system

Abstract

We provide the system. [Solution] A voice acquisition means for acquiring audio during a call, A text conversion means for converting acquired audio into text, An analytical means for analyzing the converted text and determining the possibility of fraud, A means of displaying the possibility of fraud and the analysis results, A system that includes this.

Description

Technical Field

[0004] ,

[0006] , , , ,

[0005] , , , , ,

[0003] , , ,

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

Background Art

[0002] Patent Document 1 discloses a method for controlling a persona chatbot, which is performed by at least one processor, and includes steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to an explanation of a chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Special frauds (such as oreore fraud and refund fraud) have become a social problem, and many victims have occurred. In many cases, fraud calls are carried out skillfully, and especially the elderly and the like are easily deceived, so special countermeasures are necessary. In the conventional countermeasures, it often takes a long time to notice the fraud, and it is difficult to prevent the damage in advance. Therefore, there is a need for a system that can detect the possibility of special fraud in real time during a call and warn the user.

Means for Solving the Problems

[0005] The present invention provides a system that acquires voice during a call in real time and analyzes the possibility of fraud. Specifically, it includes the following means.

[0006] A means of acquiring audio during a call.

[0007] A text conversion means that converts acquired audio into text.

[0008] An analytical means that analyzes the converted text and determines the possibility of fraud.

[0009] A means of displaying the possibility of fraud and the results of the analysis,

[0010] These methods allow for real-time analysis of voice during calls, enabling immediate warnings to users if fraud is suspected. As a result, it becomes possible to prevent victims from becoming victims. Specifically, a system will be built that acquires voice using smart devices and performs conversion and analysis using natural language processing technology. This system will enable highly accurate detection of potential fraud and rapid response.

[0011] "Voice acquisition means" refers to a device or software that has the function of recording voice during a phone call and acquiring it as digital data.

[0012] "Text conversion means" refers to a device or software that has the function of processing acquired audio data and converting it into corresponding text data.

[0013] "Analysis means" refers to a device or software that has the function of analyzing converted text data and determining the possibility of special fraud based on its content.

[0014] "Display means" refers to a device or software that has the function of visually presenting the results obtained by the analysis means to the user.

[0015] A "smart device" is a portable information terminal capable of recording and transmitting digital audio, and specifically includes smart glasses and smartphones.

[0016] "Natural language processing" is a technology that enables computers to understand, interpret, and generate human language, and is particularly useful for converting speech to text and analyzing text data.

[0017] "Analysis result" refers to judgments and information obtained based on the content analysis of text data, and particularly refers to the judgment result regarding the possibility of fraud.

Brief Explanation of Drawings

[0018] [Figure 1] It is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] It is a conceptual diagram showing an example of the main functions of a data processing device and a smart device according to the first embodiment. [Figure 3] It is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] It is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] It is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] It is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] It is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] It is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] It shows an emotion map to which multiple emotions are mapped. [Figure 10] It shows an emotion map to which multiple emotions are mapped. [Figure 11] It is a sequence diagram showing the processing flow of the data processing system in Example 1. [Figure 12] It is a sequence diagram showing the processing flow of the data processing system in Application Example 1. [Figure 13] It is a sequence diagram showing the processing flow of the data processing system in Example 2 when the emotion engine is combined. [Figure 14] It is a sequence diagram showing the processing flow of the data processing system in Application Example 2 when the emotion engine is combined.

Mode for Carrying Out the Invention

[0019] Hereinafter, an example of an embodiment of the system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

[0020] First, the terms used in the following description will be explained.

[0021] In the following embodiments, the numbered processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), and the like.

[0022] In the following embodiments, the numbered RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a work memory by the processor.

[0023] In the following embodiments, the signed storage is one or more non-volatile storage devices that store various programs and various parameters. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes.

[0024] In the following embodiments, the signed communication interface (I / F) is an interface that includes a communication processor and an antenna, etc. The communication interface manages communication between multiple computers. Examples of communication standards applicable to the communication interface include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0025] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0026] [First Embodiment]

[0027] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0028] As shown in Figure 1, the data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0029] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0030] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0031] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0032] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0033] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0034] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0035] As shown in Figure 2, in the data processing device 12, a specific processing is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" related to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0036] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0037] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0038] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0039] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. This system consists of a smart device worn by the user, a server that processes the audio data, and a terminal that displays the results. A specific embodiment of this system is described below.

[0040] Voice acquisition method

[0041] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device. The smart device records the call audio in real time and prepares to send the data to a server.

[0042] Text conversion means

[0043] The server receives audio data sent from a smart device. The server uses a speech recognition module to convert the received audio data into text. This speech recognition module typically uses natural language processing technology. Specifically, it analyzes the audio data and converts its content into corresponding text data.

[0044] Analysis means

[0045] The server inputs the transcribed conversation into an analysis device. The analysis device uses a generative AI module to analyze the content of the text data and determine the possibility of fraud. This generative AI module uses a pre-trained model to determine whether the input text data constitutes fraudulent activity. If the analysis results indicate a high probability of fraudulent activity, it summarizes the content.

[0046] Display means

[0047] The server saves the analysis results to a database and sends the results to the user's terminal. The terminal receives the analysis results sent from the server and displays them to the user. In particular, if there is a high probability of fraud, the analysis results are displayed along with a warning message.

[0048] Examples

[0049] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and analyzes it using generative AI, determining that it is highly likely to be a scam. This result is immediately sent to the user's device, displaying the analysis results along with a warning message: "Warning: This call may be a scam!"

[0050] This allows users to identify potential scams in real time and prevent becoming a victim. This system is particularly useful for the elderly and users with limited knowledge of fraud, and is expected to have a positive effect on reducing social fraud.

[0051] The following describes the processing flow.

[0052] Step 1:

[0053] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. Then, if they suspect the call may be a scam, they enable the recording function on their smart device.

[0054] Step 2:

[0055] The smart device records the audio data during a call in real time. The recorded audio data is then prepared to be sent to a server over the network.

[0056] Step 3:

[0057] The server receives audio data sent from the smart device. The received audio data is temporarily stored as an audio file.

[0058] Step 4:

[0059] The server uses a speech recognition module to convert audio data into text. The speech recognition module analyzes the audio data and generates the corresponding text data.

[0060] Step 5:

[0061] The server provides the text-converted data to the analysis system. Specifically, it inputs the text data into a generative AI module and performs the analysis.

[0062] Step 6:

[0063] The server uses a generative AI module to analyze the content of the text data. This analysis determines whether it is potentially fraudulent. If it is highly likely to be fraudulent, it summarizes the content of the conversation.

[0064] Step 7:

[0065] The server saves the analysis results (determination of the likelihood of fraud and a summary of the conversation) to a database. The saved data includes the user ID, the likelihood of fraud, and the summary text.

[0066] Step 8:

[0067] The server sends the analysis results stored in the database to the user's terminal. Network communication is used for this transmission.

[0068] Step 9:

[0069] The terminal receives the analysis results sent from the server. The received data is then analyzed for presentation to the user.

[0070] Step 10:

[0071] The device displays the analysis results to the user. In particular, if there is a high probability of fraud, a summary of the conversation will be displayed along with the message, "Warning: This call may be a scam!"

[0072] This allows users to check in real time whether a call may be related to a scam and take necessary action.

[0073] (Example 1)

[0074] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0075] In modern society, the methods used in special fraud schemes are becoming more sophisticated and diverse, and victims, particularly the elderly and others with limited knowledge of fraud, are increasingly being targeted. It is difficult to immediately detect and notify victims of fraudulent activity through ordinary phone or video calls. Therefore, there is a need for means to prevent such fraud before it occurs.

[0076] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0077] In this invention, the server includes voice acquisition means for acquiring audio during a call, voice recognition means for converting the acquired audio into text, artificial intelligence analysis means for analyzing the converted text and determining the possibility of fraud, notification means for storing the possibility of fraud and the analysis results in a database and sending them to the user's terminal, and display means for displaying the possibility of fraud and the analysis results. This makes it possible to immediately analyze the possibility of fraud during a call and notify the user.

[0078] "Audio acquisition means for acquiring audio during a call" refers to a device or method for collecting audio in real time during a call and passing that data to subsequent processing.

[0079] A "speech recognition method that converts acquired speech into text" refers to a technology or module that receives speech data and converts it into a string of characters. Generally, it utilizes natural language processing technology.

[0080] "An artificial intelligence analysis method for analyzing converted text and determining the possibility of fraud" refers to artificial intelligence technology used to analyze text data and determine the possibility of fraudulent activity based on its content. It is executed using a pre-trained model.

[0081] "A notification method for storing potential fraud and analysis results in a database and sending them to the user's terminal" refers to a method or infrastructure for securely storing analysis results and notifying the user's terminal of necessary information in a timely manner.

[0082] "Display means for displaying the possibility of fraud and the analysis results" refers to a device or application for providing the user with the results of the analysis visually.

[0083] A "wearable device" is a device that a user can wear and that has functions such as making calls and recording audio. Examples include smart glasses and smartwatches.

[0084] System Overview

[0085] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. This system consists of a wearable device worn by the user, a server that processes the audio data, and a terminal that displays the analysis results.

[0086] Hardware and software to be used

[0087] Wearable devices: Smart glasses and smartphones

[0088] Server: A computer used for speech recognition and artificial intelligence analysis.

[0089] Speech recognition software: Common examples include Google® Cloud Speech-to-Text and IBM Watson® Speech to Text.

[0090] Artificial intelligence analysis software: Generative AI models (e.g., GPT-3® and BERT)

[0091] Device: Smartphone or tablet to display the analysis results.

[0092] Processing flow

[0093] 1. Voice acquisition:

[0094] The user puts on a wearable device (such as smart glasses or a smartphone) before starting a call. If the user suspects a scam during the call, they enable the recording function of the wearable device. The wearable device records the call audio in real time and sends the data to a server.

[0095] 2. Speech recognition:

[0096] The server receives audio data transmitted from the wearable device. The server first stores the audio data, and then converts it to text using Google Cloud Speech-to-Text or IBM Watson Speech to Text.

[0097] 3. Text analysis:

[0098] After the audio is converted to text, the server inputs that text data into a generating AI model (e.g., GPT-3 or BERT). The server then uses the pre-trained model to determine the likelihood of fraud.

[0099] Example of a prompt:

[0100] "The following is a transcript of a call the user received. Please analyze it and determine if it is a scam. Then, summarize your reasoning."

[0101] Call content: "Hello, this is the bank. Your account has been used fraudulently. Please tell us your account number and PIN so we can verify it."

[0102] 4. Saving and notifying of analysis results:

[0103] The server saves the analysis results to a database. It then sends the analysis results to the user's device. If the results are highly likely to be fraudulent, a warning message is sent along with them.

[0104] 5. Display of analysis results:

[0105] The device receives the analysis results sent from the server and notifies the user. In particular, if there is a high possibility of fraud, the analysis results will be displayed along with the message, "Warning: This call may be a scam!"

[0106] Specific example

[0107] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please tell us your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server uses Google Cloud Speech-to-Text to convert the speech to text and analyzes it with a generative AI model (e.g., GPT-3). If the server determines that the call is likely a scam, it immediately sends this result to the user's smartphone, displaying a message such as "Warning: This call may be a scam!" along with the analysis results.

[0108] This system allows users to identify potential scams in real time and prevent becoming a victim. This system is particularly useful for the elderly and users with limited knowledge of fraud, and is expected to reduce the overall amount of fraud damage in society.

[0109] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0110] Step 1: Voice Acquisition

[0111] The user puts on a wearable device (such as smart glasses or a smartphone) before starting a call. If the user suspects a scam during the call, they enable the recording function of the wearable device. The wearable device records the call audio in real time and sends the audio data to a server.

[0112] Input: User's action of starting recording and call audio

[0113] Output: Audio data sent to the server

[0114] Specific actions:

[0115] The user presses the record button on the smart glasses.

[0116] The smart glasses pick up the call content with their microphone and send the compressed audio data to the server.

[0117] Step 2: Speech Recognition

[0118] The server receives audio data transmitted from the wearable device. The server first stores the audio data, and then converts it to text using Google Cloud Speech-to-Text or IBM Watson Speech to Text.

[0119] Input: Audio data

[0120] Output: Text data

[0121] Specific actions:

[0122] The server saves the audio data as a temporary file.

[0123] The server calls the Google Cloud Speech-to-Text API to convert the audio data into text data.

[0124] Example text: "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification."

[0125] Step 3: Text Analysis

[0126] After the audio data is converted to text, the server inputs that text data into a generative AI model (e.g., GPT-3 or BERT). The server uses a pre-trained model to generate prompts to determine the likelihood of fraud and sends them to the generative AI model.

[0127] Input: Text data

[0128] Output: Analysis results regarding the possibility of fraud

[0129] Specific actions:

[0130] The server creates prompt messages for the generated AI model.

[0131] Prompt example: "The following is a transcript of a call the user received. Analyze it and determine if it is a scam. Then, summarize your reasoning."

[0132] Call content: "This is the bank. Your account has been used fraudulently. Please tell us your account number and PIN so we can verify it."

[0133] The server inputs text data into a generating AI model and analyzes its potential for fraud.

[0134] Example analysis result: "Probability of fraud: High"

[0135] Step 4: Saving and notifying of analysis results

[0136] The server saves the analysis results to a database. It then sends the analysis results to the user's device. If the results are highly likely to be fraudulent, a warning message is sent along with them.

[0137] Input: Analysis results

[0138] Output: Notification message to the user terminal

[0139] Specific actions:

[0140] The server saves the analysis results to the database as "Probability of fraud: High".

[0141] The server sends a message to the user's smartphone stating that it is "highly likely to be a scam."

[0142] Step 5: Displaying the analysis results

[0143] The device receives the analysis results sent from the server and notifies the user. In particular, if there is a high possibility of fraud, the analysis results will be displayed along with the message, "Warning: This call may be a scam!"

[0144] Input: Notification message from the server

[0145] Output: Warning messages and analysis results displayed on the user screen

[0146] Specific actions:

[0147] When the device receives a message sent from the server, a pop-up notification will appear stating, "Warning: This call may be a scam!"

[0148] Detailed analysis results will be displayed on the device screen.

[0149] (Application Example 1)

[0150] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0151] Conventional call content monitoring systems lacked the means to detect potentially fraudulent calls in real time and quickly warn users. As a result, elderly users and those with limited knowledge of fraud were particularly vulnerable to becoming victims of special fraud schemes. With current technology, there was no effective way to immediately inform users of potential fraudulent activity occurring during a call and prevent it.

[0152] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0153] In this invention, the server includes: an audio acquisition means for acquiring audio during a call; a text conversion means for converting the acquired audio into text; an analysis means for analyzing the converted text and determining the possibility of fraud; a display means for displaying the possibility of fraud and the analysis results; and a warning means for generating and outputting a warning message in real time regarding the possibility of fraud. This allows the user to immediately recognize situations with a high probability of fraud during a call and prevent becoming a victim.

[0154] A "voice acquisition means" is a device that has the function of recording voice during a call in real time and acquiring that voice data.

[0155] A "text conversion means" is a module that has the function of analyzing acquired audio data and converting its content into corresponding text data.

[0156] The "analysis method" is a system that uses the converted text data and natural language processing technologies such as generative AI models to determine the possibility of fraud.

[0157] "Display means" refers to an interface for visually conveying the results of the server's analysis to the user, and is a device or application for displaying analysis results and warning messages.

[0158] A "warning mechanism" is a system that generates a warning message in real time when there is a high probability of fraud and notifies the user of it via voice or visual means.

[0159] A "smart device" is a device that a user can wear to capture audio during a call, and it refers to a wide range of devices, including smartphones and smart glasses.

[0160] To implement this invention, a smart device worn by the user, a server that processes voice data, and a terminal that displays analysis results and warning messages are required.

[0161] System Configuration

[0162] 1. Means of acquiring sound:

[0163] The user puts on a smart device (e.g., a smartphone or smart glasses) before starting a call.

[0164] If a user suspects a call may be fraudulent, they can enable the recording function on their smart device. The smart device will record the call audio in real time and send the data to a server.

[0165] 2. Text conversion means:

[0166] The server receives voice data sent from the smart device.

[0167] The server uses a speech recognition module (e.g., the speech_recognition library) to convert speech data into text. The speech data is analyzed using natural language processing techniques, and its content is converted into text data.

[0168] 3. Analysis method:

[0169] The server uses a generative AI model (e.g., a deep learning model) to analyze the transcribed conversation content.

[0170] The analysis tool determines whether the input text data constitutes fraudulent activity. If there is a high probability of fraud, it summarizes the content and obtains the analysis result.

[0171] 4. Display means:

[0172] The server saves the analysis results to a database and sends the analysis results to the user's terminal.

[0173] The device receives the analysis results sent from the server and displays them to the user. If there is a high probability of fraud, the analysis results will be displayed along with a warning message.

[0174] 5. Warning measures:

[0175] If the server determines that there is a high probability of fraud, it generates a warning message in real time and notifies the user audibly or visually (e.g., converting text to speech using the gTTS library and playing it back with the playsound library).

[0176] Specific example

[0177] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please tell us your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and analyzes it using a generative AI model. If it is determined that the call is highly likely to be fraudulent, the result is immediately sent to the user's device, displaying the analysis results along with a warning message: "Warning: This call may be fraudulent!" This allows the user to understand the possibility of a scam in real time and prevent becoming a victim.

[0178] Example of a prompt

[0179] Please convert the recorded audio data to text and follow the instructions below:

[0180] 1. Send the converted text to the API endpoint.

[0181] 2. Based on the analysis results, a warning message will be generated if there is a high probability of fraud.

[0182] 3. Play the warning message as an audio.

[0183] example:

[0184] Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification.

[0185] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0186] Step 1:

[0187] The user puts on a smart device and starts a call. If the user suspects fraud during the call, they enable the recording function on their smart device. At this point, the voice acquisition process begins. The input data is the audio from the call, and the output data is the recorded audio. Specifically, recording starts when the user presses the record button on their smartphone or smart glasses.

[0188] Step 2:

[0189] The smart device transmits recorded audio data to the server in real time. The server receives this audio data. The input data is the recorded audio data, and the output data is the audio file stored on the server. Specifically, the smart device uploads the audio data to the server via the network connection.

[0190] Step 3:

[0191] The server converts the received audio data into text data using a speech recognition module. The `speech_recognition` library is used for this conversion. The input data is audio data, and the output data is the converted text data. Specifically, the speech recognition module analyzes the audio data and converts it into the corresponding text.

[0192] Step 4:

[0193] The server inputs the converted text data into the analysis device. The analysis device uses a generative AI model to analyze the text data and determine the likelihood of fraud. Natural language processing techniques are used in this process. The input data is text data, and the output data is the analysis result indicating the likelihood of fraud. Specifically, the generative AI model analyzes the text data and determines the likelihood of fraud.

[0194] Step 5:

[0195] The server saves the analysis results to a database and sends the results to the user's terminal. The input data is the analysis results, and the output data is the analysis results displayed on the user's terminal. Specifically, the server saves the analysis results to a database and then sends that data to the user's terminal via the network.

[0196] Step 6:

[0197] The device displays the received analysis results. In particular, if there is a high probability of fraud, the analysis results are displayed along with a warning message. The input data is the analysis results sent from the server, and the output data is the warning message and analysis results visually displayed to the user. Specifically, the device displays the received data on the screen, visually conveying the information to the user.

[0198] Step 7:

[0199] If the server determines that a case is highly likely to be fraudulent, it generates a warning message in real time and notifies the user of it audibly or visually. For example, the gTTS library is used to convert the warning message into audio, and the playsound library is used to play it. The input data is the analysis results, and the output data is the generated warning message. Specifically, the server generates a warning message based on the analysis results and notifies the user of it audibly or visually.

[0200] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0201] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. By further incorporating an emotion engine that recognizes the user's emotions, this system can determine the possibility of fraud with even greater accuracy. A specific embodiment of this system is shown below.

[0202] Voice acquisition method

[0203] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device. The smart device records the call audio in real time and sends the data to a server.

[0204] Text conversion means

[0205] The server receives audio data sent from the smart device. The received audio data is temporarily stored as an audio file. Next, a speech recognition module is used to convert the audio data into text. This speech recognition module analyzes the audio data and generates the corresponding text data.

[0206] Emotional Engine

[0207] The server uses the acquired voice data to activate the emotion engine. The emotion engine analyzes the user's emotions (e.g., tension, anxiety, confusion, etc.) from the voice data in real time. The analyzed emotion data is used as reference information for fraud analysis.

[0208] Analysis means

[0209] The server provides the text-converted data to the analysis system. Specifically, it inputs the text data into a generative AI module for analysis. This analysis uses a pre-trained model to determine whether the input text data constitutes fraudulent activity. Furthermore, sentiment data obtained from the sentiment engine is also taken into consideration to determine the likelihood of fraud with even greater accuracy.

[0210] Display means

[0211] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The terminal receives the analysis results sent from the server and displays them to the user. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message.

[0212] Examples

[0213] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and simultaneously analyzes the user's emotions from the audio data. Generative AI analyzes the text data, taking the emotions into consideration, to determine that there is a high probability of fraud. This result is immediately sent to the user's device, displaying a summary of the conversation along with a warning message: "Warning: This call may be a scam!"

[0214] This allows users to check in real time whether a call may be related to a scam and take necessary action. By combining this with an emotion engine, the user's emotional state is also used as a factor in the determination, allowing for a more accurate assessment of the likelihood of fraud.

[0215] The following describes the processing flow.

[0216] Step 1:

[0217] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device.

[0218] Step 2:

[0219] The smart device records the audio data during a call in real time. The recorded audio data is then prepared to be sent to a server over the network.

[0220] Step 3:

[0221] The server receives audio data sent from the smart device. The received audio data is temporarily stored as an audio file.

[0222] Step 4:

[0223] The server uses a speech recognition module to convert audio data into text. The speech recognition module analyzes the audio data and generates the corresponding text data.

[0224] Step 5:

[0225] The server uses the acquired audio data and the text data generated from that audio data to activate the emotion engine. The emotion engine analyzes the user's emotions in real time from the audio data and generates emotion data.

[0226] Step 6:

[0227] The server provides the text-converted data and sentiment data to the analysis system. Specifically, it inputs the text data into a generative AI module and performs analysis while taking sentiment data into consideration.

[0228] Step 7:

[0229] The server uses a generative AI module to analyze the content of the text data. This analysis determines whether it is potentially fraudulent. If it is highly likely to be fraudulent, it summarizes the content of the conversation.

[0230] Step 8:

[0231] The server stores the analysis results (determination of fraud possibility and conversation summary) and sentiment data in a database. The stored data includes user ID, fraud possibility, summary text, and sentiment data.

[0232] Step 9:

[0233] The server sends the analysis results stored in the database to the user's terminal. Network communication is used for this transmission.

[0234] Step 10:

[0235] The terminal receives the analysis results sent from the server. The received data is then analyzed for presentation to the user.

[0236] Step 11:

[0237] The device displays the analysis results to the user. In particular, if there is a high probability of fraud, a message such as "WARNING: This call may be a scam!" will be displayed along with a summary of the conversation and sentiment data.

[0238] Examples

[0239] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and simultaneously analyzes the user's emotions (e.g., tension, anxiety, confusion) from the audio data. Generative AI analyzes the text data, taking into account the emotion data, to determine that there is a high probability of fraud. This result is immediately sent to the user's device, displaying a warning message, "Warning: This call may be fraudulent!", along with a summary of the conversation and the emotion data. This allows the user to check in real time whether the call is potentially fraudulent and take the necessary action.

[0240] (Example 2)

[0241] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0242] Conventional in-call fraud detection systems rely solely on text analysis of call content to determine the likelihood of fraud, resulting in low accuracy and an inability to consider emotional factors such as user anxiety and tension. Furthermore, analysis results are not always presented in real time, preventing users from responding immediately. A system is needed that addresses these challenges and provides more accurate, real-time fraud detection.

[0243] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0244] In this invention, the server includes an audio acquisition means for acquiring audio during a call, a text conversion means for converting the acquired audio into text, an emotion analysis means for analyzing the user's emotions from the audio data, an analysis means for determining the possibility of fraud based on the converted text and the analyzed emotion data, and a display means for displaying the possibility of fraud and the analysis results. This enables highly accurate fraud detection in real time by simultaneously analyzing the audio during a call and the user's emotions.

[0245] "Voice acquisition means" refers to functions or devices for collecting and recording voice during a phone call in real time.

[0246] "Text conversion means" refers to functions and technologies for converting acquired audio data into text data.

[0247] "Emotional analysis means" refers to functions and technologies for analyzing and detecting a user's emotional state from voice data.

[0248] "Analysis means" refers to analytical functions and techniques that use converted text data and sentiment data to achieve a specific purpose (in this invention, the determination of the possibility of fraud).

[0249] "Display means" refers to functions or devices that visually present analysis results or warning messages to the user.

[0250] A "portable electronic device" is an electronic device that is easy to carry and has features such as voice acquisition during calls.

[0251] Natural language processing is a field of computer science and artificial intelligence that automatically analyzes, understands, and generates human language.

[0252] "Real-time analysis" is a process that processes and analyzes data immediately upon acquisition, providing results quickly.

[0253] This invention provides a system that acquires audio during a phone call in real time and analyzes the possibility of fraud, thereby assisting users in taking immediate action. This system also incorporates an emotion engine that recognizes the user's emotions, enabling more accurate determination of fraud potential. Specific embodiments of this system are described in detail below.

[0254] Voice acquisition method

[0255] The user puts on a portable electronic device (e.g., a smart device) before starting a call. If the user suspects during the call that it may be a scam, they enable the recording function on their smart device. The smart device (e.g., smart glasses, smartphone, etc.) records the call audio in real time and sends the data to a server.

[0256] Text conversion means

[0257] The server receives audio data sent from a smart device. The received audio data is temporarily stored as an audio file. Next, a speech recognition module is used to convert the audio data into text. Common speech recognition technologies such as IBM Watson Speech to Text and Google Cloud Speech-to-Text can be used as this speech recognition module.

[0258] Emotion analysis means

[0259] The server uses the acquired audio data to activate the emotion engine. The emotion engine analyzes the user's emotions (e.g., tension, anxiety, confusion, etc.) from the audio data in real time. This emotion analysis can use emotion analysis technologies such as Microsoft® Azure® Emotion API. The analyzed emotion data is used as reference information for fraud analysis.

[0260] Analysis means

[0261] The server provides the text-converted data to a generative AI model. Specifically, the text data is input into a generative AI model (e.g., GPT-3) for analysis. This analysis utilizes a pre-trained natural language processing model. Furthermore, sentiment data obtained from sentiment analysis is also taken into consideration to determine the likelihood of fraud with high accuracy.

[0262] Display means

[0263] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The terminal receives the analysis results sent from the server and displays them to the user. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message.

[0264] Examples

[0265] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." When the user starts recording, the smart glasses record the call audio and send the data to a server. The server uses IBM Watson Speech to Text to convert the audio to text and simultaneously analyzes the user's emotions from the audio data using the Microsoft Azure Emotion API. Next, a generative AI model (e.g., GPT-3) is used to analyze the text data, and taking the emotion data into consideration, it is determined that there is a high probability of it being a scam. This analysis result is immediately sent to the user's device, displaying a summary of the conversation along with the message, "Warning: This call may be a scam!"

[0266] For example, by inputting a prompt such as, "Is there a possibility that the content of this call is fraudulent?" into the AI ​​model, highly accurate fraud detection becomes possible.

[0267] This allows users to check in real time whether a call may be related to a scam and take necessary action. By combining this with an emotion engine, the user's emotional state is also used as a factor in the determination, allowing for a more accurate assessment of the likelihood of fraud.

[0268] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0269] Detailed explanation of the processing steps

[0270] Step 1:

[0271] Acquiring audio

[0272] 1. The user puts on a portable electronic device (e.g., smart glasses or a smartphone) before initiating a call.

[0273] 2. If a user suspects a scam during a call, enable the recording function on their smart device.

[0274] 3. The smart device records the call audio in real time and sends the audio data to the server.

[0275] Input: User's voice call.

[0276] Output: Real-time audio data sent to the server.

[0277] Step 2:

[0278] Receiving and temporarily storing audio data

[0279] 1. The server receives audio data transmitted from smart devices in real time.

[0280] 2. The server temporarily saves the received audio data in a file format (e.g., audio_data.wav).

[0281] Input: Voice data transmitted from a smart device.

[0282] Output: Voice file saved on the server.

[0283] Step 3:

[0284] Text conversion of voice data

[0285] 1. The server calls the speech recognition module to analyze the saved voice data.

[0286] [[ID=2,1]] 2. The server uses IBM Watson Speech to Text or Google Cloud Speech-to-Text to convert the voice data into text data.

[0287] Input: Saved voice file.

[0288] Output: Converted text data.

[0289] Step 4:

[0290] Sentiment analysis

[0291] 1. The server activates the sentiment analysis engine using the voice data after text conversion.

[0292] 2. The server uses the Microsoft Azure Emotion API or the like to analyze the user's sentiment from the voice data in real time.

[0293] Input: Converted text data and voice data.

[0294] Output: Analyzed sentiment data (e.g., nervousness, uneasiness, confusion).

[0295] Step 5:

[0296] Fraud Analysis

[0297] 1. The server accesses a pre-trained natural language processing model (e.g., GPT-3) to provide text data to the generative AI model.

[0298] 2. The server inputs the text data into the model and performs an analysis to determine the likelihood of fraud.

[0299] 3. The server also integrates sentiment analysis data to more accurately determine the likelihood of fraud.

[0300] Input: Text data and sentiment data.

[0301] Output: Analysis results regarding the likelihood of fraud.

[0302] Step ⑥:

[0303] Display of Analysis Results

[0304] 1. The server saves the determination of the likelihood of fraud and the summary of the conversation in the database.

[0305] 2. The server sends the analysis results to the user's terminal.

[0306] 3. The terminal displays the received analysis results to the user. Especially when the likelihood of fraud is high, it displays the summary of the conversation along with a warning message.

[0307] Input: Analysis results from the server.

[0308] Output: Warning message and analysis results displayed on the user's terminal.

[0309] (Application Example 2)

[0310] Next, Application Example 2 will be described. In the following description, the data processing device 12 is referred to as the "server", and the smart device 14 is referred to as the "terminal".

[0311] Currently, telephone scams and other types of fraud are rapidly increasing, resulting in a surge in victims among individuals and businesses. To address this, a system is needed that analyzes the content of phone calls to determine the likelihood of fraud. However, conventional systems struggle to accurately detect fraud based solely on simple voice data analysis, lacking consideration for the user's emotions and circumstances. Furthermore, they lack the functionality to notify users of analysis results in real time. To address these challenges, a more accurate and real-time fraud detection system is necessary.

[0312] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0313] In this invention, the server includes voice acquisition means, text conversion means, analysis means, sentiment analysis means, and display means. This makes it possible for the user to determine the possibility of fraud in real time during a call with high accuracy and to be notified of the result immediately.

[0314] "Voice acquisition means" refers to a device that has the function of recording audio during a phone call and acquiring that data.

[0315] A "text conversion means" is a device that has the function of analyzing acquired audio data and converting it into text data.

[0316] "Analysis means" refers to a device that has the function of analyzing and determining the possibility of fraud based on the converted text data.

[0317] An "emotion analysis device" is a device that analyzes a user's emotions from voice data and provides reference data for fraud detection.

[0318] A "display means" is a device that has the function of displaying analysis results and the possibility of fraud to the user.

[0319] A "generative AI model" is a type of AI technology that is trained on large amounts of data and performs analysis based on text data and sentiment data.

[0320] A "smart device" is a portable electronic device that has internet connectivity and various sensors, and is capable of interacting with the user.

[0321] "Natural language processing" is a general term for technologies that enable computers to understand and process human language.

[0322] "Real-time" refers to processing or updating immediately without delay or lag.

[0323] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. By combining this system with an emotion analysis engine that recognizes the user's emotions, the system can determine the possibility of fraud with higher accuracy. A specific embodiment of this system is shown below.

[0324] Voice acquisition method

[0325] The user uses a smart device (e.g., a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device. The smart device records the call audio in real time and sends the data to a server.

[0326] Text conversion means

[0327] The server receives audio data sent from a smart device. The received audio data is temporarily stored as an audio file. Next, a speech recognition module (for example, Google's speech recognition service) is used to convert the audio data into text. This speech recognition module analyzes the audio data and generates the corresponding text data.

[0328] Emotion analysis means

[0329] The server uses the acquired audio data to activate an emotion analysis system. This system utilizes the Hugging Face Transformers library to analyze the user's emotions (e.g., tension, anxiety, confusion) in real time from the audio data. The analyzed emotion data is used as reference information for fraud analysis.

[0330] Analysis means

[0331] The server provides the text-converted data to the analysis system. Specifically, it uses a generative AI model to input the text data and perform the analysis. In this analysis, a pre-trained generative AI model is used to determine whether the input text data constitutes fraudulent activity. Furthermore, sentiment data obtained from the sentiment analysis system is also taken into consideration to determine the likelihood of fraud with greater accuracy.

[0332] Display means

[0333] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The user's terminal receives and displays the analysis results sent from the server. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message.

[0334] Examples

[0335] For example, consider a scenario where a user receives a call on their smartphone saying, "Hello, this is a financial institution. Your account has been compromised. Please provide your account number and password for verification." The smartphone records the call and sends the audio data to a server. The server converts the audio to text and simultaneously analyzes the user's emotions (such as anxiety or tension) from the audio data. A generative AI model analyzes the text data and, taking the emotional data into consideration, determines that there is a high probability of fraud. This result is immediately sent to the user's device, displaying a summary of the conversation along with a warning message: "Warning: This call may be a scam!"

[0336] Examples of prompts to input into a generative AI model

[0337] Hello, this is a financial institution. Your account has been compromised. Please provide your account number and password for verification. Emotion: Anxiety Score: 0.85

[0338] By inputting the above prompt text into a generating AI model, it is possible to determine the likelihood of fraud with high accuracy.

[0339] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0340] Step 1:

[0341] A user initiates a call using a smart device. If the user suspects fraud during the call, they enable the recording function on their smart device. At this time, the smart device records the call audio in real time. The input is the call audio, and the output is audio data.

[0342] Step 2:

[0343] The smart device sends the recorded audio data to the server. The server receives the audio data and temporarily stores it as an audio file. The input is the audio data, and the output is the stored audio file.

[0344] Step 3:

[0345] The server uses a speech recognition module (for example, Google's speech recognition service) to convert the stored audio data into text. The speech recognition module analyzes the audio data and generates the corresponding text data. The input is an audio file, and the output is text data.

[0346] Step 4:

[0347] The server provides the generated text data to an emotion analysis tool (e.g., the Hugging Face Transformers library). The emotion analysis tool analyzes the text data and analyzes the user's emotions (e.g., tension, anxiety, confusion, etc.) in real time. The input is text data, and the output is emotion data.

[0348] Step 5:

[0349] The server inputs text data and sentiment data into a generative AI model to analyze the likelihood of fraud. The generative AI model determines whether an activity constitutes fraud based on the text data and sentiment data. A pre-trained generative AI model is used for this purpose. The input is text data and sentiment data, and the output is the determination of the likelihood of fraud.

[0350] Step 6:

[0351] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The user's terminal receives the analysis results and displays them to the user. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message. The input is the analysis results, and the output is the message displayed on the user's terminal.

[0352] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0353] Data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0354] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0355] [Second Embodiment]

[0356] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0357] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0358] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0359] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0360] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0361] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0362] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0363] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0364] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0365] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0366] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0367] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0368] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. This system consists of a smart device worn by the user, a server that processes the audio data, and a terminal that displays the results. A specific embodiment of this system is described below.

[0369] Voice acquisition method

[0370] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device. The smart device records the call audio in real time and prepares to send the data to a server.

[0371] Text conversion means

[0372] The server receives audio data sent from a smart device. The server uses a speech recognition module to convert the received audio data into text. This speech recognition module typically uses natural language processing technology. Specifically, it analyzes the audio data and converts its content into corresponding text data.

[0373] Analysis means

[0374] The server inputs the transcribed conversation into an analysis device. The analysis device uses a generative AI module to analyze the content of the text data and determine the possibility of fraud. This generative AI module uses a pre-trained model to determine whether the input text data constitutes fraudulent activity. If the analysis results indicate a high probability of fraudulent activity, it summarizes the content.

[0375] Display means

[0376] The server saves the analysis results to a database and sends the results to the user's terminal. The terminal receives the analysis results sent from the server and displays them to the user. In particular, if there is a high probability of fraud, the analysis results are displayed along with a warning message.

[0377] Examples

[0378] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and analyzes it using generative AI, determining that it is highly likely to be a scam. This result is immediately sent to the user's device, displaying the analysis results along with a warning message: "Warning: This call may be a scam!"

[0379] This allows users to identify potential scams in real time and prevent becoming a victim. This system is particularly useful for the elderly and users with limited knowledge of fraud, and is expected to have a positive effect on reducing social fraud.

[0380] The following describes the processing flow.

[0381] Step 1:

[0382] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. Then, if they suspect the call may be a scam, they enable the recording function on their smart device.

[0383] Step 2:

[0384] The smart device records the audio data during a call in real time. The recorded audio data is then prepared to be sent to a server over the network.

[0385] Step 3:

[0386] The server receives audio data sent from the smart device. The received audio data is temporarily stored as an audio file.

[0387] Step 4:

[0388] The server uses a speech recognition module to convert audio data into text. The speech recognition module analyzes the audio data and generates the corresponding text data.

[0389] Step 5:

[0390] The server provides the text-converted data to the analysis system. Specifically, it inputs the text data into a generative AI module and performs the analysis.

[0391] Step 6:

[0392] The server uses a generative AI module to analyze the content of the text data. This analysis determines whether it is potentially fraudulent. If it is highly likely to be fraudulent, it summarizes the content of the conversation.

[0393] Step 7:

[0394] The server saves the analysis results (determination of the likelihood of fraud and a summary of the conversation) to a database. The saved data includes the user ID, the likelihood of fraud, and the summary text.

[0395] Step 8:

[0396] The server sends the analysis results stored in the database to the user's terminal. Network communication is used for this transmission.

[0397] Step 9:

[0398] The terminal receives the analysis results sent from the server. The received data is then analyzed for presentation to the user.

[0399] Step 10:

[0400] The device displays the analysis results to the user. In particular, if there is a high probability of fraud, a summary of the conversation will be displayed along with the message, "Warning: This call may be a scam!"

[0401] This allows users to check in real time whether a call may be related to a scam and take necessary action.

[0402] (Example 1)

[0403] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0404] In modern society, the methods used in special fraud schemes are becoming more sophisticated and diverse, and victims, particularly the elderly and others with limited knowledge of fraud, are increasingly being targeted. It is difficult to immediately detect and notify victims of fraudulent activity through ordinary phone or video calls. Therefore, there is a need for means to prevent such fraud before it occurs.

[0405] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0406] In this invention, the server includes voice acquisition means for acquiring audio during a call, voice recognition means for converting the acquired audio into text, artificial intelligence analysis means for analyzing the converted text and determining the possibility of fraud, notification means for storing the possibility of fraud and the analysis results in a database and sending them to the user's terminal, and display means for displaying the possibility of fraud and the analysis results. This makes it possible to immediately analyze the possibility of fraud during a call and notify the user.

[0407] "Audio acquisition means for acquiring audio during a call" refers to a device or method for collecting audio in real time during a call and passing that data to subsequent processing.

[0408] A "speech recognition method that converts acquired speech into text" refers to a technology or module that receives speech data and converts it into a string of characters. Generally, it utilizes natural language processing technology.

[0409] "An artificial intelligence analysis method for analyzing converted text and determining the possibility of fraud" refers to artificial intelligence technology used to analyze text data and determine the possibility of fraudulent activity based on its content. It is executed using a pre-trained model.

[0410] "A notification method for storing potential fraud and analysis results in a database and sending them to the user's terminal" refers to a method or infrastructure for securely storing analysis results and notifying the user's terminal of necessary information in a timely manner.

[0411] "Display means for displaying the possibility of fraud and the analysis results" refers to a device or application for providing the user with the results of the analysis visually.

[0412] A "wearable device" is a device that a user can wear and that has functions such as making calls and recording audio. Examples include smart glasses and smartwatches.

[0413] System Overview

[0414] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. This system consists of a wearable device worn by the user, a server that processes the audio data, and a terminal that displays the analysis results.

[0415] Hardware and software to be used

[0416] Wearable devices: Smart glasses and smartphones

[0417] Server: A computer used for speech recognition and artificial intelligence analysis.

[0418] Speech recognition software: Common examples include Google Cloud Speech-to-Text and IBM Watson Speech to Text.

[0419] Artificial intelligence analysis software: Generative AI models (e.g., GPT-3 and BERT)

[0420] Device: Smartphone or tablet to display the analysis results.

[0421] Processing flow

[0422] 1. Voice acquisition:

[0423] The user puts on a wearable device (such as smart glasses or a smartphone) before starting a call. If the user suspects a scam during the call, they enable the recording function of the wearable device. The wearable device records the call audio in real time and sends the data to a server.

[0424] 2. Speech recognition:

[0425] The server receives audio data transmitted from the wearable device. The server first stores the audio data, and then converts it to text using Google Cloud Speech-to-Text or IBM Watson Speech to Text.

[0426] 3. Text analysis:

[0427] After the audio is converted to text, the server inputs that text data into a generating AI model (e.g., GPT-3 or BERT). The server then uses the pre-trained model to determine the likelihood of fraud.

[0428] Example of a prompt:

[0429] "The following is a transcript of a call the user received. Please analyze it and determine if it is a scam. Then, summarize your reasoning."

[0430] Call content: "Hello, this is the bank. Your account has been used fraudulently. Please tell us your account number and PIN so we can verify it."

[0431] 4. Saving and notifying of analysis results:

[0432] The server saves the analysis results to a database. It then sends the analysis results to the user's device. If the results are highly likely to be fraudulent, a warning message is sent along with them.

[0433] 5. Display of analysis results:

[0434] The device receives the analysis results sent from the server and notifies the user. In particular, if there is a high possibility of fraud, the analysis results will be displayed along with the message, "Warning: This call may be a scam!"

[0435] Specific example

[0436] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please tell us your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server uses Google Cloud Speech-to-Text to convert the speech to text and analyzes it with a generative AI model (e.g., GPT-3). If the server determines that the call is likely a scam, it immediately sends this result to the user's smartphone, displaying a message such as "Warning: This call may be a scam!" along with the analysis results.

[0437] This system allows users to identify potential scams in real time and prevent becoming a victim. This system is particularly useful for the elderly and users with limited knowledge of fraud, and is expected to reduce the overall amount of fraud damage in society.

[0438] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0439] Step 1: Voice Acquisition

[0440] The user puts on a wearable device (such as smart glasses or a smartphone) before starting a call. If the user suspects a scam during the call, they enable the recording function of the wearable device. The wearable device records the call audio in real time and sends the audio data to a server.

[0441] Input: User's action of starting recording and call audio

[0442] Output: Audio data sent to the server

[0443] Specific actions:

[0444] The user presses the record button on the smart glasses.

[0445] The smart glasses pick up the call content with their microphone and send the compressed audio data to the server.

[0446] Step 2: Speech Recognition

[0447] The server receives audio data transmitted from the wearable device. The server first stores the audio data, and then converts it to text using Google Cloud Speech-to-Text or IBM Watson Speech to Text.

[0448] Input: Audio data

[0449] Output: Text data

[0450] Specific actions:

[0451] The server saves the audio data as a temporary file.

[0452] The server calls the Google Cloud Speech-to-Text API to convert the audio data into text data.

[0453] Example text: "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification."

[0454] Step 3: Text Analysis

[0455] After the audio data is converted to text, the server inputs that text data into a generative AI model (e.g., GPT-3 or BERT). The server uses a pre-trained model to generate prompts to determine the likelihood of fraud and sends them to the generative AI model.

[0456] Input: Text data

[0457] Output: Analysis results regarding the possibility of fraud

[0458] Specific actions:

[0459] The server creates prompt messages for the generated AI model.

[0460] Prompt example: "The following is a transcript of a call the user received. Analyze it and determine if it is a scam. Then, summarize your reasoning."

[0461] Call content: "This is the bank. Your account has been used fraudulently. Please tell us your account number and PIN so we can verify it."

[0462] The server inputs text data into a generating AI model and analyzes its potential for fraud.

[0463] Example analysis result: "Probability of fraud: High"

[0464] Step 4: Saving and notifying of analysis results

[0465] The server saves the analysis results to a database. It then sends the analysis results to the user's device. If the results are highly likely to be fraudulent, a warning message is sent along with them.

[0466] Input: Analysis results

[0467] Output: Notification message to the user terminal

[0468] Specific actions:

[0469] The server saves the analysis results to the database as "Probability of fraud: High".

[0470] The server sends a message to the user's smartphone stating that it is "highly likely to be a scam."

[0471] Step 5: Displaying the analysis results

[0472] The device receives the analysis results sent from the server and notifies the user. In particular, if there is a high possibility of fraud, the analysis results will be displayed along with the message, "Warning: This call may be a scam!"

[0473] Input: Notification message from the server

[0474] Output: Warning messages and analysis results displayed on the user screen

[0475] Specific actions:

[0476] When the device receives a message sent from the server, a pop-up notification will appear stating, "Warning: This call may be a scam!"

[0477] Detailed analysis results will be displayed on the device screen.

[0478] (Application Example 1)

[0479] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0480] Conventional call content monitoring systems lacked the means to detect potentially fraudulent calls in real time and quickly warn users. As a result, elderly users and those with limited knowledge of fraud were particularly vulnerable to becoming victims of special fraud schemes. With current technology, there was no effective way to immediately inform users of potential fraudulent activity occurring during a call and prevent it.

[0481] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0482] In this invention, the server includes: an audio acquisition means for acquiring audio during a call; a text conversion means for converting the acquired audio into text; an analysis means for analyzing the converted text and determining the possibility of fraud; a display means for displaying the possibility of fraud and the analysis results; and a warning means for generating and outputting a warning message in real time regarding the possibility of fraud. This allows the user to immediately recognize situations with a high probability of fraud during a call and prevent becoming a victim.

[0483] A "voice acquisition means" is a device that has the function of recording voice during a call in real time and acquiring that voice data.

[0484] A "text conversion means" is a module that has the function of analyzing acquired audio data and converting its content into corresponding text data.

[0485] The "analysis method" is a system that uses the converted text data and natural language processing technologies such as generative AI models to determine the possibility of fraud.

[0486] "Display means" refers to an interface for visually conveying the results of the server's analysis to the user, and is a device or application for displaying analysis results and warning messages.

[0487] A "warning mechanism" is a system that generates a warning message in real time when there is a high probability of fraud and notifies the user of it via voice or visual means.

[0488] A "smart device" is a device that a user can wear to capture audio during a call, and it refers to a wide range of devices, including smartphones and smart glasses.

[0489] To implement this invention, a smart device worn by the user, a server that processes voice data, and a terminal that displays analysis results and warning messages are required.

[0490] System Configuration

[0491] 1. Means of acquiring sound:

[0492] The user puts on a smart device (e.g., a smartphone or smart glasses) before starting a call.

[0493] If a user suspects a call may be fraudulent, they can enable the recording function on their smart device. The smart device will record the call audio in real time and send the data to a server.

[0494] 2. Text conversion means:

[0495] The server receives voice data sent from the smart device.

[0496] The server uses a speech recognition module (e.g., the speech_recognition library) to convert speech data into text. The speech data is analyzed using natural language processing techniques, and its content is converted into text data.

[0497] 3. Analysis method:

[0498] The server uses a generative AI model (e.g., a deep learning model) to analyze the transcribed conversation content.

[0499] The analysis tool determines whether the input text data constitutes fraudulent activity. If there is a high probability of fraud, it summarizes the content and obtains the analysis result.

[0500] 4. Display means:

[0501] The server saves the analysis results to a database and sends the analysis results to the user's terminal.

[0502] The device receives the analysis results sent from the server and displays them to the user. If there is a high probability of fraud, the analysis results will be displayed along with a warning message.

[0503] 5. Warning measures:

[0504] If the server determines that there is a high probability of fraud, it generates a warning message in real time and notifies the user audibly or visually (e.g., converting text to speech using the gTTS library and playing it back with the playsound library).

[0505] Specific example

[0506] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please tell us your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and analyzes it using a generative AI model. If it is determined that the call is highly likely to be fraudulent, the result is immediately sent to the user's device, displaying the analysis results along with a warning message: "Warning: This call may be fraudulent!" This allows the user to understand the possibility of a scam in real time and prevent becoming a victim.

[0507] Example of a prompt

[0508] Please convert the recorded audio data to text and follow the instructions below:

[0509] 1. Send the converted text to the API endpoint.

[0510] 2. Based on the analysis results, a warning message will be generated if there is a high probability of fraud.

[0511] 3. Play the warning message as an audio.

[0512] example:

[0513] Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification.

[0514] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0515] Step 1:

[0516] The user puts on a smart device and starts a call. If the user suspects fraud during the call, they enable the recording function on their smart device. At this point, the voice acquisition process begins. The input data is the audio from the call, and the output data is the recorded audio. Specifically, recording starts when the user presses the record button on their smartphone or smart glasses.

[0517] Step 2:

[0518] The smart device transmits recorded audio data to the server in real time. The server receives this audio data. The input data is the recorded audio data, and the output data is the audio file stored on the server. Specifically, the smart device uploads the audio data to the server via the network connection.

[0519] Step 3:

[0520] The server converts the received audio data into text data using a speech recognition module. The `speech_recognition` library is used for this conversion. The input data is audio data, and the output data is the converted text data. Specifically, the speech recognition module analyzes the audio data and converts it into the corresponding text.

[0521] Step 4:

[0522] The server inputs the converted text data into the analysis device. The analysis device uses a generative AI model to analyze the text data and determine the likelihood of fraud. Natural language processing techniques are used in this process. The input data is text data, and the output data is the analysis result indicating the likelihood of fraud. Specifically, the generative AI model analyzes the text data and determines the likelihood of fraud.

[0523] Step 5:

[0524] The server saves the analysis results to a database and sends the results to the user's terminal. The input data is the analysis results, and the output data is the analysis results displayed on the user's terminal. Specifically, the server saves the analysis results to a database and then sends that data to the user's terminal via the network.

[0525] Step 6:

[0526] The device displays the received analysis results. In particular, if there is a high probability of fraud, the analysis results are displayed along with a warning message. The input data is the analysis results sent from the server, and the output data is the warning message and analysis results visually displayed to the user. Specifically, the device displays the received data on the screen, visually conveying the information to the user.

[0527] Step 7:

[0528] If the server determines that a case is highly likely to be fraudulent, it generates a warning message in real time and notifies the user of it audibly or visually. For example, the gTTS library is used to convert the warning message into audio, and the playsound library is used to play it. The input data is the analysis results, and the output data is the generated warning message. Specifically, the server generates a warning message based on the analysis results and notifies the user of it audibly or visually.

[0529] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0530] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. By further incorporating an emotion engine that recognizes the user's emotions, this system can determine the possibility of fraud with even greater accuracy. A specific embodiment of this system is shown below.

[0531] Voice acquisition method

[0532] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device. The smart device records the call audio in real time and sends the data to a server.

[0533] Text conversion means

[0534] The server receives audio data sent from the smart device. The received audio data is temporarily stored as an audio file. Next, a speech recognition module is used to convert the audio data into text. This speech recognition module analyzes the audio data and generates the corresponding text data.

[0535] Emotional Engine

[0536] The server uses the acquired voice data to activate the emotion engine. The emotion engine analyzes the user's emotions (e.g., tension, anxiety, confusion, etc.) from the voice data in real time. The analyzed emotion data is used as reference information for fraud analysis.

[0537] Analysis means

[0538] The server provides the text-converted data to the analysis system. Specifically, it inputs the text data into a generative AI module for analysis. This analysis uses a pre-trained model to determine whether the input text data constitutes fraudulent activity. Furthermore, sentiment data obtained from the sentiment engine is also taken into consideration to determine the likelihood of fraud with even greater accuracy.

[0539] Display means

[0540] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The terminal receives the analysis results sent from the server and displays them to the user. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message.

[0541] Examples

[0542] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and simultaneously analyzes the user's emotions from the audio data. Generative AI analyzes the text data, taking the emotions into consideration, to determine that there is a high probability of fraud. This result is immediately sent to the user's device, displaying a summary of the conversation along with a warning message: "Warning: This call may be a scam!"

[0543] This allows users to check in real time whether a call may be related to a scam and take necessary action. By combining this with an emotion engine, the user's emotional state is also used as a factor in the determination, allowing for a more accurate assessment of the likelihood of fraud.

[0544] The following describes the processing flow.

[0545] Step 1:

[0546] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device.

[0547] Step 2:

[0548] The smart device records the audio data during a call in real time. The recorded audio data is then prepared to be sent to a server over the network.

[0549] Step 3:

[0550] The server receives audio data sent from the smart device. The received audio data is temporarily stored as an audio file.

[0551] Step 4:

[0552] The server uses a speech recognition module to convert audio data into text. The speech recognition module analyzes the audio data and generates the corresponding text data.

[0553] Step 5:

[0554] The server uses the acquired audio data and the text data generated from that audio data to activate the emotion engine. The emotion engine analyzes the user's emotions in real time from the audio data and generates emotion data.

[0555] Step 6:

[0556] The server provides the text-converted data and sentiment data to the analysis system. Specifically, it inputs the text data into a generative AI module and performs analysis while taking sentiment data into consideration.

[0557] Step 7:

[0558] The server uses a generative AI module to analyze the content of the text data. This analysis determines whether it is potentially fraudulent. If it is highly likely to be fraudulent, it summarizes the content of the conversation.

[0559] Step 8:

[0560] The server stores the analysis results (determination of fraud possibility and conversation summary) and sentiment data in a database. The stored data includes user ID, fraud possibility, summary text, and sentiment data.

[0561] Step 9:

[0562] The server sends the analysis results stored in the database to the user's terminal. Network communication is used for this transmission.

[0563] Step 10:

[0564] The terminal receives the analysis results sent from the server. The received data is then analyzed for presentation to the user.

[0565] Step 11:

[0566] The device displays the analysis results to the user. In particular, if there is a high probability of fraud, a message such as "WARNING: This call may be a scam!" will be displayed along with a summary of the conversation and sentiment data.

[0567] Examples

[0568] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and simultaneously analyzes the user's emotions (e.g., tension, anxiety, confusion) from the audio data. Generative AI analyzes the text data, taking into account the emotion data, to determine that there is a high probability of fraud. This result is immediately sent to the user's device, displaying a warning message, "Warning: This call may be fraudulent!", along with a summary of the conversation and the emotion data. This allows the user to check in real time whether the call is potentially fraudulent and take the necessary action.

[0569] (Example 2)

[0570] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal".

[0571] Conventional in-call fraud detection systems rely solely on text analysis of call content to determine the likelihood of fraud, resulting in low accuracy and an inability to consider emotional factors such as user anxiety and tension. Furthermore, analysis results are not always presented in real time, preventing users from responding immediately. A system is needed that addresses these challenges and provides more accurate, real-time fraud detection.

[0572] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0573] In this invention, the server includes an audio acquisition means for acquiring audio during a call, a text conversion means for converting the acquired audio into text, an emotion analysis means for analyzing the user's emotions from the audio data, an analysis means for determining the possibility of fraud based on the converted text and the analyzed emotion data, and a display means for displaying the possibility of fraud and the analysis results. This enables highly accurate fraud detection in real time by simultaneously analyzing the audio during a call and the user's emotions.

[0574] "Voice acquisition means" refers to functions or devices for collecting and recording voice during a phone call in real time.

[0575] "Text conversion means" refers to functions and technologies for converting acquired audio data into text data.

[0576] "Emotional analysis means" refers to functions and technologies for analyzing and detecting a user's emotional state from voice data.

[0577] "Analysis means" refers to analytical functions and techniques that use converted text data and sentiment data to achieve a specific purpose (in this invention, the determination of the possibility of fraud).

[0578] "Display means" refers to functions or devices that visually present analysis results or warning messages to the user.

[0579] A "portable electronic device" is an electronic device that is easy to carry and has features such as voice acquisition during calls.

[0580] Natural language processing is a field of computer science and artificial intelligence that automatically analyzes, understands, and generates human language.

[0581] "Real-time analysis" is a process that processes and analyzes data immediately upon acquisition, providing results quickly.

[0582] This invention provides a system that acquires audio during a phone call in real time and analyzes the possibility of fraud, thereby assisting users in taking immediate action. This system also incorporates an emotion engine that recognizes the user's emotions, enabling more accurate determination of fraud potential. Specific embodiments of this system are described in detail below.

[0583] Voice acquisition method

[0584] The user puts on a portable electronic device (e.g., a smart device) before starting a call. If the user suspects during the call that it may be a scam, they enable the recording function on their smart device. The smart device (e.g., smart glasses, smartphone, etc.) records the call audio in real time and sends the data to a server.

[0585] Text conversion means

[0586] The server receives audio data sent from a smart device. The received audio data is temporarily stored as an audio file. Next, a speech recognition module is used to convert the audio data into text. Common speech recognition technologies such as IBM Watson Speech to Text and Google Cloud Speech-to-Text can be used as this speech recognition module.

[0587] Emotion analysis means

[0588] The server uses the acquired audio data to activate the emotion engine. The emotion engine analyzes the user's emotions (e.g., tension, anxiety, confusion, etc.) from the audio data in real time. This emotion analysis can use emotion analysis technologies such as the Microsoft Azure Emotion API. The analyzed emotion data is used as reference information for fraud analysis.

[0589] Analysis means

[0590] The server provides the text-converted data to a generative AI model. Specifically, the text data is input into a generative AI model (e.g., GPT-3) for analysis. This analysis utilizes a pre-trained natural language processing model. Furthermore, sentiment data obtained from sentiment analysis is also taken into consideration to determine the likelihood of fraud with high accuracy.

[0591] Display means

[0592] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The terminal receives the analysis results sent from the server and displays them to the user. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message.

[0593] Examples

[0594] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." When the user starts recording, the smart glasses record the call audio and send the data to a server. The server uses IBM Watson Speech to Text to convert the audio to text and simultaneously analyzes the user's emotions from the audio data using the Microsoft Azure Emotion API. Next, a generative AI model (e.g., GPT-3) is used to analyze the text data, and taking the emotion data into consideration, it is determined that there is a high probability of it being a scam. This analysis result is immediately sent to the user's device, displaying a summary of the conversation along with the message, "Warning: This call may be a scam!"

[0595] For example, by inputting a prompt such as, "Is there a possibility that the content of this call is fraudulent?" into the AI ​​model, highly accurate fraud detection becomes possible.

[0596] This allows users to check in real time whether a call may be related to a scam and take necessary action. By combining this with an emotion engine, the user's emotional state is also used as a factor in the determination, allowing for a more accurate assessment of the likelihood of fraud.

[0597] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0598] Detailed explanation of the processing steps

[0599] Step 1:

[0600] Acquiring audio

[0601] 1. The user puts on a portable electronic device (e.g., smart glasses or a smartphone) before initiating a call.

[0602] 2. If a user suspects a scam during a call, enable the recording function on their smart device.

[0603] 3. The smart device records the call audio in real time and sends the audio data to the server.

[0604] Input: User's voice call.

[0605] Output: Real-time audio data sent to the server.

[0606] Step 2:

[0607] Receiving and temporarily storing audio data

[0608] 1. The server receives audio data transmitted from smart devices in real time.

[0609] 2. The server temporarily saves the received audio data in a file format (e.g., audio_data.wav).

[0610] Input: Voice data sent from a smart device.

[0611] Output: Audio file stored on the server.

[0612] Step 3:

[0613] Converting audio data to text

[0614] 1. The server invokes the speech recognition module to analyze the stored audio data.

[0615] 2. The server uses IBM Watson Speech to Text or Google Cloud Speech-to-Text to convert the audio data into text data.

[0616] Input: Saved audio file.

[0617] Output: Converted text data.

[0618] Step 4:

[0619] Emotion analysis

[0620] 1. The server starts the sentiment analysis engine using the audio data after it has been converted to text.

[0621] 2. The server analyzes the user's emotions in real time from the voice data using the Microsoft Azure Emotion API, etc.

[0622] Input: Converted text data and audio data.

[0623] Output: Analyzed emotion data (e.g., tension, anxiety, confusion).

[0624] Step 5:

[0625] Fraud analysis

[0626] 1. The server accesses a pre-trained natural language processing model (e.g., GPT-3) to provide text data to the generative AI model.

[0627] 2. The server inputs text data into the model and performs analysis to determine the likelihood of fraud.

[0628] 3. The server also integrates sentiment analysis data to more accurately determine the likelihood of fraud.

[0629] Input: Text data and sentiment data.

[0630] Output: Analysis results regarding the possibility of fraud.

[0631] Step 6:

[0632] Display of analysis results

[0633] 1. The server stores the determination of the possibility of fraud and a summary of the conversation in a database.

[0634] 2. The server sends the analysis results to the user's terminal.

[0635] 3. The device displays the received analysis results to the user. In particular, if there is a high probability of fraud, it displays a summary of the conversation along with a warning message.

[0636] Input: Analysis results from the server.

[0637] Output: Warning messages and analysis results displayed on the user's terminal.

[0638] (Application Example 2)

[0639] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0640] Currently, telephone scams and other types of fraud are rapidly increasing, resulting in a surge in victims among individuals and businesses. To address this, a system is needed that analyzes the content of phone calls to determine the likelihood of fraud. However, conventional systems struggle to accurately detect fraud based solely on simple voice data analysis, lacking consideration for the user's emotions and circumstances. Furthermore, they lack the functionality to notify users of analysis results in real time. To address these challenges, a more accurate and real-time fraud detection system is necessary.

[0641] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0642] In this invention, the server includes voice acquisition means, text conversion means, analysis means, sentiment analysis means, and display means. This makes it possible for the user to determine the possibility of fraud in real time during a call with high accuracy and to be notified of the result immediately.

[0643] "Voice acquisition means" refers to a device that has the function of recording audio during a phone call and acquiring that data.

[0644] A "text conversion means" is a device that has the function of analyzing acquired audio data and converting it into text data.

[0645] An "analysis device" is a device that has the function of analyzing and determining the possibility of fraud based on the converted text data.

[0646] An "emotion analysis device" is a device that analyzes a user's emotions from voice data and provides reference data for fraud detection.

[0647] A "display means" is a device that has the function of displaying analysis results and the possibility of fraud to the user.

[0648] A "generative AI model" is a type of AI technology that is trained on large amounts of data and performs analysis based on text data and sentiment data.

[0649] A "smart device" is a portable electronic device that has internet connectivity and various sensors, and is capable of interacting with the user.

[0650] "Natural language processing" is a general term for technologies that enable computers to understand and process human language.

[0651] "Real-time" refers to processing or updating immediately without delay or lag.

[0652] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. By combining this system with an emotion analysis engine that recognizes the user's emotions, the system can determine the possibility of fraud with higher accuracy. A specific embodiment of this system is shown below.

[0653] Voice acquisition method

[0654] The user uses a smart device (e.g., a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device. The smart device records the call audio in real time and sends the data to a server.

[0655] Text conversion means

[0656] The server receives audio data sent from a smart device. The received audio data is temporarily stored as an audio file. Next, a speech recognition module (for example, Google's speech recognition service) is used to convert the audio data into text. This speech recognition module analyzes the audio data and generates the corresponding text data.

[0657] Emotion analysis means

[0658] The server uses the acquired audio data to activate an emotion analysis system. This system utilizes the Hugging Face Transformers library to analyze the user's emotions (e.g., tension, anxiety, confusion) in real time from the audio data. The analyzed emotion data is used as reference information for fraud analysis.

[0659] Analysis means

[0660] The server provides the text-converted data to the analysis system. Specifically, it uses a generative AI model to input the text data and perform the analysis. In this analysis, a pre-trained generative AI model is used to determine whether the input text data constitutes fraudulent activity. Furthermore, sentiment data obtained from the sentiment analysis system is also taken into consideration to determine the likelihood of fraud with greater accuracy.

[0661] Display means

[0662] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The user's terminal receives and displays the analysis results sent from the server. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message.

[0663] Examples

[0664] For example, consider a scenario where a user receives a call on their smartphone saying, "Hello, this is a financial institution. Your account has been compromised. Please provide your account number and password for verification." The smartphone records the call and sends the audio data to a server. The server converts the audio to text and simultaneously analyzes the user's emotions (such as anxiety or tension) from the audio data. A generative AI model analyzes the text data and, taking the emotional data into consideration, determines that there is a high probability of fraud. This result is immediately sent to the user's device, displaying a summary of the conversation along with a warning message: "Warning: This call may be a scam!"

[0665] Examples of prompts to input into a generative AI model

[0666] Hello, this is a financial institution. Your account has been compromised. Please provide your account number and password for verification. Emotion: Anxiety Score: 0.85

[0667] By inputting the above prompt text into a generating AI model, it is possible to determine the likelihood of fraud with high accuracy.

[0668] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0669] Step 1:

[0670] A user initiates a call using a smart device. If the user suspects fraud during the call, they enable the recording function on their smart device. At this time, the smart device records the call audio in real time. The input is the call audio, and the output is audio data.

[0671] Step 2:

[0672] The smart device sends the recorded audio data to the server. The server receives the audio data and temporarily stores it as an audio file. The input is the audio data, and the output is the stored audio file.

[0673] Step 3:

[0674] The server uses a speech recognition module (for example, Google's speech recognition service) to convert the stored audio data into text. The speech recognition module analyzes the audio data and generates the corresponding text data. The input is an audio file, and the output is text data.

[0675] Step 4:

[0676] The server provides the generated text data to an emotion analysis tool (e.g., the Hugging Face Transformers library). The emotion analysis tool analyzes the text data and analyzes the user's emotions (e.g., tension, anxiety, confusion, etc.) in real time. The input is text data, and the output is emotion data.

[0677] Step 5:

[0678] The server inputs text data and sentiment data into a generative AI model to analyze the likelihood of fraud. The generative AI model determines whether an activity constitutes fraud based on the text data and sentiment data. A pre-trained generative AI model is used for this purpose. The input is text data and sentiment data, and the output is the determination of the likelihood of fraud.

[0679] Step 6:

[0680] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The user's terminal receives the analysis results and displays them to the user. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message. The input is the analysis results, and the output is the message displayed on the user's terminal.

[0681] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0682] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0683] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0684] [Third Embodiment]

[0685] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0686] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0687] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0688] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0689] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0690] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0691] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0692] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0693] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0694] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0695] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0696] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[0697] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. This system consists of a smart device worn by the user, a server that processes the audio data, and a terminal that displays the results. A specific embodiment of this system is described below.

[0698] Voice acquisition method

[0699] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device. The smart device records the call audio in real time and prepares to send the data to a server.

[0700] Text conversion means

[0701] The server receives audio data sent from a smart device. The server uses a speech recognition module to convert the received audio data into text. This speech recognition module typically uses natural language processing technology. Specifically, it analyzes the audio data and converts its content into corresponding text data.

[0702] Analysis means

[0703] The server inputs the transcribed conversation into an analysis device. The analysis device uses a generative AI module to analyze the content of the text data and determine the possibility of fraud. This generative AI module uses a pre-trained model to determine whether the input text data constitutes fraudulent activity. If the analysis results indicate a high probability of fraudulent activity, it summarizes the content.

[0704] Display means

[0705] The server saves the analysis results to a database and sends the results to the user's terminal. The terminal receives the analysis results sent from the server and displays them to the user. In particular, if there is a high probability of fraud, the analysis results are displayed along with a warning message.

[0706] Examples

[0707] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and analyzes it using generative AI, determining that it is highly likely to be a scam. This result is immediately sent to the user's device, displaying the analysis results along with a warning message: "Warning: This call may be a scam!"

[0708] This allows users to identify potential scams in real time and prevent becoming a victim. This system is particularly useful for the elderly and users with limited knowledge of fraud, and is expected to have a positive effect on reducing social fraud.

[0709] The following describes the processing flow.

[0710] Step 1:

[0711] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. Then, if they suspect the call may be a scam, they enable the recording function on their smart device.

[0712] Step 2:

[0713] The smart device records the audio data during a call in real time. The recorded audio data is then prepared to be sent to a server over the network.

[0714] Step 3:

[0715] The server receives audio data sent from the smart device. The received audio data is temporarily stored as an audio file.

[0716] Step 4:

[0717] The server uses a speech recognition module to convert audio data into text. The speech recognition module analyzes the audio data and generates the corresponding text data.

[0718] Step 5:

[0719] The server provides the text-converted data to the analysis system. Specifically, it inputs the text data into a generative AI module and performs the analysis.

[0720] Step 6:

[0721] The server uses a generative AI module to analyze the content of the text data. This analysis determines whether it is potentially fraudulent. If it is highly likely to be fraudulent, it summarizes the content of the conversation.

[0722] Step 7:

[0723] The server saves the analysis results (determination of the likelihood of fraud and a summary of the conversation) to a database. The saved data includes the user ID, the likelihood of fraud, and the summary text.

[0724] Step 8:

[0725] The server sends the analysis results stored in the database to the user's terminal. Network communication is used for this transmission.

[0726] Step 9:

[0727] The terminal receives the analysis results sent from the server. The received data is then analyzed for presentation to the user.

[0728] Step 10:

[0729] The device displays the analysis results to the user. In particular, if there is a high probability of fraud, a summary of the conversation will be displayed along with the message, "Warning: This call may be a scam!"

[0730] This allows users to check in real time whether a call may be related to a scam and take necessary action.

[0731] (Example 1)

[0732] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0733] In modern society, the methods used in special fraud schemes are becoming more sophisticated and diverse, and victims, particularly the elderly and others with limited knowledge of fraud, are increasingly being targeted. It is difficult to immediately detect and notify victims of fraudulent activity through ordinary phone or video calls. Therefore, there is a need for means to prevent such fraud before it occurs.

[0734] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0735] In this invention, the server includes voice acquisition means for acquiring audio during a call, voice recognition means for converting the acquired audio into text, artificial intelligence analysis means for analyzing the converted text and determining the possibility of fraud, notification means for storing the possibility of fraud and the analysis results in a database and sending them to the user's terminal, and display means for displaying the possibility of fraud and the analysis results. This makes it possible to immediately analyze the possibility of fraud during a call and notify the user.

[0736] "Audio acquisition means for acquiring audio during a call" refers to a device or method for collecting audio in real time during a call and passing that data to subsequent processing.

[0737] A "speech recognition method that converts acquired speech into text" refers to a technology or module that receives speech data and converts it into a string of characters. Generally, it utilizes natural language processing technology.

[0738] "An artificial intelligence analysis method for analyzing converted text and determining the possibility of fraud" refers to artificial intelligence technology used to analyze text data and determine the possibility of fraudulent activity based on its content. It is executed using a pre-trained model.

[0739] "A notification method for storing potential fraud and analysis results in a database and sending them to the user's terminal" refers to a method or infrastructure for securely storing analysis results and notifying the user's terminal of necessary information in a timely manner.

[0740] "Display means for displaying the possibility of fraud and the analysis results" refers to a device or application for providing the user with the results of the analysis visually.

[0741] A "wearable device" is a device that a user can wear and that has functions such as making calls and recording audio. Examples include smart glasses and smartwatches.

[0742] System Overview

[0743] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. This system consists of a wearable device worn by the user, a server that processes the audio data, and a terminal that displays the analysis results.

[0744] Hardware and software to be used

[0745] Wearable devices: Smart glasses and smartphones

[0746] Server: A computer used for speech recognition and artificial intelligence analysis.

[0747] Speech recognition software: Common examples include Google Cloud Speech-to-Text and IBM Watson Speech to Text.

[0748] Artificial intelligence analysis software: Generative AI models (e.g., GPT-3 and BERT)

[0749] Device: Smartphone or tablet to display the analysis results.

[0750] Processing flow

[0751] 1. Voice acquisition:

[0752] The user puts on a wearable device (such as smart glasses or a smartphone) before starting a call. If the user suspects a scam during the call, they enable the recording function of the wearable device. The wearable device records the call audio in real time and sends the data to a server.

[0753] 2. Speech recognition:

[0754] The server receives audio data transmitted from the wearable device. The server first stores the audio data, and then converts it to text using Google Cloud Speech-to-Text or IBM Watson Speech to Text.

[0755] 3. Text analysis:

[0756] After the audio is converted to text, the server inputs that text data into a generating AI model (e.g., GPT-3 or BERT). The server then uses the pre-trained model to determine the likelihood of fraud.

[0757] Example of a prompt:

[0758] "The following is a transcript of a call the user received. Please analyze it and determine if it is a scam. Then, summarize your reasoning."

[0759] Call content: "Hello, this is the bank. Your account has been used fraudulently. Please tell us your account number and PIN so we can verify it."

[0760] 4. Saving and notifying of analysis results:

[0761] The server saves the analysis results to a database. It then sends the analysis results to the user's device. If the results are highly likely to be fraudulent, a warning message is sent along with them.

[0762] 5. Display of analysis results:

[0763] The device receives the analysis results sent from the server and notifies the user. In particular, if there is a high possibility of fraud, the analysis results will be displayed along with the message, "Warning: This call may be a scam!"

[0764] Specific example

[0765] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please tell us your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server uses Google Cloud Speech-to-Text to convert the speech to text and analyzes it with a generative AI model (e.g., GPT-3). If the server determines that the call is likely a scam, it immediately sends this result to the user's smartphone, displaying a message such as "Warning: This call may be a scam!" along with the analysis results.

[0766] This system allows users to identify potential scams in real time and prevent becoming a victim. This system is particularly useful for the elderly and users with limited knowledge of fraud, and is expected to reduce the overall amount of fraud damage in society.

[0767] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0768] Step 1: Voice Acquisition

[0769] The user puts on a wearable device (such as smart glasses or a smartphone) before starting a call. If the user suspects a scam during the call, they enable the recording function of the wearable device. The wearable device records the call audio in real time and sends the audio data to a server.

[0770] Input: User's action of starting recording and call audio

[0771] Output: Audio data sent to the server

[0772] Specific actions:

[0773] The user presses the record button on the smart glasses.

[0774] The smart glasses pick up the call content with their microphone and send the compressed audio data to the server.

[0775] Step 2: Speech Recognition

[0776] The server receives audio data transmitted from the wearable device. The server first stores the audio data, and then converts it to text using Google Cloud Speech-to-Text or IBM Watson Speech to Text.

[0777] Input: Audio data

[0778] Output: Text data

[0779] Specific actions:

[0780] The server saves the audio data as a temporary file.

[0781] The server calls the Google Cloud Speech-to-Text API to convert the audio data into text data.

[0782] Example text: "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification."

[0783] Step 3: Text Analysis

[0784] After the audio data is converted to text, the server inputs that text data into a generative AI model (e.g., GPT-3 or BERT). The server uses a pre-trained model to generate prompts to determine the likelihood of fraud and sends them to the generative AI model.

[0785] Input: Text data

[0786] Output: Analysis results regarding the possibility of fraud

[0787] Specific actions:

[0788] The server creates prompt messages for the generated AI model.

[0789] Prompt example: "The following is a transcript of a call the user received. Analyze it and determine if it is a scam. Then, summarize your reasoning."

[0790] Call content: "This is the bank. Your account has been used fraudulently. Please tell us your account number and PIN so we can verify it."

[0791] The server inputs text data into a generating AI model and analyzes its potential for fraud.

[0792] Example analysis result: "Probability of fraud: High"

[0793] Step 4: Saving and notifying of analysis results

[0794] The server saves the analysis results to a database. It then sends the analysis results to the user's device. If the results are highly likely to be fraudulent, a warning message is sent along with them.

[0795] Input: Analysis results

[0796] Output: Notification message to the user terminal

[0797] Specific actions:

[0798] The server saves the analysis results to the database as "Probability of fraud: High".

[0799] The server sends a message to the user's smartphone stating that it is "highly likely to be a scam."

[0800] Step 5: Displaying the analysis results

[0801] The device receives the analysis results sent from the server and notifies the user. In particular, if there is a high possibility of fraud, the analysis results will be displayed along with the message, "Warning: This call may be a scam!"

[0802] Input: Notification message from the server

[0803] Output: Warning messages and analysis results displayed on the user screen

[0804] Specific actions:

[0805] When the device receives a message sent from the server, a pop-up notification will appear stating, "Warning: This call may be a scam!"

[0806] Detailed analysis results will be displayed on the device screen.

[0807] (Application Example 1)

[0808] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0809] Conventional call content monitoring systems lacked the means to detect potentially fraudulent calls in real time and quickly warn users. As a result, elderly users and those with limited knowledge of fraud were particularly vulnerable to becoming victims of special fraud schemes. With current technology, there was no effective way to immediately inform users of potential fraudulent activity occurring during a call and prevent it.

[0810] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0811] In this invention, the server includes: an audio acquisition means for acquiring audio during a call; a text conversion means for converting the acquired audio into text; an analysis means for analyzing the converted text and determining the possibility of fraud; a display means for displaying the possibility of fraud and the analysis results; and a warning means for generating and outputting a warning message in real time regarding the possibility of fraud. This allows the user to immediately recognize situations with a high probability of fraud during a call and prevent becoming a victim.

[0812] A "voice acquisition means" is a device that has the function of recording voice during a call in real time and acquiring that voice data.

[0813] A "text conversion means" is a module that has the function of analyzing acquired audio data and converting its content into corresponding text data.

[0814] The "analysis method" is a system that uses the converted text data and natural language processing technologies such as generative AI models to determine the possibility of fraud.

[0815] "Display means" refers to an interface for visually conveying the results of the server's analysis to the user, and is a device or application for displaying analysis results and warning messages.

[0816] A "warning mechanism" is a system that generates a warning message in real time when there is a high probability of fraud and notifies the user of it via voice or visual means.

[0817] A "smart device" is a device that a user can wear to capture audio during a call, and it refers to a wide range of devices, including smartphones and smart glasses.

[0818] To implement this invention, a smart device worn by the user, a server that processes voice data, and a terminal that displays analysis results and warning messages are required.

[0819] System Configuration

[0820] 1. Means of acquiring sound:

[0821] The user puts on a smart device (e.g., a smartphone or smart glasses) before starting a call.

[0822] If a user suspects a call may be fraudulent, they can enable the recording function on their smart device. The smart device will record the call audio in real time and send the data to a server.

[0823] 2. Text conversion means:

[0824] The server receives voice data sent from the smart device.

[0825] The server uses a speech recognition module (e.g., the speech_recognition library) to convert speech data into text. The speech data is analyzed using natural language processing techniques, and its content is converted into text data.

[0826] 3. Analysis method:

[0827] The server uses a generative AI model (e.g., a deep learning model) to analyze the transcribed conversation content.

[0828] The analysis tool determines whether the input text data constitutes fraudulent activity. If there is a high probability of fraud, it summarizes the content and obtains the analysis result.

[0829] 4. Display means:

[0830] The server saves the analysis results to a database and sends the analysis results to the user's terminal.

[0831] The device receives the analysis results sent from the server and displays them to the user. If there is a high probability of fraud, the analysis results will be displayed along with a warning message.

[0832] 5. Warning measures:

[0833] If the server determines that there is a high probability of fraud, it generates a warning message in real time and notifies the user audibly or visually (e.g., converting text to speech using the gTTS library and playing it back with the playsound library).

[0834] Specific example

[0835] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please tell us your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and analyzes it using a generative AI model. If it is determined that the call is highly likely to be fraudulent, the result is immediately sent to the user's device, displaying the analysis results along with a warning message: "Warning: This call may be fraudulent!" This allows the user to understand the possibility of a scam in real time and prevent becoming a victim.

[0836] Example of a prompt

[0837] Please convert the recorded audio data to text and follow the instructions below:

[0838] 1. Send the converted text to the API endpoint.

[0839] 2. Based on the analysis results, a warning message will be generated if there is a high probability of fraud.

[0840] 3. Play the warning message as an audio.

[0841] example:

[0842] Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification.

[0843] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0844] Step 1:

[0845] The user puts on a smart device and starts a call. If the user suspects fraud during the call, they enable the recording function on their smart device. At this point, the voice acquisition process begins. The input data is the audio from the call, and the output data is the recorded audio. Specifically, recording starts when the user presses the record button on their smartphone or smart glasses.

[0846] Step 2:

[0847] The smart device transmits recorded audio data to the server in real time. The server receives this audio data. The input data is the recorded audio data, and the output data is the audio file stored on the server. Specifically, the smart device uploads the audio data to the server via the network connection.

[0848] Step 3:

[0849] The server converts the received audio data into text data using a speech recognition module. The `speech_recognition` library is used for this conversion. The input data is audio data, and the output data is the converted text data. Specifically, the speech recognition module analyzes the audio data and converts it into the corresponding text.

[0850] Step 4:

[0851] The server inputs the converted text data into the analysis device. The analysis device uses a generative AI model to analyze the text data and determine the likelihood of fraud. Natural language processing techniques are used in this process. The input data is text data, and the output data is the analysis result indicating the likelihood of fraud. Specifically, the generative AI model analyzes the text data and determines the likelihood of fraud.

[0852] Step 5:

[0853] The server saves the analysis results to a database and sends the results to the user's terminal. The input data is the analysis results, and the output data is the analysis results displayed on the user's terminal. Specifically, the server saves the analysis results to a database and then sends that data to the user's terminal via the network.

[0854] Step 6:

[0855] The device displays the received analysis results. In particular, if there is a high probability of fraud, the analysis results are displayed along with a warning message. The input data is the analysis results sent from the server, and the output data is the warning message and analysis results visually displayed to the user. Specifically, the device displays the received data on the screen, visually conveying the information to the user.

[0856] Step 7:

[0857] If the server determines that a case is highly likely to be fraudulent, it generates a warning message in real time and notifies the user of it audibly or visually. For example, the gTTS library is used to convert the warning message into audio, and the playsound library is used to play it. The input data is the analysis results, and the output data is the generated warning message. Specifically, the server generates a warning message based on the analysis results and notifies the user of it audibly or visually.

[0858] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0859] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. By further incorporating an emotion engine that recognizes the user's emotions, this system can determine the possibility of fraud with even greater accuracy. A specific embodiment of this system is shown below.

[0860] Voice acquisition method

[0861] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device. The smart device records the call audio in real time and sends the data to a server.

[0862] Text conversion means

[0863] The server receives audio data sent from the smart device. The received audio data is temporarily stored as an audio file. Next, a speech recognition module is used to convert the audio data into text. This speech recognition module analyzes the audio data and generates the corresponding text data.

[0864] Emotional Engine

[0865] The server uses the acquired voice data to activate the emotion engine. The emotion engine analyzes the user's emotions (e.g., tension, anxiety, confusion, etc.) from the voice data in real time. The analyzed emotion data is used as reference information for fraud analysis.

[0866] Analysis means

[0867] The server provides the text-converted data to the analysis system. Specifically, it inputs the text data into a generative AI module for analysis. This analysis uses a pre-trained model to determine whether the input text data constitutes fraudulent activity. Furthermore, sentiment data obtained from the sentiment engine is also taken into consideration to determine the likelihood of fraud with even greater accuracy.

[0868] Display means

[0869] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The terminal receives the analysis results sent from the server and displays them to the user. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message.

[0870] Examples

[0871] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and simultaneously analyzes the user's emotions from the audio data. Generative AI analyzes the text data, taking the emotions into consideration, to determine that there is a high probability of fraud. This result is immediately sent to the user's device, displaying a summary of the conversation along with a warning message: "Warning: This call may be a scam!"

[0872] This allows users to check in real time whether a call may be related to a scam and take necessary action. By combining this with an emotion engine, the user's emotional state is also used as a factor in the determination, allowing for a more accurate assessment of the likelihood of fraud.

[0873] The following describes the processing flow.

[0874] Step 1:

[0875] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device.

[0876] Step 2:

[0877] The smart device records the audio data during a call in real time. The recorded audio data is then prepared to be sent to a server over the network.

[0878] Step 3:

[0879] The server receives audio data sent from the smart device. The received audio data is temporarily stored as an audio file.

[0880] Step 4:

[0881] The server uses a speech recognition module to convert audio data into text. The speech recognition module analyzes the audio data and generates the corresponding text data.

[0882] Step 5:

[0883] The server uses the acquired audio data and the text data generated from that audio data to activate the emotion engine. The emotion engine analyzes the user's emotions in real time from the audio data and generates emotion data.

[0884] Step 6:

[0885] The server provides the text-converted data and sentiment data to the analysis system. Specifically, it inputs the text data into a generative AI module and performs analysis while taking sentiment data into consideration.

[0886] Step 7:

[0887] The server uses a generative AI module to analyze the content of the text data. This analysis determines whether it is potentially fraudulent. If it is highly likely to be fraudulent, it summarizes the content of the conversation.

[0888] Step 8:

[0889] The server stores the analysis results (determination of fraud possibility and conversation summary) and sentiment data in a database. The stored data includes user ID, fraud possibility, summary text, and sentiment data.

[0890] Step 9:

[0891] The server sends the analysis results stored in the database to the user's terminal. Network communication is used for this transmission.

[0892] Step 10:

[0893] The terminal receives the analysis results sent from the server. The received data is then analyzed for presentation to the user.

[0894] Step 11:

[0895] The device displays the analysis results to the user. In particular, if there is a high probability of fraud, a message such as "WARNING: This call may be a scam!" will be displayed along with a summary of the conversation and sentiment data.

[0896] Examples

[0897] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and simultaneously analyzes the user's emotions (e.g., tension, anxiety, confusion) from the audio data. Generative AI analyzes the text data, taking into account the emotion data, to determine that there is a high probability of fraud. This result is immediately sent to the user's device, displaying a warning message, "Warning: This call may be fraudulent!", along with a summary of the conversation and the emotion data. This allows the user to check in real time whether the call is potentially fraudulent and take the necessary action.

[0898] (Example 2)

[0899] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0900] Conventional in-call fraud detection systems rely solely on text analysis of call content to determine the likelihood of fraud, resulting in low accuracy and an inability to consider emotional factors such as user anxiety and tension. Furthermore, analysis results are not always presented in real time, preventing users from responding immediately. A system is needed that addresses these challenges and provides more accurate, real-time fraud detection.

[0901] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0902] In this invention, the server includes an audio acquisition means for acquiring audio during a call, a text conversion means for converting the acquired audio into text, an emotion analysis means for analyzing the user's emotions from the audio data, an analysis means for determining the possibility of fraud based on the converted text and the analyzed emotion data, and a display means for displaying the possibility of fraud and the analysis results. This enables highly accurate fraud detection in real time by simultaneously analyzing the audio during a call and the user's emotions.

[0903] "Voice acquisition means" refers to functions or devices for collecting and recording voice during a phone call in real time.

[0904] "Text conversion means" refers to functions and technologies for converting acquired audio data into text data.

[0905] "Emotional analysis means" refers to functions and technologies for analyzing and detecting a user's emotional state from voice data.

[0906] "Analysis means" refers to analytical functions and techniques that use converted text data and sentiment data to achieve a specific purpose (in this invention, the determination of the possibility of fraud).

[0907] "Display means" refers to functions or devices that visually present analysis results or warning messages to the user.

[0908] A "portable electronic device" is an electronic device that is easy to carry and has features such as voice acquisition during calls.

[0909] Natural language processing is a field of computer science and artificial intelligence that automatically analyzes, understands, and generates human language.

[0910] "Real-time analysis" is a process that processes and analyzes data immediately upon acquisition, providing results quickly.

[0911] This invention provides a system that acquires audio during a phone call in real time and analyzes the possibility of fraud, thereby assisting users in taking immediate action. This system also incorporates an emotion engine that recognizes the user's emotions, enabling more accurate determination of fraud potential. Specific embodiments of this system are described in detail below.

[0912] Voice acquisition method

[0913] The user puts on a portable electronic device (e.g., a smart device) before starting a call. If the user suspects during the call that it may be a scam, they enable the recording function on their smart device. The smart device (e.g., smart glasses, smartphone, etc.) records the call audio in real time and sends the data to a server.

[0914] Text conversion means

[0915] The server receives audio data sent from a smart device. The received audio data is temporarily stored as an audio file. Next, a speech recognition module is used to convert the audio data into text. Common speech recognition technologies such as IBM Watson Speech to Text and Google Cloud Speech-to-Text can be used as this speech recognition module.

[0916] Emotion analysis means

[0917] The server uses the acquired audio data to activate the emotion engine. The emotion engine analyzes the user's emotions (e.g., tension, anxiety, confusion, etc.) from the audio data in real time. This emotion analysis can use emotion analysis technologies such as the Microsoft Azure Emotion API. The analyzed emotion data is used as reference information for fraud analysis.

[0918] Analysis means

[0919] The server provides the text-converted data to a generative AI model. Specifically, the text data is input into a generative AI model (e.g., GPT-3) for analysis. This analysis utilizes a pre-trained natural language processing model. Furthermore, sentiment data obtained from sentiment analysis is also taken into consideration to determine the likelihood of fraud with high accuracy.

[0920] Display means

[0921] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The terminal receives the analysis results sent from the server and displays them to the user. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message.

[0922] Examples

[0923] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." When the user starts recording, the smart glasses record the call audio and send the data to a server. The server uses IBM Watson Speech to Text to convert the audio to text and simultaneously analyzes the user's emotions from the audio data using the Microsoft Azure Emotion API. Next, a generative AI model (e.g., GPT-3) is used to analyze the text data, and taking the emotion data into consideration, it is determined that there is a high probability of it being a scam. This analysis result is immediately sent to the user's device, displaying a summary of the conversation along with the message, "Warning: This call may be a scam!"

[0924] For example, by inputting a prompt such as, "Is there a possibility that the content of this call is fraudulent?" into the AI ​​model, highly accurate fraud detection becomes possible.

[0925] This allows users to check in real time whether a call may be related to a scam and take necessary action. By combining this with an emotion engine, the user's emotional state is also used as a factor in the determination, allowing for a more accurate assessment of the likelihood of fraud.

[0926] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0927] Detailed explanation of the processing steps

[0928] Step 1:

[0929] Acquiring audio

[0930] 1. The user puts on a portable electronic device (e.g., smart glasses or a smartphone) before initiating a call.

[0931] 2. If a user suspects a scam during a call, enable the recording function on their smart device.

[0932] 3. The smart device records the call audio in real time and sends the audio data to the server.

[0933] Input: User's voice call.

[0934] Output: Real-time audio data sent to the server.

[0935] Step 2:

[0936] Receiving and temporarily storing audio data

[0937] 1. The server receives audio data transmitted from smart devices in real time.

[0938] 2. The server temporarily saves the received audio data in a file format (e.g., audio_data.wav).

[0939] Input: Voice data sent from a smart device.

[0940] Output: Audio file stored on the server.

[0941] Step 3:

[0942] Converting audio data to text

[0943] 1. The server invokes the speech recognition module to analyze the stored audio data.

[0944] 2. The server uses IBM Watson Speech to Text or Google Cloud Speech-to-Text to convert the audio data into text data.

[0945] Input: Saved audio file.

[0946] Output: Converted text data.

[0947] Step 4:

[0948] Emotion analysis

[0949] 1. The server starts the sentiment analysis engine using the audio data after it has been converted to text.

[0950] 2. The server analyzes the user's emotions in real time from the voice data using the Microsoft Azure Emotion API, etc.

[0951] Input: Converted text data and audio data.

[0952] Output: Analyzed emotion data (e.g., tension, anxiety, confusion).

[0953] Step 5:

[0954] Fraud analysis

[0955] 1. The server accesses a pre-trained natural language processing model (e.g., GPT-3) to provide text data to the generative AI model.

[0956] 2. The server inputs text data into the model and performs analysis to determine the likelihood of fraud.

[0957] 3. The server also integrates sentiment analysis data to more accurately determine the likelihood of fraud.

[0958] Input: Text data and sentiment data.

[0959] Output: Analysis results regarding the possibility of fraud.

[0960] Step 6:

[0961] Display of analysis results

[0962] 1. The server stores the determination of the possibility of fraud and a summary of the conversation in a database.

[0963] 2. The server sends the analysis results to the user's terminal.

[0964] 3. The device displays the received analysis results to the user. In particular, if there is a high probability of fraud, it displays a summary of the conversation along with a warning message.

[0965] Input: Analysis results from the server.

[0966] Output: Warning messages and analysis results displayed on the user's terminal.

[0967] (Application Example 2)

[0968] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the headset-type terminal 314 will be referred to as the "terminal."

[0969] Currently, telephone scams and other types of fraud are rapidly increasing, resulting in a surge in victims among individuals and businesses. To address this, a system is needed that analyzes the content of phone calls to determine the likelihood of fraud. However, conventional systems struggle to accurately detect fraud based solely on simple voice data analysis, lacking consideration for the user's emotions and circumstances. Furthermore, they lack the functionality to notify users of analysis results in real time. To address these challenges, a more accurate and real-time fraud detection system is necessary.

[0970] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0971] In this invention, the server includes voice acquisition means, text conversion means, analysis means, sentiment analysis means, and display means. This makes it possible for the user to determine the possibility of fraud in real time during a call with high accuracy and to be notified of the result immediately.

[0972] "Voice acquisition means" refers to a device that has the function of recording audio during a phone call and acquiring that data.

[0973] A "text conversion means" is a device that has the function of analyzing acquired audio data and converting it into text data.

[0974] An "analysis device" is a device that has the function of analyzing and determining the possibility of fraud based on the converted text data.

[0975] An "emotion analysis device" is a device that analyzes a user's emotions from voice data and provides reference data for fraud detection.

[0976] A "display means" is a device that has the function of displaying analysis results and the possibility of fraud to the user.

[0977] A "generative AI model" is a type of AI technology that is trained on large amounts of data and performs analysis based on text data and sentiment data.

[0978] A "smart device" is a portable electronic device that has internet connectivity and various sensors, and is capable of interacting with the user.

[0979] "Natural language processing" is a general term for technologies that enable computers to understand and process human language.

[0980] "Real-time" refers to processing or updating immediately without delay or lag.

[0981] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. By combining this system with an emotion analysis engine that recognizes the user's emotions, the system can determine the possibility of fraud with higher accuracy. A specific embodiment of this system is shown below.

[0982] Voice acquisition method

[0983] The user uses a smart device (e.g., a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device. The smart device records the call audio in real time and sends the data to a server.

[0984] Text conversion means

[0985] The server receives audio data sent from a smart device. The received audio data is temporarily stored as an audio file. Next, a speech recognition module (for example, Google's speech recognition service) is used to convert the audio data into text. This speech recognition module analyzes the audio data and generates the corresponding text data.

[0986] Emotion analysis means

[0987] The server uses the acquired audio data to activate an emotion analysis system. This system utilizes the Hugging Face Transformers library to analyze the user's emotions (e.g., tension, anxiety, confusion) in real time from the audio data. The analyzed emotion data is used as reference information for fraud analysis.

[0988] Analysis means

[0989] The server provides the text-converted data to the analysis system. Specifically, it uses a generative AI model to input the text data and perform the analysis. In this analysis, a pre-trained generative AI model is used to determine whether the input text data constitutes fraudulent activity. Furthermore, sentiment data obtained from the sentiment analysis system is also taken into consideration to determine the likelihood of fraud with greater accuracy.

[0990] Display means

[0991] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The user's terminal receives and displays the analysis results sent from the server. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message.

[0992] Examples

[0993] For example, consider a scenario where a user receives a call on their smartphone saying, "Hello, this is a financial institution. Your account has been compromised. Please provide your account number and password for verification." The smartphone records the call and sends the audio data to a server. The server converts the audio to text and simultaneously analyzes the user's emotions (such as anxiety or tension) from the audio data. A generative AI model analyzes the text data and, taking the emotional data into consideration, determines that there is a high probability of fraud. This result is immediately sent to the user's device, displaying a summary of the conversation along with a warning message: "Warning: This call may be a scam!"

[0994] Examples of prompts to input into a generative AI model

[0995] Hello, this is a financial institution. Your account has been compromised. Please provide your account number and password for verification. Emotion: Anxiety Score: 0.85

[0996] By inputting the above prompt text into a generating AI model, it is possible to determine the likelihood of fraud with high accuracy.

[0997] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0998] Step 1:

[0999] A user initiates a call using a smart device. If the user suspects fraud during the call, they enable the recording function on their smart device. At this time, the smart device records the call audio in real time. The input is the call audio, and the output is audio data.

[1000] Step 2:

[1001] The smart device sends the recorded audio data to the server. The server receives the audio data and temporarily stores it as an audio file. The input is the audio data, and the output is the stored audio file.

[1002] Step 3:

[1003] The server uses a speech recognition module (for example, Google's speech recognition service) to convert the stored audio data into text. The speech recognition module analyzes the audio data and generates the corresponding text data. The input is an audio file, and the output is text data.

[1004] Step 4:

[1005] The server provides the generated text data to an emotion analysis tool (e.g., the Hugging Face Transformers library). The emotion analysis tool analyzes the text data and analyzes the user's emotions (e.g., tension, anxiety, confusion, etc.) in real time. The input is text data, and the output is emotion data.

[1006] Step 5:

[1007] The server inputs text data and sentiment data into a generative AI model to analyze the likelihood of fraud. The generative AI model determines whether an activity constitutes fraud based on the text data and sentiment data. A pre-trained generative AI model is used for this purpose. The input is text data and sentiment data, and the output is the determination of the likelihood of fraud.

[1008] Step 6:

[1009] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The user's terminal receives the analysis results and displays them to the user. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message. The input is the analysis results, and the output is the message displayed on the user's terminal.

[1010] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1011] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1012] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[1013] [Fourth Embodiment]

[1014] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[1015] As shown in Figure 7, the data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1016] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1017] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[1018] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1019] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1020] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1021] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[1022] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1023] The specific processing program 56 is an example of a "program" relating to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1024] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1025] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1026] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1027] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. This system consists of a smart device worn by the user, a server that processes the audio data, and a terminal that displays the results. A specific embodiment of this system is described below.

[1028] Voice acquisition method

[1029] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device. The smart device records the call audio in real time and prepares to send the data to a server.

[1030] Text conversion means

[1031] The server receives audio data sent from a smart device. The server uses a speech recognition module to convert the received audio data into text. This speech recognition module typically uses natural language processing technology. Specifically, it analyzes the audio data and converts its content into corresponding text data.

[1032] Analysis means

[1033] The server inputs the transcribed conversation into an analysis device. The analysis device uses a generative AI module to analyze the content of the text data and determine the possibility of fraud. This generative AI module uses a pre-trained model to determine whether the input text data constitutes fraudulent activity. If the analysis results indicate a high probability of fraudulent activity, it summarizes the content.

[1034] Display means

[1035] The server saves the analysis results to a database and sends the results to the user's terminal. The terminal receives the analysis results sent from the server and displays them to the user. In particular, if there is a high probability of fraud, the analysis results are displayed along with a warning message.

[1036] Examples

[1037] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and analyzes it using generative AI, determining that it is highly likely to be a scam. This result is immediately sent to the user's device, displaying the analysis results along with a warning message: "Warning: This call may be a scam!"

[1038] This allows users to identify potential scams in real time and prevent becoming a victim. This system is particularly useful for the elderly and users with limited knowledge of fraud, and is expected to have a positive effect on reducing social fraud.

[1039] The following describes the processing flow.

[1040] Step 1:

[1041] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. Then, if they suspect the call may be a scam, they enable the recording function on their smart device.

[1042] Step 2:

[1043] The smart device records the audio data during a call in real time. The recorded audio data is then prepared to be sent to a server over the network.

[1044] Step 3:

[1045] The server receives audio data sent from the smart device. The received audio data is temporarily stored as an audio file.

[1046] Step 4:

[1047] The server uses a speech recognition module to convert audio data into text. The speech recognition module analyzes the audio data and generates the corresponding text data.

[1048] Step 5:

[1049] The server provides the text-converted data to the analysis system. Specifically, it inputs the text data into a generative AI module and performs the analysis.

[1050] Step 6:

[1051] The server uses a generative AI module to analyze the content of the text data. This analysis determines whether it is potentially fraudulent. If it is highly likely to be fraudulent, it summarizes the content of the conversation.

[1052] Step 7:

[1053] The server saves the analysis results (determination of the likelihood of fraud and a summary of the conversation) to a database. The saved data includes the user ID, the likelihood of fraud, and the summary text.

[1054] Step 8:

[1055] The server sends the analysis results stored in the database to the user's terminal. Network communication is used for this transmission.

[1056] Step 9:

[1057] The terminal receives the analysis results sent from the server. The received data is then analyzed for presentation to the user.

[1058] Step 10:

[1059] The device displays the analysis results to the user. In particular, if there is a high probability of fraud, a summary of the conversation will be displayed along with the message, "Warning: This call may be a scam!"

[1060] This allows users to check in real time whether a call may be related to a scam and take necessary action.

[1061] (Example 1)

[1062] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1063] In modern society, the methods used in special fraud schemes are becoming more sophisticated and diverse, and victims, particularly the elderly and others with limited knowledge of fraud, are increasingly being targeted. It is difficult to immediately detect and notify victims of fraudulent activity through ordinary phone or video calls. Therefore, there is a need for means to prevent such fraud before it occurs.

[1064] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1065] In this invention, the server includes voice acquisition means for acquiring audio during a call, voice recognition means for converting the acquired audio into text, artificial intelligence analysis means for analyzing the converted text and determining the possibility of fraud, notification means for storing the possibility of fraud and the analysis results in a database and sending them to the user's terminal, and display means for displaying the possibility of fraud and the analysis results. This makes it possible to immediately analyze the possibility of fraud during a call and notify the user.

[1066] "Audio acquisition means for acquiring audio during a call" refers to a device or method for collecting audio in real time during a call and passing that data to subsequent processing.

[1067] A "speech recognition method that converts acquired speech into text" refers to a technology or module that receives speech data and converts it into a string of characters. Generally, it utilizes natural language processing technology.

[1068] "An artificial intelligence analysis method for analyzing converted text and determining the possibility of fraud" refers to artificial intelligence technology used to analyze text data and determine the possibility of fraudulent activity based on its content. It is executed using a pre-trained model.

[1069] "A notification method for storing potential fraud and analysis results in a database and sending them to the user's terminal" refers to a method or infrastructure for securely storing analysis results and notifying the user's terminal of necessary information in a timely manner.

[1070] "Display means for displaying the possibility of fraud and the analysis results" refers to a device or application for providing the user with the results of the analysis visually.

[1071] A "wearable device" is a device that a user can wear and that has functions such as making calls and recording audio. Examples include smart glasses and smartwatches.

[1072] System Overview

[1073] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. This system consists of a wearable device worn by the user, a server that processes the audio data, and a terminal that displays the analysis results.

[1074] Hardware and software to be used

[1075] Wearable devices: Smart glasses and smartphones

[1076] Server: A computer used for speech recognition and artificial intelligence analysis.

[1077] Speech recognition software: Common examples include Google Cloud Speech-to-Text and IBM Watson Speech to Text.

[1078] Artificial intelligence analysis software: Generative AI models (e.g., GPT-3 and BERT)

[1079] Device: Smartphone or tablet to display the analysis results.

[1080] Processing flow

[1081] 1. Voice acquisition:

[1082] The user puts on a wearable device (such as smart glasses or a smartphone) before starting a call. If the user suspects a scam during the call, they enable the recording function of the wearable device. The wearable device records the call audio in real time and sends the data to a server.

[1083] 2. Speech recognition:

[1084] The server receives audio data transmitted from the wearable device. The server first stores the audio data, and then converts it to text using Google Cloud Speech-to-Text or IBM Watson Speech to Text.

[1085] 3. Text analysis:

[1086] After the audio is converted to text, the server inputs that text data into a generating AI model (e.g., GPT-3 or BERT). The server then uses the pre-trained model to determine the likelihood of fraud.

[1087] Example of a prompt:

[1088] "The following is a transcript of a call the user received. Please analyze it and determine if it is a scam. Then, summarize your reasoning."

[1089] Call content: "Hello, this is the bank. Your account has been used fraudulently. Please tell us your account number and PIN so we can verify it."

[1090] 4. Saving and notifying of analysis results:

[1091] The server saves the analysis results to a database. It then sends the analysis results to the user's device. If the results are highly likely to be fraudulent, a warning message is sent along with them.

[1092] 5. Display of analysis results:

[1093] The device receives the analysis results sent from the server and notifies the user. In particular, if there is a high possibility of fraud, the analysis results will be displayed along with the message, "Warning: This call may be a scam!"

[1094] Specific example

[1095] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please tell us your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server uses Google Cloud Speech-to-Text to convert the speech to text and analyzes it with a generative AI model (e.g., GPT-3). If the server determines that the call is likely a scam, it immediately sends this result to the user's smartphone, displaying a message such as "Warning: This call may be a scam!" along with the analysis results.

[1096] This system allows users to identify potential scams in real time and prevent becoming a victim. This system is particularly useful for the elderly and users with limited knowledge of fraud, and is expected to reduce the overall amount of fraud damage in society.

[1097] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1098] Step 1: Voice Acquisition

[1099] The user puts on a wearable device (such as smart glasses or a smartphone) before starting a call. If the user suspects a scam during the call, they enable the recording function of the wearable device. The wearable device records the call audio in real time and sends the audio data to a server.

[1100] Input: User's action of starting recording and call audio

[1101] Output: Audio data sent to the server

[1102] Specific actions:

[1103] The user presses the record button on the smart glasses.

[1104] The smart glasses pick up the call content with their microphone and send the compressed audio data to the server.

[1105] Step 2: Speech Recognition

[1106] The server receives audio data transmitted from the wearable device. The server first stores the audio data, and then converts it to text using Google Cloud Speech-to-Text or IBM Watson Speech to Text.

[1107] Input: Audio data

[1108] Output: Text data

[1109] Specific actions:

[1110] The server saves the audio data as a temporary file.

[1111] The server calls the Google Cloud Speech-to-Text API to convert the audio data into text data.

[1112] Example text: "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification."

[1113] Step 3: Text Analysis

[1114] After the audio data is converted to text, the server inputs that text data into a generative AI model (e.g., GPT-3 or BERT). The server uses a pre-trained model to generate prompts to determine the likelihood of fraud and sends them to the generative AI model.

[1115] Input: Text data

[1116] Output: Analysis results regarding the possibility of fraud

[1117] Specific actions:

[1118] The server creates prompt messages for the generated AI model.

[1119] Prompt example: "The following is a transcript of a call the user received. Analyze it and determine if it is a scam. Then, summarize your reasoning."

[1120] Call content: "This is the bank. Your account has been used fraudulently. Please tell us your account number and PIN so we can verify it."

[1121] The server inputs text data into a generating AI model and analyzes its potential for fraud.

[1122] Example analysis result: "Probability of fraud: High"

[1123] Step 4: Saving and notifying of analysis results

[1124] The server saves the analysis results to a database. It then sends the analysis results to the user's device. If the results are highly likely to be fraudulent, a warning message is sent along with them.

[1125] Input: Analysis results

[1126] Output: Notification message to the user terminal

[1127] Specific actions:

[1128] The server saves the analysis results to the database as "Probability of fraud: High".

[1129] The server sends a message to the user's smartphone stating that it is "highly likely to be a scam."

[1130] Step 5: Displaying the analysis results

[1131] The device receives the analysis results sent from the server and notifies the user. In particular, if there is a high possibility of fraud, the analysis results will be displayed along with the message, "Warning: This call may be a scam!"

[1132] Input: Notification message from the server

[1133] Output: Warning messages and analysis results displayed on the user screen

[1134] Specific actions:

[1135] When the device receives a message sent from the server, a pop-up notification will appear stating, "Warning: This call may be a scam!"

[1136] Detailed analysis results will be displayed on the device screen.

[1137] (Application Example 1)

[1138] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1139] Conventional call content monitoring systems lacked the means to detect potentially fraudulent calls in real time and quickly warn users. As a result, elderly users and those with limited knowledge of fraud were particularly vulnerable to becoming victims of special fraud schemes. With current technology, there was no effective way to immediately inform users of potential fraudulent activity occurring during a call and prevent it.

[1140] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1141] In this invention, the server includes: an audio acquisition means for acquiring audio during a call; a text conversion means for converting the acquired audio into text; an analysis means for analyzing the converted text and determining the possibility of fraud; a display means for displaying the possibility of fraud and the analysis results; and a warning means for generating and outputting a warning message in real time regarding the possibility of fraud. This allows the user to immediately recognize situations with a high probability of fraud during a call and prevent becoming a victim.

[1142] A "voice acquisition means" is a device that has the function of recording voice during a call in real time and acquiring that voice data.

[1143] A "text conversion means" is a module that has the function of analyzing acquired audio data and converting its content into corresponding text data.

[1144] The "analysis method" is a system that uses the converted text data and natural language processing technologies such as generative AI models to determine the possibility of fraud.

[1145] "Display means" refers to an interface for visually conveying the results of the server's analysis to the user, and is a device or application for displaying analysis results and warning messages.

[1146] A "warning mechanism" is a system that generates a warning message in real time when there is a high probability of fraud and notifies the user of it via voice or visual means.

[1147] A "smart device" is a device that a user can wear to capture audio during a call, and it refers to a wide range of devices, including smartphones and smart glasses.

[1148] To implement this invention, a smart device worn by the user, a server that processes voice data, and a terminal that displays analysis results and warning messages are required.

[1149] System Configuration

[1150] 1. Means of acquiring sound:

[1151] The user puts on a smart device (e.g., a smartphone or smart glasses) before starting a call.

[1152] If a user suspects a call may be fraudulent, they can enable the recording function on their smart device. The smart device will record the call audio in real time and send the data to a server.

[1153] 2. Text conversion means:

[1154] The server receives voice data sent from the smart device.

[1155] The server uses a speech recognition module (e.g., the speech_recognition library) to convert speech data into text. The speech data is analyzed using natural language processing techniques, and its content is converted into text data.

[1156] 3. Analysis method:

[1157] The server uses a generative AI model (e.g., a deep learning model) to analyze the transcribed conversation content.

[1158] The analysis tool determines whether the input text data constitutes fraudulent activity. If there is a high probability of fraud, it summarizes the content and obtains the analysis result.

[1159] 4. Display means:

[1160] The server saves the analysis results to a database and sends the analysis results to the user's terminal.

[1161] The device receives the analysis results sent from the server and displays them to the user. If there is a high probability of fraud, the analysis results will be displayed along with a warning message.

[1162] 5. Warning measures:

[1163] If the server determines that there is a high probability of fraud, it generates a warning message in real time and notifies the user audibly or visually (e.g., converting text to speech using the gTTS library and playing it back with the playsound library).

[1164] Specific example

[1165] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please tell us your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and analyzes it using a generative AI model. If it is determined that the call is highly likely to be fraudulent, the result is immediately sent to the user's device, displaying the analysis results along with a warning message: "Warning: This call may be fraudulent!" This allows the user to understand the possibility of a scam in real time and prevent becoming a victim.

[1166] Example of a prompt

[1167] Please convert the recorded audio data to text and follow the instructions below:

[1168] 1. Send the converted text to the API endpoint.

[1169] 2. Based on the analysis results, a warning message will be generated if there is a high probability of fraud.

[1170] 3. Play the warning message as an audio.

[1171] example:

[1172] Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification.

[1173] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1174] Step 1:

[1175] The user puts on a smart device and starts a call. If the user suspects fraud during the call, they enable the recording function on their smart device. At this point, the voice acquisition process begins. The input data is the audio from the call, and the output data is the recorded audio. Specifically, recording starts when the user presses the record button on their smartphone or smart glasses.

[1176] Step 2:

[1177] The smart device transmits recorded audio data to the server in real time. The server receives this audio data. The input data is the recorded audio data, and the output data is the audio file stored on the server. Specifically, the smart device uploads the audio data to the server via the network connection.

[1178] Step 3:

[1179] The server converts the received audio data into text data using a speech recognition module. The `speech_recognition` library is used for this conversion. The input data is audio data, and the output data is the converted text data. Specifically, the speech recognition module analyzes the audio data and converts it into the corresponding text.

[1180] Step 4:

[1181] The server inputs the converted text data into the analysis device. The analysis device uses a generative AI model to analyze the text data and determine the likelihood of fraud. Natural language processing techniques are used in this process. The input data is text data, and the output data is the analysis result indicating the likelihood of fraud. Specifically, the generative AI model analyzes the text data and determines the likelihood of fraud.

[1182] Step 5:

[1183] The server saves the analysis results to a database and sends the results to the user's terminal. The input data is the analysis results, and the output data is the analysis results displayed on the user's terminal. Specifically, the server saves the analysis results to a database and then sends that data to the user's terminal via the network.

[1184] Step 6:

[1185] The device displays the received analysis results. In particular, if there is a high probability of fraud, the analysis results are displayed along with a warning message. The input data is the analysis results sent from the server, and the output data is the warning message and analysis results visually displayed to the user. Specifically, the device displays the received data on the screen, visually conveying the information to the user.

[1186] Step 7:

[1187] If the server determines that a case is highly likely to be fraudulent, it generates a warning message in real time and notifies the user of it audibly or visually. For example, the gTTS library is used to convert the warning message into audio, and the playsound library is used to play it. The input data is the analysis results, and the output data is the generated warning message. Specifically, the server generates a warning message based on the analysis results and notifies the user of it audibly or visually.

[1188] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1189] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. By further incorporating an emotion engine that recognizes the user's emotions, this system can determine the possibility of fraud with even greater accuracy. A specific embodiment of this system is shown below.

[1190] Voice acquisition method

[1191] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device. The smart device records the call audio in real time and sends the data to a server.

[1192] Text conversion means

[1193] The server receives audio data sent from the smart device. The received audio data is temporarily stored as an audio file. Next, a speech recognition module is used to convert the audio data into text. This speech recognition module analyzes the audio data and generates the corresponding text data.

[1194] Emotional Engine

[1195] The server uses the acquired voice data to activate the emotion engine. The emotion engine analyzes the user's emotions (e.g., tension, anxiety, confusion, etc.) from the voice data in real time. The analyzed emotion data is used as reference information for fraud analysis.

[1196] Analysis means

[1197] The server provides the text-converted data to the analysis system. Specifically, it inputs the text data into a generative AI module for analysis. This analysis uses a pre-trained model to determine whether the input text data constitutes fraudulent activity. Furthermore, sentiment data obtained from the sentiment engine is also taken into consideration to determine the likelihood of fraud with even greater accuracy.

[1198] Display means

[1199] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The terminal receives the analysis results sent from the server and displays them to the user. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message.

[1200] Examples

[1201] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and simultaneously analyzes the user's emotions from the audio data. Generative AI analyzes the text data, taking the emotions into consideration, to determine that there is a high probability of fraud. This result is immediately sent to the user's device, displaying a summary of the conversation along with a warning message: "Warning: This call may be a scam!"

[1202] This allows users to check in real time whether a call may be related to a scam and take necessary action. By combining this with an emotion engine, the user's emotional state is also used as a factor in the determination, allowing for a more accurate assessment of the likelihood of fraud.

[1203] The following describes the processing flow.

[1204] Step 1:

[1205] The user puts on a smart device (e.g., smart glasses or a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device.

[1206] Step 2:

[1207] The smart device records the audio data during a call in real time. The recorded audio data is then prepared to be sent to a server over the network.

[1208] Step 3:

[1209] The server receives audio data sent from the smart device. The received audio data is temporarily stored as an audio file.

[1210] Step 4:

[1211] The server uses a speech recognition module to convert audio data into text. The speech recognition module analyzes the audio data and generates the corresponding text data.

[1212] Step 5:

[1213] The server uses the acquired audio data and the text data generated from that audio data to activate the emotion engine. The emotion engine analyzes the user's emotions in real time from the audio data and generates emotion data.

[1214] Step 6:

[1215] The server provides the text-converted data and sentiment data to the analysis system. Specifically, it inputs the text data into a generative AI module and performs analysis while taking sentiment data into consideration.

[1216] Step 7:

[1217] The server uses a generative AI module to analyze the content of the text data. This analysis determines whether it is potentially fraudulent. If it is highly likely to be fraudulent, it summarizes the content of the conversation.

[1218] Step 8:

[1219] The server stores the analysis results (determination of fraud possibility and conversation summary) and sentiment data in a database. The stored data includes user ID, fraud possibility, summary text, and sentiment data.

[1220] Step 9:

[1221] The server sends the analysis results stored in the database to the user's terminal. Network communication is used for this transmission.

[1222] Step 10:

[1223] The terminal receives the analysis results sent from the server. The received data is then analyzed for presentation to the user.

[1224] Step 11:

[1225] The device displays the analysis results to the user. In particular, if there is a high probability of fraud, a message such as "WARNING: This call may be a scam!" will be displayed along with a summary of the conversation and sentiment data.

[1226] Examples

[1227] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." The smart glasses record the call and send the audio data to a server. The server converts the audio to text and simultaneously analyzes the user's emotions (e.g., tension, anxiety, confusion) from the audio data. Generative AI analyzes the text data, taking into account the emotion data, to determine that there is a high probability of fraud. This result is immediately sent to the user's device, displaying a warning message, "Warning: This call may be fraudulent!", along with a summary of the conversation and the emotion data. This allows the user to check in real time whether the call is potentially fraudulent and take the necessary action.

[1228] (Example 2)

[1229] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1230] Conventional in-call fraud detection systems rely solely on text analysis of call content to determine the likelihood of fraud, resulting in low accuracy and an inability to consider emotional factors such as user anxiety and tension. Furthermore, analysis results are not always presented in real time, preventing users from responding immediately. A system is needed that addresses these challenges and provides more accurate, real-time fraud detection.

[1231] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1232] In this invention, the server includes an audio acquisition means for acquiring audio during a call, a text conversion means for converting the acquired audio into text, an emotion analysis means for analyzing the user's emotions from the audio data, an analysis means for determining the possibility of fraud based on the converted text and the analyzed emotion data, and a display means for displaying the possibility of fraud and the analysis results. This enables highly accurate fraud detection in real time by simultaneously analyzing the audio during a call and the user's emotions.

[1233] "Voice acquisition means" refers to functions or devices for collecting and recording voice during a phone call in real time.

[1234] "Text conversion means" refers to functions and technologies for converting acquired audio data into text data.

[1235] "Emotional analysis means" refers to functions and technologies for analyzing and detecting a user's emotional state from voice data.

[1236] "Analysis means" refers to analytical functions and techniques that use converted text data and sentiment data to achieve a specific purpose (in this invention, the determination of the possibility of fraud).

[1237] "Display means" refers to functions or devices that visually present analysis results or warning messages to the user.

[1238] A "portable electronic device" is an electronic device that is easy to carry and has features such as voice acquisition during calls.

[1239] Natural language processing is a field of computer science and artificial intelligence that automatically analyzes, understands, and generates human language.

[1240] "Real-time analysis" is a process that processes and analyzes data immediately upon acquisition, providing results quickly.

[1241] This invention provides a system that acquires audio during a phone call in real time and analyzes the possibility of fraud, thereby assisting users in taking immediate action. This system also incorporates an emotion engine that recognizes the user's emotions, enabling more accurate determination of fraud potential. Specific embodiments of this system are described in detail below.

[1242] Voice acquisition method

[1243] The user puts on a portable electronic device (e.g., a smart device) before starting a call. If the user suspects during the call that it may be a scam, they enable the recording function on their smart device. The smart device (e.g., smart glasses, smartphone, etc.) records the call audio in real time and sends the data to a server.

[1244] Text conversion means

[1245] The server receives audio data sent from a smart device. The received audio data is temporarily stored as an audio file. Next, a speech recognition module is used to convert the audio data into text. Common speech recognition technologies such as IBM Watson Speech to Text and Google Cloud Speech-to-Text can be used as this speech recognition module.

[1246] Emotion analysis means

[1247] The server uses the acquired audio data to activate the emotion engine. The emotion engine analyzes the user's emotions (e.g., tension, anxiety, confusion, etc.) from the audio data in real time. This emotion analysis can use emotion analysis technologies such as the Microsoft Azure Emotion API. The analyzed emotion data is used as reference information for fraud analysis.

[1248] Analysis means

[1249] The server provides the text-converted data to a generative AI model. Specifically, the text data is input into a generative AI model (e.g., GPT-3) for analysis. This analysis utilizes a pre-trained natural language processing model. Furthermore, sentiment data obtained from sentiment analysis is also taken into consideration to determine the likelihood of fraud with high accuracy.

[1250] Display means

[1251] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The terminal receives the analysis results sent from the server and displays them to the user. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message.

[1252] Examples

[1253] For example, consider a scenario where a user wearing smart glasses receives a call saying, "Hello, this is the bank. Your account has been used fraudulently. Please provide your account number and PIN for verification." When the user starts recording, the smart glasses record the call audio and send the data to a server. The server uses IBM Watson Speech to Text to convert the audio to text and simultaneously analyzes the user's emotions from the audio data using the Microsoft Azure Emotion API. Next, a generative AI model (e.g., GPT-3) is used to analyze the text data, and taking the emotion data into consideration, it is determined that there is a high probability of it being a scam. This analysis result is immediately sent to the user's device, displaying a summary of the conversation along with the message, "Warning: This call may be a scam!"

[1254] For example, by inputting a prompt such as, "Is there a possibility that the content of this call is fraudulent?" into the AI ​​model, highly accurate fraud detection becomes possible.

[1255] This allows users to check in real time whether a call may be related to a scam and take necessary action. By combining this with an emotion engine, the user's emotional state is also used as a factor in the determination, allowing for a more accurate assessment of the likelihood of fraud.

[1256] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1257] Detailed explanation of the processing steps

[1258] Step 1:

[1259] Acquiring audio

[1260] 1. The user puts on a portable electronic device (e.g., smart glasses or a smartphone) before initiating a call.

[1261] 2. If a user suspects a scam during a call, enable the recording function on their smart device.

[1262] 3. The smart device records the call audio in real time and sends the audio data to the server.

[1263] Input: User's voice call.

[1264] Output: Real-time audio data sent to the server.

[1265] Step 2:

[1266] Receiving and temporarily storing audio data

[1267] 1. The server receives audio data transmitted from smart devices in real time.

[1268] 2. The server temporarily saves the received audio data in a file format (e.g., audio_data.wav).

[1269] Input: Voice data sent from a smart device.

[1270] Output: Audio file stored on the server.

[1271] Step 3:

[1272] Converting audio data to text

[1273] 1. The server invokes the speech recognition module to analyze the stored audio data.

[1274] 2. The server uses IBM Watson Speech to Text or Google Cloud Speech-to-Text to convert the audio data into text data.

[1275] Input: Saved audio file.

[1276] Output: Converted text data.

[1277] Step 4:

[1278] Emotion analysis

[1279] 1. The server starts the sentiment analysis engine using the audio data after it has been converted to text.

[1280] 2. The server analyzes the user's emotions in real time from the voice data using the Microsoft Azure Emotion API, etc.

[1281] Input: Converted text data and audio data.

[1282] Output: Analyzed emotion data (e.g., tension, anxiety, confusion).

[1283] Step 5:

[1284] Fraud analysis

[1285] 1. The server accesses a pre-trained natural language processing model (e.g., GPT-3) to provide text data to the generative AI model.

[1286] 2. The server inputs text data into the model and performs analysis to determine the likelihood of fraud.

[1287] 3. The server also integrates sentiment analysis data to more accurately determine the likelihood of fraud.

[1288] Input: Text data and sentiment data.

[1289] Output: Analysis results regarding the possibility of fraud.

[1290] Step 6:

[1291] Display of analysis results

[1292] 1. The server stores the determination of the possibility of fraud and a summary of the conversation in a database.

[1293] 2. The server sends the analysis results to the user's terminal.

[1294] 3. The device displays the received analysis results to the user. In particular, if there is a high probability of fraud, it displays a summary of the conversation along with a warning message.

[1295] Input: Analysis results from the server.

[1296] Output: Warning messages and analysis results displayed on the user's terminal.

[1297] (Application Example 2)

[1298] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1299] Currently, telephone scams and other types of fraud are rapidly increasing, resulting in a surge in victims among individuals and businesses. To address this, a system is needed that analyzes the content of phone calls to determine the likelihood of fraud. However, conventional systems struggle to accurately detect fraud based solely on simple voice data analysis, lacking consideration for the user's emotions and circumstances. Furthermore, they lack the functionality to notify users of analysis results in real time. To address these challenges, a more accurate and real-time fraud detection system is necessary.

[1300] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1301] In this invention, the server includes voice acquisition means, text conversion means, analysis means, sentiment analysis means, and display means. This makes it possible for the user to determine the possibility of fraud in real time during a call with high accuracy and to be notified of the result immediately.

[1302] "Voice acquisition means" refers to a device that has the function of recording audio during a phone call and acquiring that data.

[1303] A "text conversion means" is a device that has the function of analyzing acquired audio data and converting it into text data.

[1304] An "analysis device" is a device that has the function of analyzing and determining the possibility of fraud based on the converted text data.

[1305] An "emotion analysis device" is a device that analyzes a user's emotions from voice data and provides reference data for fraud detection.

[1306] A "display means" is a device that has the function of displaying analysis results and the possibility of fraud to the user.

[1307] A "generative AI model" is a type of AI technology that is trained on large amounts of data and performs analysis based on text data and sentiment data.

[1308] A "smart device" is a portable electronic device that has internet connectivity and various sensors, and is capable of interacting with the user.

[1309] "Natural language processing" is a general term for technologies that enable computers to understand and process human language.

[1310] "Real-time" refers to processing or updating immediately without delay or lag.

[1311] This invention relates to a system that acquires audio during a phone call in real time and analyzes the possibility of fraud. By combining this system with an emotion analysis engine that recognizes the user's emotions, the system can determine the possibility of fraud with higher accuracy. A specific embodiment of this system is shown below.

[1312] Voice acquisition method

[1313] The user uses a smart device (e.g., a smartphone) before starting a call. If the user suspects the call may be a scam, they enable the recording function on their smart device. The smart device records the call audio in real time and sends the data to a server.

[1314] Text conversion means

[1315] The server receives audio data sent from a smart device. The received audio data is temporarily stored as an audio file. Next, a speech recognition module (for example, Google's speech recognition service) is used to convert the audio data into text. This speech recognition module analyzes the audio data and generates the corresponding text data.

[1316] Emotion analysis means

[1317] The server uses the acquired audio data to activate an emotion analysis system. This system utilizes the Hugging Face Transformers library to analyze the user's emotions (e.g., tension, anxiety, confusion) in real time from the audio data. The analyzed emotion data is used as reference information for fraud analysis.

[1318] Analysis means

[1319] The server provides the text-converted data to the analysis system. Specifically, it uses a generative AI model to input the text data and perform the analysis. In this analysis, a pre-trained generative AI model is used to determine whether the input text data constitutes fraudulent activity. Furthermore, sentiment data obtained from the sentiment analysis system is also taken into consideration to determine the likelihood of fraud with greater accuracy.

[1320] Display means

[1321] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The user's terminal receives and displays the analysis results sent from the server. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message.

[1322] Examples

[1323] For example, consider a scenario where a user receives a call on their smartphone saying, "Hello, this is a financial institution. Your account has been compromised. Please provide your account number and password for verification." The smartphone records the call and sends the audio data to a server. The server converts the audio to text and simultaneously analyzes the user's emotions (such as anxiety or tension) from the audio data. A generative AI model analyzes the text data and, taking the emotional data into consideration, determines that there is a high probability of fraud. This result is immediately sent to the user's device, displaying a summary of the conversation along with a warning message: "Warning: This call may be a scam!"

[1324] Examples of prompts to input into a generative AI model

[1325] Hello, this is a financial institution. Your account has been compromised. Please provide your account number and password for verification. Emotion: Anxiety Score: 0.85

[1326] By inputting the above prompt text into a generating AI model, it is possible to determine the likelihood of fraud with high accuracy.

[1327] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1328] Step 1:

[1329] A user initiates a call using a smart device. If the user suspects fraud during the call, they enable the recording function on their smart device. At this time, the smart device records the call audio in real time. The input is the call audio, and the output is audio data.

[1330] Step 2:

[1331] The smart device sends the recorded audio data to the server. The server receives the audio data and temporarily stores it as an audio file. The input is the audio data, and the output is the stored audio file.

[1332] Step 3:

[1333] The server uses a speech recognition module (for example, Google's speech recognition service) to convert the stored audio data into text. The speech recognition module analyzes the audio data and generates the corresponding text data. The input is an audio file, and the output is text data.

[1334] Step 4:

[1335] The server provides the generated text data to an emotion analysis tool (e.g., the Hugging Face Transformers library). The emotion analysis tool analyzes the text data and analyzes the user's emotions (e.g., tension, anxiety, confusion, etc.) in real time. The input is text data, and the output is emotion data.

[1336] Step 5:

[1337] The server inputs text data and sentiment data into a generative AI model to analyze the likelihood of fraud. The generative AI model determines whether an activity constitutes fraud based on the text data and sentiment data. A pre-trained generative AI model is used for this purpose. The input is text data and sentiment data, and the output is the determination of the likelihood of fraud.

[1338] Step 6:

[1339] The server stores the analysis results (assessment of the likelihood of fraud and a summary of the conversation) in a database and sends the results to the user's terminal. The user's terminal receives the analysis results and displays them to the user. In particular, if there is a high possibility of fraud, the analysis results are displayed along with a warning message. The input is the analysis results, and the output is the message displayed on the user's terminal.

[1340] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1341] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1342] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[1343] Furthermore, the emotion identification model 59, acting as an emotion engine, may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1344] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[1345] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[1346] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[1347] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[1348] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[1349] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[1350] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[1351] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[1352] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[1353] 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.

[1354] Furthermore, it is not necessary to store the entirety of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[1355] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[1356] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[1357] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[1358] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[1359] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[1360] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted as being incorporated by reference.

[1361] The following is further disclosed regarding the embodiments described above.

[1362] (Claim 1)

[1363] A means of acquiring audio during a call,

[1364] A text conversion means for converting acquired audio into text,

[1365] An analytical means for analyzing the converted text and determining the possibility of fraud,

[1366] A means of displaying the possibility of fraud and the analysis results,

[1367] A system that includes this.

[1368] (Claim 2)

[1369] The system according to claim 1, wherein a smart device is used as a means for acquiring voice.

[1370] (Claim 3)

[1371] The system according to claim 1, which uses natural language processing as an analytical means to determine the possibility of fraud.

[1372] "Example 1"

[1373] (Claim 1)

[1374] A means of acquiring audio during a call,

[1375] A speech recognition means that converts acquired audio into text,

[1376] An artificial intelligence analysis tool that analyzes the converted text and determines the possibility of fraud,

[1377] A notification mechanism that stores the potential fraud and analysis results in a database and sends them to the user's terminal,

[1378] A means of displaying the possibility of fraud and the analysis results,

[1379] A system that includes this.

[1380] (Claim 2)

[1381] The system according to claim 1, wherein a wearable device is used as a means of acquiring voice.

[1382] (Claim 3)

[1383] The system according to claim 1, which uses a generated AI model as an artificial intelligence analysis means to determine the possibility of fraud.

[1384] "Application Example 1"

[1385] (Claim 1)

[1386] A means of acquiring audio during a call,

[1387] A text conversion means for converting acquired audio into text,

[1388] An analytical means for analyzing the converted text and determining the possibility of fraud,

[1389] A means of displaying the possibility of fraud and the analysis results,

[1390] A system including a warning mechanism that generates and outputs warning messages in real time regarding potential fraud.

[1391] (Claim 2)

[1392] The system according to claim 1, wherein a smart device is used as a means for acquiring voice.

[1393] (Claim 3)

[1394] The system according to claim 1, which uses natural language processing as an analytical means to determine the possibility of fraud.

[1395] "Example 2 of combining an emotion engine"

[1396] (Claim 1)

[1397] A means of acquiring audio during a call,

[1398] A text conversion means for converting acquired audio into text,

[1399] A means of analyzing user emotions from voice data,

[1400] An analytical means for determining the possibility of fraud based on converted text and analyzed sentiment data,

[1401] A means of displaying the possibility of fraud and the analysis results,

[1402] A system that includes this.

[1403] (Claim 2)

[1404] The system according to claim 1, wherein a portable electronic device is used as a means for acquiring voice.

[1405] (Claim 3)

[1406] The system according to claim 1, which uses natural language processing as an analytical means to determine the possibility of fraud.

[1407] "Application example 2 when combining with an emotional engine"

[1408] Claims (with added features of application examples)

[1409] (Claim 1)

[1410] A means of acquiring sound,

[1411] A text conversion means for converting acquired audio into text,

[1412] An analytical means for analyzing the converted text and determining the possibility of fraud,

[1413] A sentiment analysis method that analyzes sentiment data and provides reference data for fraud detection,

[1414] A means of displaying the possibility of fraud and the analysis results,

[1415] A system that includes this.

[1416] (Claim 2)

[1417] The system according to claim 1, which uses a smart device as a means of acquiring voice and is installed on a smartphone.

[1418] (Claim 3)

[1419] The system according to claim 1, which uses natural language processing and a generative AI model as analytical means to determine the possibility of fraud, including emotional data. [Explanation of Symbols]

[1420] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots< / url:> < / url:> < / url:> < / url:>

Claims

1. A means of acquiring audio during a call, A text conversion means for converting acquired audio into text, An analytical means for analyzing the converted text and determining the possibility of fraud, A means of displaying the possibility of fraud and the analysis results, A system that includes this.

2. The system according to claim 1, wherein a smart device is used as a means for acquiring voice.

3. The system according to claim 1, which uses natural language processing as an analytical means to determine the possibility of fraud.

Images