate content from the converted text data. The input is text data, and the output is text data with inappropriate words and phrases flagged. In this step, the server analyzes the text data to identify inappropriate content.

[1663] Step 5:

[1664] emotion recognition

[1665] The server uses an emotion recognition engine (e.g., a voice emotion recognition library) to analyze the user's emotion from the voice data. The input is voice data, and the output is emotion data. In this step, the server analyzes the emotion of the voice and determines the emotion the user expressed when speaking.

[1666] Step 6:

[1667] Correction of statement

[1668] The server converts inappropriate comments into appropriate ones based on the emotion data. The input is text data with inappropriate words and emotion data flagged, and the output is the corrected text data. In this step, the server replaces the inappropriate comments with more appropriate words based on the emotion.

[1669] Step 7:

[1670] Speech synthesis

[1671] The server converts the corrected text data back into speech data using a speech synthesis engine (e.g., Google Text-to-Speech API). The input is the corrected text data, and the output is the synthesized speech data. In this step, the server converts the text data back into speech.

[1672] Step 8:

[1673] Audio data distribution

[1674] The device receives new audio data obtained from the server and delivers it to the listener in real time. The input is the synthesized audio data, and the output is the modified audio delivered to the listener. In this step, the device delivers the new audio in real time.

[1675] 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.

[1676] 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.

[1677] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1678] 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.

[1679] FIG. 9 illustrates 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 behaviors 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.

[1680] 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.

[1681] 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).

[1682] 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.

[1683] 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."

[1684] 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.

[1685] 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).

[1686] 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.

[1687] 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.

[1688] 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.

[1689] 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.

[1690] 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.

[1691] 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.

[1692] 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.

[1693] 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.

[1694] 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.

[1695] 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.

[1696] The following is further disclosed regarding the above embodiment.

[1697] (Claim 1)

[1698] a means for acquiring audio via a microphone;

[1699] means for analyzing the captured audio and detecting inappropriate speech;

[1700] A means to automatically convert detected inappropriate language into friendly language;

[1701] A means for delivering the converted audio to the other party;

[1702] A system including:

[1703] (Claim 2)

[1704] A means for converting speech into text data;

[1705] means for analyzing the text data to identify inappropriate words and phrases;

[1706] A means of replacing identified inappropriate words with kinder words;

[1707] means for converting the replaced text data back into speech;

[1708] 10. The system of claim 1, comprising:

[1709] (Claim 3)

[1710] means for processing the captured audio data in real time;

[1711] A means to update the audio stream in real time;

[1712] 10. The system of claim 1, comprising:

[1713] "Example 1"

[1714] (Claim 1)

[1715] a means for acquiring audio via a microphone;

[1716] means for converting the captured audio into a digital signal;

[1717] means for transmitting the converted audio to a server via the Internet;

[1718] A means for the server to convert the voice data into text data;

[1719] A means for the server to analyze the text data and detect inappropriate comments;

[1720] A means of replacing detected inappropriate comments with kinder words from a database;

[1721] means for reconverting the replaced text data into audio data;

[1722] A means for transmitting the reconverted voice data to the terminal and distributing it to the other party;

[1723] A system including:

[1724] (Claim 2)

[1725] means for processing the captured audio data in real time;

[1726] A means to update the audio stream in real time;

[1727] 10. The system of claim 1, comprising:

[1728] (Claim 3)

[1729] A means for analyzing text data using a natural language processing engine;

[1730] means for reconverting the voice data using a speech synthesis engine;

[1731] 10. The system of claim 1, comprising:

[1732] "Application Example 1"

[1733] (Claim 1)

[1734] a means for acquiring audio via a microphone;

[1735] A means for analyzing the acquired voice using a voice recognition module and detecting inappropriate remarks;

[1736] A means to automatically convert detected inappropriate remarks into friendly language using a generative AI model;

[1737] A means for converting the converted text data into speech again using a speech synthesis module;

[1738] A means to deliver the converted voice to the other party in real time,

[1739] a database means for providing prompt sentences to the generative AI model;

[1740] A system including:

[1741] (Claim 2)

[1742] A means for converting speech into text data;

[1743] means for analyzing the text data to identify inappropriate words and phrases;

[1744] A means of replacing identified inappropriate words with kinder words;

[1745] means for converting the replaced text data back into speech;

[1746] 10. The system of claim 1, comprising:

[1747] (Claim 3)

[1748] means for processing the captured audio data in real time;

[1749] A means to update the audio stream in real time;

[1750] 10. The system of claim 1, comprising:

[1751] "Example 2: Combining Emotion Engines"

[1752] (Claim 1)

[1753] a means for acquiring audio via a microphone;

[1754] means for analyzing the captured audio and detecting inappropriate speech;

[1755] A means to automatically convert detected inappropriate language into friendly language;

[1756] means for analyzing a user's emotions and generating emotion data;

[1757] a means for adjusting the selection of kind words based on the emotion data;

[1758] A means for delivering the converted audio to the other party;

[1759] A system including:

[1760] (Claim 2)

[1761] A means for converting speech into text data;

[1762] means for analyzing the text data to identify inappropriate words and phrases;

[1763] A means of replacing identified inappropriate words with kinder words;

[1764] means for converting the replaced text data back into speech;

[1765] means for modifying the text data based on the emotion data;

[1766] 10. The system of claim 1, comprising:

[1767] (Claim 3)

[1768] means for processing the captured audio data in real time;

[1769] A means to update the audio stream in real time;

[1770] 10. The system of claim 1, comprising:

[1771] "Application example 2 when combining emotion engines"

[1772] Claim amendment

[1773] (Claim 1)

[1774] a means for acquiring audio via a microphone;

[1775] means for analyzing the captured audio and detecting inappropriate speech;

[1776] means for recognizing a user's emotion upon detection of inappropriate speech;

[1777] A means for converting detected inappropriate language into more benign language based on the recognized sentiment;

[1778] A means for delivering the converted audio to the other party;

[1779] A system including:

[1780] (Claim 2)

[1781] A means for converting speech into text data;

[1782] means for analyzing the text data to identify inappropriate words and phrases;

[1783] a means for replacing the identified inappropriate words and phrases with kinder words based on the sentiment data;

[1784] means for converting the replaced text data back into speech;

[1785] 10. The system of claim 1, comprising:

[1786] (Claim 3)

[1787] means for processing the captured audio data in real time;

[1788] A means to update the audio stream in real time;

[1789] 10. The system of claim 1, including novel application-specific portions. [Explanation of symbols]

[1790] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. a means for acquiring audio via a microphone; means for analyzing the captured audio and detecting inappropriate speech; A means to automatically convert detected inappropriate language into friendly language; A means for delivering the converted audio to the other party; A system including:

2. A means for converting speech into text data; means for analyzing the text data to identify inappropriate words and phrases; A means of replacing identified inappropriate words with kinder words; means for converting the replaced text data back into speech; The system of claim 1 , comprising:

3. means for processing the captured audio data in real time; A means to update the audio stream in real time; The system of claim 1 , comprising:

Images