system

Abstract

We provide the system. [Solution] Means for acquiring voice information during a call, A means of identifying emotions by analyzing acquired audio information, A means of assessing the likelihood of fraud based on identified emotions, A means of altering the content of a conversation using voice reproduction when the likelihood of fraud exceeds a certain standard, A means of notifying users of suspected fraud and displaying information to alleviate their anxiety, A means of sending dynamic warning messages to fraudulent communicators, A system that includes this.

Description

Technical Field

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

Background Art

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

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In recent years, fraud using telephones has become a social problem, and many people have suffered from it. These frauds are carried out through voice, and by sophisticated means, they give the recipient anxiety and confusion, and as a result, may cause the provision of false information or the extortion of money. Conventional methods cannot adequately address this problem, and there is a need for a technology that can detect and prevent fraud in real time. Therefore, it is required to provide a new system that analyzes voice data during a call and responds quickly and effectively when there is suspicion of fraud.

Means for Solving the Problems

[0005] This invention provides a system that analyzes voice data from acquired calls to identify emotions and evaluates fraud risk based on the identified emotions. If the fraud risk exceeds a certain threshold, the system dynamically modifies the conversation content using voice emulation to prevent fraudulent activity. Furthermore, it provides psychological reassurance to the user by displaying a message that notifies them of the suspected fraud and alleviates their anxiety. This system also includes a function to convert voice signals into text data and compare it with known fraud patterns, which can improve the accuracy of fraud detection.

[0006] "Voice data" refers to data that represents the voice signals collected during a phone call in digital format.

[0007] "Emotion" refers to a psychological state analyzed from speech and the linguistic content of the speaker, and includes feelings such as confusion and anxiety.

[0008] "Fraud risk" is an evaluation metric used to determine whether there is a possibility of fraud in the content of a phone call.

[0009] A "threshold" is a standard value set to determine a level of risk when quantifying the risk of fraud.

[0010] "Voice emulation" is a technology that uses AI to generate synthesized speech and alters the content of a conversation by using a voice different from the original speaker.

[0011] "User" refers to an individual or legal entity that uses this system.

[0012] "Text data" refers to string information converted from audio signals using speech recognition technology.

[0013] A "fraud pattern" is a set of linguistic features and techniques used to identify fraudulent activities, based on past fraud cases. [Brief explanation of the drawing]

[0014] [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 a plurality of emotions are mapped. [Figure 10] It shows an emotion map to which a plurality of 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 an 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 an emotion engine is combined.

Modes for Carrying Out the Invention

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

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

[0017] In the following embodiments, a tagged processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), and the like.

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

[0019] In the following embodiments, a tagged storage is one or more non-volatile storage devices that store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes, etc.

[0020] In the following embodiments, the signed communication interface (I / F) is an interface that includes a communication processor and an antenna, etc. The communication interface manages communication between multiple computers. Examples of communication standards applicable to the communication interface include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0021] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0022] [First Embodiment]

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

[0024] As shown in Figure 1, the data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0025] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0026] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0027] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0028] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0029] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

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

[0031] As shown in Figure 2, in the data processing device 12, a specific processing is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" related to the technology of this disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 according to the specific processing program 56 executed on the RAM 30.

[0032] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0033] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0034] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0035] This invention provides an AI agent system for effectively combating fraudulent phone calls. This system is server-centric and operates in conjunction with the user's terminal. The program's processing and its application are described in detail below.

[0036] The server acquires audio data of calls initiated from the user's terminal in real time. This audio data is converted into a clean signal using noise reduction technology and then converted into text data using speech recognition technology. This text data is analyzed using natural language processing (NLP) to identify the speaker's emotions. The emotion identification algorithm particularly identifies emotions such as confusion and anxiety, and this information is used to assess the risk of fraud.

[0037] When the fraud risk exceeds a certain threshold, the server dynamically adjusts the conversation using voice emulation. This voice emulation generates synthesized speech and sends messages to deter fraudsters. Specifically, it generates messages such as, "This call is being monitored by security." During this process, the user's device also displays reassuring messages notifying them that the caller may be fraudulent and to alleviate their anxiety.

[0038] For example, if a user receives a phone call that may be fraudulent, the server automatically executes the process described above and intervenes with voice emulation to thwart the fraudster's intentions. As a result, it is possible to prevent fraud and reduce the user's psychological burden.

[0039] In this way, the system proposed in the patent provides an effective defense against telephone fraud by integrating speech recognition, natural language processing, and speech synthesis technologies. This enhances user trust and contributes to improved social safety.

[0040] The following describes the processing flow.

[0041] Step 1:

[0042] The server acquires audio data for calls initiated on the user's terminal. The audio signal is sent to the server in real time and prepared for data processing.

[0043] Step 2:

[0044] The server performs noise reduction on the acquired audio data to clarify the audio signal. This generates clean audio data that improves the accuracy of speech recognition.

[0045] Step 3:

[0046] The server converts clean audio data into text data using speech recognition technology. This process extracts the content of the audio as a string of characters, preparing it for subsequent processing.

[0047] Step 4:

[0048] The server analyzes the generated text data using natural language processing (NLP) techniques to identify the speaker's emotions. In particular, it scrutinizes feelings of confusion and anxiety, and measures their intensity if present.

[0049] Step 5:

[0050] The server assesses fraud risk based on the analyzed sentiment data. It calculates a risk score, which also involves matching the data against existing fraud patterns to detect fraudulent phrases and unusual patterns.

[0051] Step 6:

[0052] If the server determines that the fraud risk has exceeded a set threshold, it dynamically alters the call content using AI-powered voice emulation. At this time, it sends a counter-message to the fraudster in synthesized speech.

[0053] Step 7:

[0054] The server displays a reassuring message on the user's device to alleviate their concerns and notify them of the potential for fraud. This allows the user to take appropriate action quickly.

[0055] Step 8:

[0056] The server ultimately logs the call and stores data for analyzing fraud risk. This allows for later review of the details and, if necessary, assist in legal proceedings.

[0057] (Example 1)

[0058] 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."

[0059] In modern society, fraudulent activities conducted via communication networks are on the rise, with phone-based fraud being a particular social problem. These scams are becoming more sophisticated, making it difficult for ordinary users to detect them on their own. Furthermore, fraudulent activities induce psychological confusion and anxiety in users, making it difficult for them to take prompt and appropriate action. Therefore, there is a need for a system that can detect potential fraud on behalf of users and respond quickly.

[0060] 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.

[0061] In this invention, the server includes means for acquiring voice information during a call, means for removing noise from the acquired voice information and converting it into text information using speech recognition technology, means for analyzing the text information to identify emotions, means for modifying the conversation content using speech synthesis when the risk of fraud exceeds a certain threshold, and means for displaying information to notify the user of the suspicion of fraud and alleviate their anxiety. This makes it possible to detect the risk of fraud in real time during a call and ensure the safety of the user.

[0062] A "call" is the act of transmitting voice information bidirectionally over a communication network, and is a means of communication through voice.

[0063] "Auditory information" refers to data that is recorded or transmitted as sound, including human speech and other sound signals.

[0064] "Noise reduction" is a process that makes audio clearer by removing unwanted background noise and interference sounds from audio data.

[0065] "Speech recognition technology" is a technology that converts speech information into a digital format, analyzes the content of the speech, and converts it into text information.

[0066] "Textual information" refers to string data extracted through speech recognition, and includes information expressed in human language.

[0067] "Emotion identification" is the process of analyzing textual information and audio data to identify the speaker's emotional state (for example, confusion, anxiety, joy, etc.).

[0068] "Fraud risk" is an indicator that numerically or qualitatively assesses the likelihood that the person on the other end of a call is likely to commit fraud.

[0069] A "threshold" is a boundary value that is set as a reference point, and when that value is exceeded, a specific action is performed.

[0070] "Speech synthesis" is a technology that converts text data into speech and generates new speech data.

[0071] "Changing the content of a conversation" refers to the act of manipulating or editing voice messages sent during a call to convey intended information.

[0072] A "user" is a person or group that makes a call using this system and is protected from fraud.

[0073] "Information that alleviates anxiety" refers to messages and notifications presented in a way that provides users with a sense of psychological reassurance.

[0074] This invention is a system for detecting call fraud via a communication network and protecting users. This system operates in cooperation with a server, a user's terminal, and the network.

[0075] The server acquires audio information from calls transmitted from the user's terminal in real time. The audio information is streamed to the server via a digital processing unit. This system uses speech recognition technology to convert the audio information into text. Specifically, it uses a processing library to remove noise and the Google® Cloud Speech-to-Text API to transcribe the text.

[0076] Text information undergoes sentiment analysis on the server using natural language processing (NLP) techniques. The system utilizes Python's natural language processing libraries, NLTK and spaCy, to identify the speaker's emotions. This system specifically assesses fraud risk by detecting emotions such as confusion and anxiety. If the fraud risk assessment exceeds a threshold, speech synthesis technology is used to generate a deterrent warning message for fraudsters. Specifically, speech synthesis software is used to broadcast the message, "This call is being monitored by security."

[0077] The server also notifies the user's device of the suspected fraud and displays a reassuring message on the screen. For example, it might say, "This may be a suspicious call. Please rest assured, we are handling it."

[0078] One concrete example of this system's use is that when a user receives a suspicious phone call, they can input a prompt message into the AI ​​model such as, "I have been notified that there has been an unauthorized payment on my credit card. Please analyze the sentiment and assess the risk of fraud."

[0079] Without requiring any special action from the user, the system automatically detects potential fraud and takes appropriate action, preventing fraud and allowing users to continue their calls with peace of mind. This contributes to improving user safety and social trust.

[0080] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0081] Step 1:

[0082] The server acquires the audio information of the call transmitted from the user's terminal. This audio information is streamed via the VoIP protocol and input to the server in digital format. As output, the server prepares the audio data to pass to the next processing step while maintaining real-time performance.

[0083] Step 2:

[0084] The server processes the acquired audio information to remove noise. Specifically, it uses an audio processing library to apply filters that reduce background noise, processing the audio to make it clearer. The input is audio data containing noise, and the output is clean audio data with the noise minimized.

[0085] Step 3:

[0086] The server converts clean audio data into text information using speech recognition technology. Specifically, it uses speech recognition software to convert audio data into text format. The input is processed audio data, and the output is the corresponding text data. This conversion enables subsequent natural language processing.

[0087] Step 4:

[0088] The server applies natural language processing (NLP) techniques to the generated text information to analyze emotions. It uses a Python NLP library to analyze the text, focusing on specific emotions, particularly confusion and anxiety. The input is text data, and the output is the identification and intensity of the detected emotions.

[0089] Step 5:

[0090] The server assesses fraud risk using the results of sentiment analysis. A fraud risk assessment algorithm is used to quantify the risk level based on the identified emotions. The input is identified sentiment information, and the output is a numerical evaluation of fraud risk.

[0091] Step 6:

[0092] If the fraud risk assessment exceeds a certain threshold, the server uses speech synthesis technology to generate a warning message and insert it into the call. Specifically, it uses speech synthesis software to synthesize a message such as "This call is being monitored by security" to deter fraudsters. The input is the result of the fraud risk assessment, and the output is the generated warning message.

[0093] Step 7:

[0094] The server notifies the user's device of a suspected scam and displays a reassuring message on the screen. This notification allows the user to understand the potential for fraud and take prompt action. The input is the result of the fraud risk assessment, and the output is the reassuring message displayed on the device.

[0095] (Application Example 1)

[0096] 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."

[0097] In recent years, the methods used in telephone fraud have become more sophisticated, making it difficult for individual users to deal with them. Therefore, there is a need for effective means to prevent damage from telephone fraud. This invention aims to provide a new defense system that can quickly and accurately assess the possibility of telephone fraud and protect users.

[0098] 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.

[0099] In this invention, the server includes means for acquiring voice information in a call, means for analyzing the acquired voice information to identify emotions, means for evaluating the likelihood of fraud based on the identified emotions, means for modifying the conversation content using voice reconstruction if the likelihood of fraud exceeds a certain standard, means for displaying information to notify the user of the suspicion of fraud and alleviate anxiety, and means for sending a dynamic warning message to the fraudulent caller. This makes it possible to effectively deter special fraud calls and provide peace of mind to the user.

[0100] "Voice information in a call" refers to the entirety of voice data transmitted via communication means.

[0101] "Means of acquisition" refers to the technology for receiving and storing audio information in real time.

[0102] "Methods for analyzing and identifying emotions" refers to algorithms that process audio data to identify the speaker's psychological state.

[0103] "Methods for assessing the likelihood of fraud" refer to methods for quantifying the risk of fraudulent activity based on analyzed sentiment information.

[0104] "Methods for altering conversation content using voice reproduction" refers to technology that generates synthesized speech and automatically provides a new message to the other party on the call.

[0105] "A means of notifying users of suspected fraud and displaying information to alleviate their anxiety" refers to a method of providing reassurance by displaying a warning message on the user's device when fraud is suspected.

[0106] "Means of sending dynamic warning messages to fraudulent communicators" refers to technology that, when fraudulent activity is detected, sends a message in real time to warn those attempting to commit fraud.

[0107] To implement this invention, it is necessary to construct a system that acquires and analyzes voice information during a call in real time. This system consists of a user terminal including a microphone and communication devices for acquiring voice information, and a server for analyzing the voice data and evaluating the risk of fraud.

[0108] The server uses speech recognition software and natural language processing (NLP) algorithms to convert speech information into text and analyze its sentiment. This analysis assesses the likelihood of fraud, and if it exceeds a certain threshold, a warning message using voice reproduction is generated for the fraudster. This warning message is sent to the fraudster in real time using synthesized speech technology.

[0109] The user's device immediately receives a notification if fraud is suspected, and reassuring information is displayed. This information is displayed dynamically based on voice analysis results, so the user can always stay informed of the latest situation. Furthermore, dynamic warning messages are sent directly to fraudsters, acting to deter fraudulent activity.

[0110] As a concrete example, when a user receives a call from an unknown number, the server analyzes the conversation and determines that it may be a scam. As a result, the server issues a synthesized voice warning that "this call is being monitored," and simultaneously displays a notification to the user stating "this may be a scam." This kind of action allows users to prevent themselves from becoming victims of fraud.

[0111] An example of a prompt message for a generative AI model is: "Audio data has been input. Perform speech recognition and convert it to text data. Then, use sentiment analysis to assess the fraud risk, and if the risk is high, generate a warning message and notify the user." This prompt allows the server to perform appropriate analysis and evaluation.

[0112] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0113] Step 1:

[0114] The server receives in-call audio information transmitted from the user terminal in real time. The input audio information is transmitted in digital format and processed by the server's speech recognition software. At this stage, noise reduction is performed on the audio signal to obtain clear audio data.

[0115] Step 2:

[0116] The server converts the received clear audio data into text information using speech recognition technology. By converting the audio data to text, natural language processing algorithms become applicable. In this process, the entire content of the audio is recorded as text and used as input data for the next analysis step.

[0117] Step 3:

[0118] The server analyzes the transcribed audio information using natural language processing (NLP) algorithms to identify the speaker's emotions. Sentiment analysis evaluates keywords and context within the text, particularly extracting feelings of confusion and anxiety associated with fraud. This outputs the emotional data necessary for assessing fraud risk.

[0119] Step 4:

[0120] The server assesses the likelihood of fraud based on emotions. The assessment algorithm compares the sentiment analysis results with known fraud patterns to quantify the fraud risk. This output is used to determine whether the fraud risk exceeds a set threshold.

[0121] Step 5:

[0122] The server uses voice synthesis technology to generate and send a warning message to the fraudulent caller when the fraud risk exceeds a certain threshold. A synthesized voice engine creates messages such as "This call is being monitored" and transmits them to the fraudulent caller in real time.

[0123] Step 6:

[0124] The device displays a notification to the user regarding identified suspected fraud. This notification is displayed on the device screen as a visually reassuring message. By reviewing the information on the screen, the user can prepare for the possibility of a fraudulent call.

[0125] 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.

[0126] This invention is an AI agent system for preventing fraudulent phone calls, integrating voice data acquisition, emotion recognition, fraud risk assessment, voice emulation, user notification, and user emotion analysis using an emotion engine. The system consists of a server, a terminal, and a user interface.

[0127] The server first acquires audio data of calls made on the user's terminal in real time. This audio data is then subjected to noise reduction processing and converted into text data using speech recognition technology. This converted text data is then analyzed using natural language processing (NLP) technology to identify the speaker's emotions, particularly confusion or anxiety.

[0128] In addition, the emotion engine analyzes the user's voice and speech patterns, and also evaluates the user's emotional state. By identifying the user's stress and tension, it more accurately assesses the likelihood of fraud based on fraud risk and emotional state. If the fraud risk exceeds a certain threshold, the server dynamically changes the content of the call using AI-powered voice emulation. This sends a message that will cause scammers to hesitate.

[0129] As a concrete example, when a user receives a potentially fraudulent phone call, the server uses an emotion engine to detect signs of the user's anxiety and thoroughly reassess the fraud risk. If it determines that the fraud risk is high, it then uses voice emulation to broadcast a synthesized voice message such as "This call is being monitored" to psychologically unsettle the fraudster. Furthermore, a reassuring message is displayed on the user's device to alleviate their anxiety.

[0130] Thus, the present invention can prevent fraudulent activities and enhance psychological protection for users by recognizing the emotional state of the user and implementing dynamic fraud prevention measures based on that state.

[0131] The following describes the processing flow.

[0132] Step 1:

[0133] The server receives a call initiation notification from the user's device and begins acquiring audio data. It receives the audio stream in real time and prepares it for data processing.

[0134] Step 2:

[0135] The server performs noise reduction on the acquired audio data to generate a clear audio signal. This prepares the system for improving the accuracy of speech recognition.

[0136] Step 3:

[0137] The server converts clear audio data into text data using speech recognition technology. The converted text is then used for subsequent analysis.

[0138] Step 4:

[0139] The server analyzes text data using natural language processing (NLP) techniques to identify the speaker's emotions. In particular, it focuses on analyzing emotions such as confusion and anxiety.

[0140] Step 5:

[0141] The server uses an emotion engine to analyze the user's voice tone and speech patterns to recognize the user's emotional state. This allows it to determine whether the user is experiencing stress or tension.

[0142] Step 6:

[0143] The server assesses the fraud risk based on the analysis results. It compares the results against known fraudulent phrases and patterns to calculate a risk score. It also considers the user's emotional state to improve the accuracy of the assessment.

[0144] Step 7:

[0145] If the fraud risk exceeds a set threshold, the server modifies the conversation using AI-powered voice emulation. Specifically, it generates synthesized messages such as "This call is being monitored by security" to psychologically unsettle the fraudster.

[0146] Step 8:

[0147] The server notifies the user's device of the suspected fraud and displays a reassuring message. This allows the user to take appropriate action quickly.

[0148] Step 9:

[0149] The server logs call data and analysis results, preparing for future analysis and provision to the police. The data is stored securely with user privacy in mind.

[0150] (Example 2)

[0151] 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".

[0152] In modern times, fraudulent activities such as special fraud and wire transfer fraud are on the rise. These scams are often conducted via voice, and victims are frequently deceived and lose money. With current technology, it is difficult to detect the risk of fraud in real time during a call and take appropriate action. Furthermore, it is not easy for victims themselves to recognize the risk of fraud, which can cause psychological distress. Therefore, there is a need to prevent fraudulent activities and provide users with a sense of security.

[0153] 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.

[0154] In this invention, the server includes means for acquiring voice information in a call, means for applying noise reduction processing to the acquired voice information, and means for converting the noise-reduced voice information into text information using speech recognition technology. This enables real-time assessment of fraud risk and, if necessary, generates messages using synthesized speech to deter fraudulent activity, thereby providing psychological protection to the user.

[0155] "Audio information" refers to data based on the wavelengths of sounds contained during a call, and is used to represent the content of the conversation.

[0156] "Noise reduction processing" is a technique used to reduce or remove unnecessary background noise and interfering sounds from audio information in order to obtain a clear audio signal.

[0157] "Speech recognition technology" is a technology that analyzes speech information and converts its linguistic content into text format.

[0158] "Text information" refers to string data in which speech information converted by speech recognition technology is linguistically represented.

[0159] "Emotional analysis" is an analytical technique used to identify a speaker's emotional state from text or audio information.

[0160] "Vocal characteristics" refer to information that describes the tone, speed, and pitch of a speaker's voice, and are used to analyze their emotional state and intentions.

[0161] "Fraud risk" is an indicator that shows the likelihood of fraudulent activity based on the content of the call and the emotional state of the speaker.

[0162] "Synthesized speech" refers to artificially generated voice data, which is an acoustic signal used to convey a specific message.

[0163] "Psychological security" refers to the sense of safety and peace of mind provided to system users, and a state in which they feel protected from fraudulent activities.

[0164] This invention relates to a system for preventing special fraud, and includes a series of processes for acquiring and analyzing voice information and assessing fraud risk. It consists of three main elements: a server, a terminal, and a user.

[0165] Server Functions

[0166] The server focuses on acquiring, analyzing, and evaluating audio information. The server receives audio information transmitted in real time from terminals. Digital signal processing technology is used for audio processing, and a dedicated algorithm is implemented to remove noise from the audio information. For speech recognition technology, a cloud-based speech recognition service is used to convert speech into text. Specifically, Google Cloud Speech-to-Text API or other commercial speech recognition engines can be used.

[0167] The server further analyzes text information based on sentiment analysis technology to identify emotions from the user's statements. This process utilizes natural language processing libraries, with NLTK and other machine learning models proving effective. Additionally, open-source speech characteristic analysis tools are used to analyze the user's voice characteristics, enabling the detection of user stress, tension, and other factors. In assessing fraud risk, these analysis results are integrated to determine the likelihood of fraudulent activity.

[0168] If the risk of fraud is high, the server uses AI speech synthesis technology to generate and transmit a synthesized voice message, such as "This call is being monitored." Commercial speech synthesis engines such as Amazon Polly can be used for this purpose. The generated voice message aims to have a psychological effect to deter fraudulent activity.

[0169] Device functions

[0170] The terminal functions as an interface with the user, collecting and transmitting voice information. Voice information acquired during a call is filtered for noise before being sent to the server. It also plays a role in alleviating anxiety by displaying reassuring messages received from the server to the user.

[0171] User roles

[0172] Users are users of the system and are monitored by the system during calls. To minimize the burden on users while providing protection from fraudulent activities, the system is designed to make them feel safe through reassuring messages.

[0173] Specific example

[0174] For example, if a user receives a suspicious call, the server immediately analyzes the audio and text to detect signs of fraud. If it is deemed high-risk, a synthesized voice message such as "This call is being monitored" is sent to the caller. In addition, the user's device displays "You are currently on a secure call" to provide reassurance.

[0175] Examples of prompts for generative AI models

[0176] "What is needed to develop a prototype of a call content analysis system for evaluating the risk of fraudulent phone calls?"

[0177] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0178] Step 1: Obtain audio information

[0179] The device detects the start of a call and captures the call's audio information in real time using its internal microphone. The captured audio information is then sent to the server as a digital signal. The input is raw audio, and the output is digitized audio data.

[0180] Step 2: Noise Reduction and Speech Recognition

[0181] The server performs noise reduction processing on the received audio data. This reduces background noise and improves the accuracy of speech recognition. Next, the noise-reduced audio data is converted into text information using speech recognition technology. Specifically, a speech recognition API is used to convert sound waves into character data. The input is noise-reduced audio data, and the output is text data.

[0182] Step 3: Emotion Analysis

[0183] The server analyzes the generated text information to identify the speaker's emotions. Using natural language processing techniques, it calculates emotional indicators from words and phrases in the text. A sentiment analysis library is used for this analysis. The input is text data, and the output is an emotional score and an inferred emotional state.

[0184] Step 4: User Voice Characteristics Assessment

[0185] The server uses a voice analysis tool to evaluate the user's voice characteristics. It analyzes voice tone, tempo, pitch, etc., to identify signs of stress or tension. The input is voice data, and the output is the evaluation result regarding the user's voice characteristics.

[0186] Step 5: Assessing the risk of fraud

[0187] The server integrates the sentiment analysis results and voice characteristic evaluation results to assess fraud risk. This assessment uses an algorithm that compares the results with fraud patterns to calculate a risk score. The inputs are the sentiment score and voice characteristic evaluation results, and the output is the fraud risk score.

[0188] Step 6: Intervention using voice emulation

[0189] If the server determines that there is a high risk of fraud, it uses AI speech synthesis technology to generate a message such as "This call is being monitored" and plays the synthesized voice to the caller. The input is the fraud risk score, and the output is synthesized voice data.

[0190] Step 7: Displaying user notifications and safety messages

[0191] The terminal, based on information sent from the server, displays a reassuring message to the user on the screen, such as "You are currently on a secure call," providing a sense of psychological security. The input is instruction data from the server, and the output is the reassuring message displayed on the screen.

[0192] (Application Example 2)

[0193] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0194] Fraudulent use of electronic payments poses a significant threat to users. Preventing losses due to fraudulent activity and ensuring user confidence are crucial. However, current technology lacks the means to rapidly assess the possibility of fraud in real time and respond appropriately to users. As a result, users are not fully protected from fraudulent activity.

[0195] 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.

[0196] In this invention, the server includes a processing unit for acquiring voice data, a processing unit for analyzing the acquired voice data and identifying emotions, and a processing unit for evaluating the possibility of fraudulent activity based on the identified emotions. This makes it possible to detect changes in the user's emotions in real time and respond quickly if there are signs of fraudulent activity.

[0197] A "processing device for acquiring voice data" is a device that has the function of collecting voice information from phone calls and conversations and converting it into a digital data format that can be used within the system.

[0198] A "processing device for identifying emotions" is a device that has the function of identifying the emotional state of a speaker based on collected audio data and analyzing it as data.

[0199] A "processing device for evaluating the likelihood of fraudulent activity" is an electronic device that evaluates the likelihood of fraud in transactions or communications based on identified emotions and other relevant data, and makes decisions on actions based on that evaluation.

[0200] A "processing device that adjusts dialogue content using speech synthesis technology" is a device that operates technology used to generate voice data and send warnings and notifications to the user.

[0201] A "device that displays information to provide a sense of security" is a device that has the function of visually presenting messages and information to alleviate anxiety in the user.

[0202] This invention relates to an information processing system for preventing misuse. The invention mainly consists of a server and a user terminal, and performs a series of processes to analyze voice data, identify emotions to evaluate the possibility of misuse, and notify the user.

[0203] The server first acquires audio data transmitted from the user's terminal in real time. Mobile devices such as smartphones and tablets can be used for this process. The acquired audio data is then noise-reduced using FFT-based noise reduction technology. This audio data is then converted into text using the Google Cloud Speech-to-Text API.

[0204] Next, the server analyzes the converted text information using natural language processing techniques such as Hugging Face's Transformers library to identify the user's emotions. Based on the results of this emotion analysis, the server assesses the likelihood of fraudulent activity. This assessment includes comparing and matching the data against pre-registered, known fraudulent patterns.

[0205] If the server determines that there is a high probability of fraud, it will send a warning to the user's terminal via synthesized voice. The user will be notified of a message such as, "This transaction is being monitored. Do you really want to proceed?", which is generated using speech synthesis technology.

[0206] The user's terminal simultaneously displays visually reassuring information to reduce the likelihood of the user becoming involved in fraudulent activity. Furthermore, if fraudulent activity is suspected, the transaction is temporarily suspended, giving the user time to review the situation.

[0207] For example, when making a high-value online purchase, the system can issue a warning if the user is likely to be redirected to a fraudulent website, allowing them to reconfirm the transaction and prevent fraud.

[0208] An example of a prompt message is: "Create a program that analyzes the user's voice in real time, detects changes in emotion, and outputs a warning message in real time if there is a possibility of fraud."

[0209] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0210] Step 1:

[0211] The device acquires the user's voice data. This voice data is an analog signal and is collected in real time through the device's microphone. The input is the user's voice signal, and the output is digital voice data sent to the server. Noise reduction technology is used to reduce background noise and ensure clear voice acquisition.

[0212] Step 2:

[0213] The server converts the acquired audio data into text using the Google Cloud Speech-to-Text API. The input is digital audio data, and the output is text data generated as a result of analysis by the API. This conversion process makes it possible to accurately extract the user's speech from the audio.

[0214] Step 3:

[0215] The server performs emotion recognition processing based on the converted text data. Here, natural language processing technologies such as Hugging Face's Transformers are used to analyze and identify the user's emotional state (e.g., anxiety, confusion). The input is text data, and the output is emotion labels and emotion scores. This allows for an accurate understanding of the user's psychological state.

[0216] Step 4:

[0217] The server assesses the likelihood of fraud by comparing the user's emotional state with known fraud patterns. Inputs include emotional labels, emotional scores, and past fraud pattern data. Output is a risk score, which quantifies the likelihood of fraud. If this score exceeds a certain threshold, the server proceeds to the next step.

[0218] Step 5:

[0219] If the server determines that there is a high probability of fraudulent activity, it will use speech synthesis technology to create a warning message. For example, it might generate a warning such as "This transaction is being monitored." The input is a predefined message template and a fraud score, and the output is a synthesized voice file. This allows the user to recognize potential fraud.

[0220] Step 6:

[0221] The terminal receives warning messages from the server and notifies the user visually and audibly. Input is synthesized speech files and text messages, and output is the information presented through the terminal's speaker or display. This draws the user's attention to fraudulent activity and prompts them to review or cancel transactions as needed.

[0222] 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.

[0223] 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.

[0224] 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.

[0225] [Second Embodiment]

[0226] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0227] 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.

[0228] 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).

[0229] 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.

[0230] 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.

[0231] 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).

[0232] 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.

[0233] 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.

[0234] 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.

[0235] 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.

[0236] 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.

[0237] 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".

[0238] This invention provides an AI agent system for effectively combating fraudulent phone calls. This system is server-centric and operates in conjunction with the user's terminal. The program's processing and its application are described in detail below.

[0239] The server acquires audio data of calls initiated from the user's terminal in real time. This audio data is converted into a clean signal using noise reduction technology and then converted into text data using speech recognition technology. This text data is analyzed using natural language processing (NLP) to identify the speaker's emotions. The emotion identification algorithm particularly identifies emotions such as confusion and anxiety, and this information is used to assess the risk of fraud.

[0240] When the fraud risk exceeds a certain threshold, the server dynamically adjusts the conversation using voice emulation. This voice emulation generates synthesized speech and sends messages to deter fraudsters. Specifically, it generates messages such as, "This call is being monitored by security." During this process, the user's device also displays reassuring messages notifying them that the caller may be fraudulent and to alleviate their anxiety.

[0241] For example, if a user receives a phone call that may be fraudulent, the server automatically executes the process described above and intervenes with voice emulation to thwart the fraudster's intentions. As a result, it is possible to prevent fraud and reduce the user's psychological burden.

[0242] In this way, the system proposed in the patent provides an effective defense against telephone fraud by integrating speech recognition, natural language processing, and speech synthesis technologies. This enhances user trust and contributes to improved social safety.

[0243] The following describes the processing flow.

[0244] Step 1:

[0245] The server acquires audio data for calls initiated on the user's terminal. The audio signal is sent to the server in real time and prepared for data processing.

[0246] Step 2:

[0247] The server performs noise reduction on the acquired audio data to clarify the audio signal. This generates clean audio data that improves the accuracy of speech recognition.

[0248] Step 3:

[0249] The server converts clean audio data into text data using speech recognition technology. This process extracts the content of the audio as a string of characters, preparing it for subsequent processing.

[0250] Step 4:

[0251] The server analyzes the generated text data using natural language processing (NLP) techniques to identify the speaker's emotions. In particular, it scrutinizes feelings of confusion and anxiety, and measures their intensity if present.

[0252] Step 5:

[0253] The server assesses fraud risk based on the analyzed sentiment data. It calculates a risk score, which also involves matching the data against existing fraud patterns to detect fraudulent phrases and unusual patterns.

[0254] Step 6:

[0255] If the server determines that the fraud risk has exceeded a set threshold, it dynamically alters the call content using AI-powered voice emulation. At this time, it sends a counter-message to the fraudster in synthesized speech.

[0256] Step 7:

[0257] The server displays a reassuring message on the user's device to alleviate their concerns and notify them of the potential for fraud. This allows the user to take appropriate action quickly.

[0258] Step 8:

[0259] The server ultimately logs the call and stores data for analyzing fraud risk. This allows for later review of the details and, if necessary, assist in legal proceedings.

[0260] (Example 1)

[0261] 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".

[0262] In modern society, fraudulent activities conducted via communication networks are on the rise, with phone-based fraud being a particular social problem. These scams are becoming more sophisticated, making it difficult for ordinary users to detect them on their own. Furthermore, fraudulent activities induce psychological confusion and anxiety in users, making it difficult for them to take prompt and appropriate action. Therefore, there is a need for a system that can detect potential fraud on behalf of users and respond quickly.

[0263] 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.

[0264] In this invention, the server includes means for acquiring voice information during a call, means for removing noise from the acquired voice information and converting it into text information using speech recognition technology, means for analyzing the text information to identify emotions, means for modifying the conversation content using speech synthesis when the risk of fraud exceeds a certain threshold, and means for displaying information to notify the user of the suspicion of fraud and alleviate their anxiety. This makes it possible to detect the risk of fraud in real time during a call and ensure the safety of the user.

[0265] A "call" is the act of transmitting voice information bidirectionally over a communication network, and is a means of communication through voice.

[0266] "Auditory information" refers to data that is recorded or transmitted as sound, including human speech and other sound signals.

[0267] "Noise reduction" is a process that makes audio clearer by removing unwanted background noise and interference sounds from audio data.

[0268] "Speech recognition technology" is a technology that converts speech information into a digital format, analyzes the content of the speech, and converts it into text information.

[0269] "Textual information" refers to string data extracted through speech recognition, and includes information expressed in human language.

[0270] "Emotion identification" is the process of analyzing textual information and audio data to identify the speaker's emotional state (for example, confusion, anxiety, joy, etc.).

[0271] "Fraud risk" is an indicator that numerically or qualitatively assesses the likelihood that the person on the other end of a call is likely to commit fraud.

[0272] A "threshold" is a boundary value that is set as a reference point, and when that value is exceeded, a specific action is performed.

[0273] "Speech synthesis" is a technology that converts text data into speech and generates new speech data.

[0274] "Changing the content of a conversation" refers to the act of manipulating or editing voice messages sent during a call to convey intended information.

[0275] A "user" is a person or group that makes a call using this system and is protected from fraud.

[0276] "Information that alleviates anxiety" refers to messages and notifications presented in a way that provides users with a sense of psychological reassurance.

[0277] This invention is a system for detecting call fraud via a communication network and protecting users. This system operates in cooperation with a server, a user's terminal, and the network.

[0278] The server acquires audio information from calls transmitted from the user's terminal in real time. The audio information is streamed to the server via a digital processing unit. This system uses speech recognition technology to convert the audio information into text. Specifically, it uses a processing library to remove noise and the Google Cloud Speech-to-Text API to transcribe the text.

[0279] Text information undergoes sentiment analysis on the server using natural language processing (NLP) techniques. The system utilizes Python's natural language processing libraries, NLTK and spaCy, to identify the speaker's emotions. This system specifically assesses fraud risk by detecting emotions such as confusion and anxiety. If the fraud risk assessment exceeds a threshold, speech synthesis technology is used to generate a deterrent warning message for fraudsters. Specifically, speech synthesis software is used to broadcast the message, "This call is being monitored by security."

[0280] The server also notifies the user's device of the suspected fraud and displays a reassuring message on the screen. For example, it might say, "This may be a suspicious call. Please rest assured, we are handling it."

[0281] One concrete example of this system's use is that when a user receives a suspicious phone call, they can input a prompt message into the AI ​​model such as, "I have been notified that there has been an unauthorized payment on my credit card. Please analyze the sentiment and assess the risk of fraud."

[0282] Without the user performing any special operations, the system automatically detects the possibility of fraud and takes appropriate actions to prevent fraud losses and enable the user to continue the call with confidence. This contributes to improving the safety and social reliability of the user.

[0283] The flow of the specific process in Example 1 will be described using FIG. 11.

[0284] Step 1:

[0285] The server acquires the voice information of the call transmitted from the user's terminal. This voice information is streamed via the VoIP protocol and input to the server in digital format. As output, the server prepares to pass the voice data to the next processing step while maintaining real-time performance.

[0286] Step 2:

[0287] The server performs a process to remove noise from the acquired voice information. Specifically, an audio processing library is used to apply a filter to reduce background noise and process the voice to be clear. The input is voice data containing noise, and the output is clean voice data with minimized noise.

[0288] Step 3:

[0289] The server converts the clean voice data into character information using voice recognition technology. Specifically, voice recognition software is used to perform a process of converting the voice data into text format. The input is the processed voice data, and the output is the corresponding text data. This conversion enables subsequent natural language processing.

[0290] Step 4:

[0291] The server applies natural language processing (NLP) techniques to the generated text information to analyze emotions. It uses a Python NLP library to analyze the text, focusing on specific emotions, particularly confusion and anxiety. The input is text data, and the output is the identification and intensity of the detected emotions.

[0292] Step 5:

[0293] The server assesses fraud risk using the results of sentiment analysis. A fraud risk assessment algorithm is used to quantify the risk level based on the identified emotions. The input is identified sentiment information, and the output is a numerical evaluation of fraud risk.

[0294] Step 6:

[0295] If the fraud risk assessment exceeds a certain threshold, the server uses speech synthesis technology to generate a warning message and insert it into the call. Specifically, it uses speech synthesis software to synthesize a message such as "This call is being monitored by security" to deter fraudsters. The input is the result of the fraud risk assessment, and the output is the generated warning message.

[0296] Step 7:

[0297] The server notifies the user's device of a suspected scam and displays a reassuring message on the screen. This notification allows the user to understand the potential for fraud and take prompt action. The input is the result of the fraud risk assessment, and the output is the reassuring message displayed on the device.

[0298] (Application Example 1)

[0299] 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."

[0300] In recent years, the tactics of special fraud calls have become sophisticated, and it is not easy for individual users to deal with them. Therefore, an effective means to prevent damage caused by telephone fraud is required. The purpose of the present invention is to provide a new defense system that quickly and accurately evaluates the possibility of telephone fraud and protects users.

[0301] The specific processing by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0302] In this invention, the server includes means for acquiring voice information in a call, means for analyzing the acquired voice information to identify emotions, means for evaluating the possibility of fraud based on the identified emotions, means for changing the conversation content using voice reproduction when the possibility of fraud exceeds a certain criterion, means for displaying information that notifies the user of suspicion of fraud and relieves anxiety, and means for transmitting a dynamic warning message to the fraud communicator. Thereby, it becomes possible to effectively repel special fraud calls and provide peace of mind to users.

[0303] "Voice information in a call" refers to the entire voice data transmitted via a communication means.

[0304] "Means for acquiring" refers to a technology for receiving and storing voice information in real time.

[0305] "Means for analyzing and identifying emotions" refers to an algorithm for processing voice data to identify the speaker's psychological state.

[0306] "Means for evaluating the possibility of fraud" refers to a method of quantifying the risk of fraud based on the analyzed emotion information.

[0307] "Means for changing the conversation content using voice reproduction" refers to a technology for generating synthetic voice and automatically providing a new message to the call partner.

[0308] "A means of notifying users of suspected fraud and displaying information to alleviate their anxiety" refers to a method of providing reassurance by displaying a warning message on the user's device when fraud is suspected.

[0309] "Means of sending dynamic warning messages to fraudulent communicators" refers to technology that, when fraudulent activity is detected, sends a message in real time to warn those attempting to commit fraud.

[0310] To implement this invention, it is necessary to construct a system that acquires and analyzes voice information during a call in real time. This system consists of a user terminal including a microphone and communication devices for acquiring voice information, and a server for analyzing the voice data and evaluating the risk of fraud.

[0311] The server uses speech recognition software and natural language processing (NLP) algorithms to convert speech information into text and analyze its sentiment. This analysis assesses the likelihood of fraud, and if it exceeds a certain threshold, a warning message using voice reproduction is generated for the fraudster. This warning message is sent to the fraudster in real time using synthesized speech technology.

[0312] The user's device immediately receives a notification if fraud is suspected, and reassuring information is displayed. This information is displayed dynamically based on voice analysis results, so the user can always stay informed of the latest situation. Furthermore, dynamic warning messages are sent directly to fraudsters, acting to deter fraudulent activity.

[0313] As a concrete example, when a user receives a call from an unknown number, the server analyzes the conversation and determines that it may be a scam. As a result, the server issues a synthesized voice warning that "this call is being monitored," and simultaneously displays a notification to the user stating "this may be a scam." This kind of action allows users to prevent themselves from becoming victims of fraud.

[0314] An example of a prompt message for a generative AI model is: "Audio data has been input. Perform speech recognition and convert it to text data. Then, use sentiment analysis to assess the fraud risk, and if the risk is high, generate a warning message and notify the user." This prompt allows the server to perform appropriate analysis and evaluation.

[0315] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0316] Step 1:

[0317] The server receives in-call audio information transmitted from the user terminal in real time. The input audio information is transmitted in digital format and processed by the server's speech recognition software. At this stage, noise reduction is performed on the audio signal to obtain clear audio data.

[0318] Step 2:

[0319] The server converts the received clear audio data into text information using speech recognition technology. By converting the audio data to text, natural language processing algorithms become applicable. In this process, the entire content of the audio is recorded as text and used as input data for the next analysis step.

[0320] Step 3:

[0321] The server analyzes the transcribed audio information using natural language processing (NLP) algorithms to identify the speaker's emotions. Sentiment analysis evaluates keywords and context within the text, particularly extracting feelings of confusion and anxiety associated with fraud. This outputs the emotional data necessary for assessing fraud risk.

[0322] Step 4:

[0323] The server assesses the likelihood of fraud based on emotions. The assessment algorithm compares the sentiment analysis results with known fraud patterns to quantify the fraud risk. This output is used to determine whether the fraud risk exceeds a set threshold.

[0324] Step 5:

[0325] The server uses voice synthesis technology to generate and send a warning message to the fraudulent caller when the fraud risk exceeds a certain threshold. A synthesized voice engine creates messages such as "This call is being monitored" and transmits them to the fraudulent caller in real time.

[0326] Step 6:

[0327] The device displays a notification to the user regarding identified suspected fraud. This notification is displayed on the device screen as a visually reassuring message. By reviewing the information on the screen, the user can prepare for the possibility of a fraudulent call.

[0328] 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.

[0329] This invention is an AI agent system for preventing fraudulent phone calls, integrating voice data acquisition, emotion recognition, fraud risk assessment, voice emulation, user notification, and user emotion analysis using an emotion engine. The system consists of a server, a terminal, and a user interface.

[0330] The server first acquires audio data of calls made on the user's terminal in real time. This audio data is then subjected to noise reduction processing and converted into text data using speech recognition technology. This converted text data is then analyzed using natural language processing (NLP) technology to identify the speaker's emotions, particularly confusion or anxiety.

[0331] In addition, the emotion engine analyzes the user's voice and speech patterns, and also evaluates the user's emotional state. By identifying the user's stress and tension, it more accurately assesses the likelihood of fraud based on fraud risk and emotional state. If the fraud risk exceeds a certain threshold, the server dynamically changes the content of the call using AI-powered voice emulation. This sends a message that will cause scammers to hesitate.

[0332] As a concrete example, when a user receives a potentially fraudulent phone call, the server uses an emotion engine to detect signs of the user's anxiety and thoroughly reassess the fraud risk. If it determines that the fraud risk is high, it then uses voice emulation to broadcast a synthesized voice message such as "This call is being monitored" to psychologically unsettle the fraudster. Furthermore, a reassuring message is displayed on the user's device to alleviate their anxiety.

[0333] Thus, the present invention can prevent fraudulent activities and enhance psychological protection for users by recognizing the emotional state of the user and implementing dynamic fraud prevention measures based on that state.

[0334] The following describes the processing flow.

[0335] Step 1:

[0336] The server receives a call initiation notification from the user's device and begins acquiring audio data. It receives the audio stream in real time and prepares it for data processing.

[0337] Step 2:

[0338] The server performs noise reduction on the acquired audio data to generate a clear audio signal. This prepares the system for improving the accuracy of speech recognition.

[0339] Step 3:

[0340] The server converts clear audio data into text data using speech recognition technology. The converted text is then used for subsequent analysis.

[0341] Step 4:

[0342] The server analyzes text data using natural language processing (NLP) techniques to identify the speaker's emotions. In particular, it focuses on analyzing emotions such as confusion and anxiety.

[0343] Step 5:

[0344] The server uses an emotion engine to analyze the user's voice tone and speech patterns to recognize the user's emotional state. This allows it to determine whether the user is experiencing stress or tension.

[0345] Step 6:

[0346] The server assesses the fraud risk based on the analysis results. It compares the results against known fraudulent phrases and patterns to calculate a risk score. It also considers the user's emotional state to improve the accuracy of the assessment.

[0347] Step 7:

[0348] If the fraud risk exceeds a set threshold, the server modifies the conversation using AI-powered voice emulation. Specifically, it generates synthesized messages such as "This call is being monitored by security" to psychologically unsettle the fraudster.

[0349] Step 8:

[0350] The server notifies the user's device of the suspected fraud and displays a reassuring message. This allows the user to take appropriate action quickly.

[0351] Step 9:

[0352] The server logs call data and analysis results, preparing for future analysis and provision to the police. The data is stored securely with user privacy in mind.

[0353] (Example 2)

[0354] 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".

[0355] In modern times, fraudulent activities such as special fraud and wire transfer fraud are on the rise. These scams are often conducted via voice, and victims are frequently deceived and lose money. With current technology, it is difficult to detect the risk of fraud in real time during a call and take appropriate action. Furthermore, it is not easy for victims themselves to recognize the risk of fraud, which can cause psychological distress. Therefore, there is a need to prevent fraudulent activities and provide users with a sense of security.

[0356] 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.

[0357] In this invention, the server includes means for acquiring voice information in a call, means for applying noise reduction processing to the acquired voice information, and means for converting the noise-reduced voice information into text information using speech recognition technology. This enables real-time assessment of fraud risk and, if necessary, generates messages using synthesized speech to deter fraudulent activity, thereby providing psychological protection to the user.

[0358] "Audio information" refers to data based on the wavelengths of sounds contained during a call, and is used to represent the content of the conversation.

[0359] "Noise reduction processing" is a technique used to reduce or remove unnecessary background noise and interfering sounds from audio information in order to obtain a clear audio signal.

[0360] "Speech recognition technology" is a technology that analyzes speech information and converts its linguistic content into text format.

[0361] "Text information" refers to string data in which speech information converted by speech recognition technology is linguistically represented.

[0362] "Emotional analysis" is an analytical technique used to identify a speaker's emotional state from text or audio information.

[0363] "Vocal characteristics" refer to information that describes the tone, speed, and pitch of a speaker's voice, and are used to analyze their emotional state and intentions.

[0364] "Fraud risk" is an indicator that shows the likelihood of fraudulent activity based on the content of the call and the emotional state of the speaker.

[0365] "Synthesized speech" refers to artificially generated voice data, which is an acoustic signal used to convey a specific message.

[0366] "Psychological security" refers to the sense of safety and peace of mind provided to system users, and a state in which they feel protected from fraudulent activities.

[0367] This invention relates to a system for preventing special fraud, and includes a series of processes for acquiring and analyzing voice information and assessing fraud risk. It consists of three main elements: a server, a terminal, and a user.

[0368] Server Functions

[0369] The server focuses on acquiring, analyzing, and evaluating audio information. The server receives audio information transmitted in real time from terminals. Digital signal processing technology is used for audio processing, and a dedicated algorithm is implemented to remove noise from the audio information. For speech recognition technology, a cloud-based speech recognition service is used to convert speech into text. Specifically, Google Cloud Speech-to-Text API or other commercial speech recognition engines can be used.

[0370] The server further analyzes text information based on sentiment analysis technology to identify emotions from the user's statements. This process utilizes natural language processing libraries, with NLTK and other machine learning models proving effective. Additionally, open-source speech characteristic analysis tools are used to analyze the user's voice characteristics, enabling the detection of user stress, tension, and other factors. In assessing fraud risk, these analysis results are integrated to determine the likelihood of fraudulent activity.

[0371] If the risk of fraud is high, the server uses AI speech synthesis technology to generate and transmit a synthesized voice message, such as "This call is being monitored." Commercial speech synthesis engines such as Amazon Polly can be used for this purpose. The generated voice message aims to have a psychological effect to deter fraudulent activity.

[0372] Device functions

[0373] The terminal functions as an interface with the user, collecting and transmitting voice information. Voice information acquired during a call is filtered for noise before being sent to the server. It also plays a role in alleviating anxiety by displaying reassuring messages received from the server to the user.

[0374] User roles

[0375] Users are users of the system and are monitored by the system during calls. To minimize the burden on users while providing protection from fraudulent activities, the system is designed to make them feel safe through reassuring messages.

[0376] Specific example

[0377] For example, if a user receives a suspicious call, the server immediately analyzes the audio and text to detect signs of fraud. If it is deemed high-risk, a synthesized voice message such as "This call is being monitored" is sent to the caller. In addition, the user's device displays "You are currently on a secure call" to provide reassurance.

[0378] Examples of prompts for generative AI models

[0379] "What is needed to develop a prototype of a call content analysis system for evaluating the risk of fraudulent phone calls?"

[0380] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0381] Step 1: Obtain audio information

[0382] The device detects the start of a call and captures the call's audio information in real time using its internal microphone. The captured audio information is then sent to the server as a digital signal. The input is raw audio, and the output is digitized audio data.

[0383] Step 2: Noise Reduction and Speech Recognition

[0384] The server performs noise reduction processing on the received audio data. This reduces background noise and improves the accuracy of speech recognition. Next, the noise-reduced audio data is converted into text information using speech recognition technology. Specifically, a speech recognition API is used to convert sound waves into character data. The input is noise-reduced audio data, and the output is text data.

[0385] Step 3: Emotion Analysis

[0386] The server analyzes the generated text information to identify the speaker's emotions. Using natural language processing techniques, it calculates emotional indicators from words and phrases in the text. A sentiment analysis library is used for this analysis. The input is text data, and the output is an emotional score and an inferred emotional state.

[0387] Step 4: User Voice Characteristics Assessment

[0388] The server uses a voice analysis tool to evaluate the user's voice characteristics. It analyzes voice tone, tempo, pitch, etc., to identify signs of stress or tension. The input is voice data, and the output is the evaluation result regarding the user's voice characteristics.

[0389] Step 5: Assessing the risk of fraud

[0390] The server integrates the sentiment analysis results and voice characteristic evaluation results to assess fraud risk. This assessment uses an algorithm that compares the results with fraud patterns to calculate a risk score. The inputs are the sentiment score and voice characteristic evaluation results, and the output is the fraud risk score.

[0391] Step 6: Intervention using voice emulation

[0392] If the server determines that there is a high risk of fraud, it uses AI speech synthesis technology to generate a message such as "This call is being monitored" and plays the synthesized voice to the caller. The input is the fraud risk score, and the output is synthesized voice data.

[0393] Step 7: Displaying user notifications and safety messages

[0394] The terminal, based on information sent from the server, displays a reassuring message to the user on the screen, such as "You are currently on a secure call," providing a sense of psychological security. The input is instruction data from the server, and the output is the reassuring message displayed on the screen.

[0395] (Application Example 2)

[0396] 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."

[0397] Fraudulent use of electronic payments poses a significant threat to users. Preventing losses due to fraudulent activity and ensuring user confidence are crucial. However, current technology lacks the means to rapidly assess the possibility of fraud in real time and respond appropriately to users. As a result, users are not fully protected from fraudulent activity.

[0398] 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.

[0399] In this invention, the server includes a processing unit for acquiring voice data, a processing unit for analyzing the acquired voice data and identifying emotions, and a processing unit for evaluating the possibility of fraudulent activity based on the identified emotions. This makes it possible to detect changes in the user's emotions in real time and respond quickly if there are signs of fraudulent activity.

[0400] A "processing device for acquiring voice data" is a device that has the function of collecting voice information from phone calls and conversations and converting it into a digital data format that can be used within the system.

[0401] A "processing device for identifying emotions" is a device that has the function of identifying the emotional state of a speaker based on collected audio data and analyzing it as data.

[0402] A "processing device for evaluating the likelihood of fraudulent activity" is an electronic device that evaluates the likelihood of fraud in transactions or communications based on identified emotions and other relevant data, and makes decisions on actions based on that evaluation.

[0403] A "processing device that adjusts dialogue content using speech synthesis technology" is a device that operates technology used to generate voice data and send warnings and notifications to the user.

[0404] A "device that displays information to provide a sense of security" is a device that has the function of visually presenting messages and information to alleviate anxiety in the user.

[0405] This invention relates to an information processing system for preventing misuse. The invention mainly consists of a server and a user terminal, and performs a series of processes to analyze voice data, identify emotions to evaluate the possibility of misuse, and notify the user.

[0406] The server first acquires audio data transmitted from the user's terminal in real time. Mobile devices such as smartphones and tablets can be used for this process. The acquired audio data is then noise-reduced using FFT-based noise reduction technology. This audio data is then converted into text using the Google Cloud Speech-to-Text API.

[0407] Next, the server analyzes the converted text information using natural language processing techniques such as Hugging Face's Transformers library to identify the user's emotions. Based on the results of this emotion analysis, the server assesses the likelihood of fraudulent activity. This assessment includes comparing and matching the data against pre-registered, known fraudulent patterns.

[0408] If the server determines that there is a high probability of fraud, it will send a warning to the user's terminal via synthesized voice. The user will be notified of a message such as, "This transaction is being monitored. Do you really want to proceed?", which is generated using speech synthesis technology.

[0409] The user's terminal simultaneously displays visually reassuring information to reduce the likelihood of the user becoming involved in fraudulent activity. Furthermore, if fraudulent activity is suspected, the transaction is temporarily suspended, giving the user time to review the situation.

[0410] For example, when making a high-value online purchase, the system can issue a warning if the user is likely to be redirected to a fraudulent website, allowing them to reconfirm the transaction and prevent fraud.

[0411] An example of a prompt message is: "Create a program that analyzes the user's voice in real time, detects changes in emotion, and outputs a warning message in real time if there is a possibility of fraud."

[0412] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0413] Step 1:

[0414] The device acquires the user's voice data. This voice data is an analog signal and is collected in real time through the device's microphone. The input is the user's voice signal, and the output is digital voice data sent to the server. Noise reduction technology is used to reduce background noise and ensure clear voice acquisition.

[0415] Step 2:

[0416] The server converts the acquired audio data into text using the Google Cloud Speech-to-Text API. The input is digital audio data, and the output is text data generated as a result of analysis by the API. This conversion process makes it possible to accurately extract the user's speech from the audio.

[0417] Step 3:

[0418] The server performs emotion recognition processing based on the converted text data. Here, natural language processing technologies such as Hugging Face's Transformers are used to analyze and identify the user's emotional state (e.g., anxiety, confusion). The input is text data, and the output is emotion labels and emotion scores. This allows for an accurate understanding of the user's psychological state.

[0419] Step 4:

[0420] The server assesses the likelihood of fraud by comparing the user's emotional state with known fraud patterns. Inputs include emotional labels, emotional scores, and past fraud pattern data. Output is a risk score, which quantifies the likelihood of fraud. If this score exceeds a certain threshold, the server proceeds to the next step.

[0421] Step 5:

[0422] If the server determines that there is a high probability of fraudulent activity, it will use speech synthesis technology to create a warning message. For example, it might generate a warning such as "This transaction is being monitored." The input is a predefined message template and a fraud score, and the output is a synthesized voice file. This allows the user to recognize potential fraud.

[0423] Step 6:

[0424] The terminal receives warning messages from the server and notifies the user visually and audibly. Input is synthesized speech files and text messages, and output is the information presented through the terminal's speaker or display. This draws the user's attention to fraudulent activity and prompts them to review or cancel transactions as needed.

[0425] 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.

[0426] 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.

[0427] 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.

[0428] [Third Embodiment]

[0429] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0430] 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.

[0431] 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).

[0432] 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.

[0433] 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.

[0434] 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).

[0435] 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.

[0436] 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.

[0437] 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.

[0438] 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.

[0439] 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.

[0440] 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".

[0441] This invention provides an AI agent system for effectively combating fraudulent phone calls. This system is server-centric and operates in conjunction with the user's terminal. The program's processing and its application are described in detail below.

[0442] The server acquires audio data of calls initiated from the user's terminal in real time. This audio data is converted into a clean signal using noise reduction technology and then converted into text data using speech recognition technology. This text data is analyzed using natural language processing (NLP) to identify the speaker's emotions. The emotion identification algorithm particularly identifies emotions such as confusion and anxiety, and this information is used to assess the risk of fraud.

[0443] When the fraud risk exceeds a certain threshold, the server dynamically adjusts the conversation using voice emulation. This voice emulation generates synthesized speech and sends messages to deter fraudsters. Specifically, it generates messages such as, "This call is being monitored by security." During this process, the user's device also displays reassuring messages notifying them that the caller may be fraudulent and to alleviate their anxiety.

[0444] For example, if a user receives a phone call that may be fraudulent, the server automatically executes the process described above and intervenes with voice emulation to thwart the fraudster's intentions. As a result, it is possible to prevent fraud and reduce the user's psychological burden.

[0445] In this way, the system proposed in the patent provides an effective defense against telephone fraud by integrating speech recognition, natural language processing, and speech synthesis technologies. This enhances user trust and contributes to improved social safety.

[0446] The following describes the processing flow.

[0447] Step 1:

[0448] The server acquires audio data for calls initiated on the user's terminal. The audio signal is sent to the server in real time and prepared for data processing.

[0449] Step 2:

[0450] The server performs noise reduction on the acquired audio data to clarify the audio signal. This generates clean audio data that improves the accuracy of speech recognition.

[0451] Step 3:

[0452] The server converts clean audio data into text data using speech recognition technology. This process extracts the content of the audio as a string of characters, preparing it for subsequent processing.

[0453] Step 4:

[0454] The server analyzes the generated text data using natural language processing (NLP) techniques to identify the speaker's emotions. In particular, it scrutinizes feelings of confusion and anxiety, and measures their intensity if present.

[0455] Step 5:

[0456] The server assesses fraud risk based on the analyzed sentiment data. It calculates a risk score, which also involves matching the data against existing fraud patterns to detect fraudulent phrases and unusual patterns.

[0457] Step 6:

[0458] If the server determines that the fraud risk has exceeded a set threshold, it dynamically alters the call content using AI-powered voice emulation. At this time, it sends a counter-message to the fraudster in synthesized speech.

[0459] Step 7:

[0460] The server displays a reassuring message on the user's device to alleviate their concerns and notify them of the potential for fraud. This allows the user to take appropriate action quickly.

[0461] Step 8:

[0462] The server ultimately logs the call and stores data for analyzing fraud risk. This allows for later review of the details and, if necessary, assist in legal proceedings.

[0463] (Example 1)

[0464] 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."

[0465] In modern society, fraudulent activities conducted via communication networks are on the rise, with phone-based fraud being a particular social problem. These scams are becoming more sophisticated, making it difficult for ordinary users to detect them on their own. Furthermore, fraudulent activities induce psychological confusion and anxiety in users, making it difficult for them to take prompt and appropriate action. Therefore, there is a need for a system that can detect potential fraud on behalf of users and respond quickly.

[0466] 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.

[0467] In this invention, the server includes means for acquiring voice information during a call, means for removing noise from the acquired voice information and converting it into text information using speech recognition technology, means for analyzing the text information to identify emotions, means for modifying the conversation content using speech synthesis when the risk of fraud exceeds a certain threshold, and means for displaying information to notify the user of the suspicion of fraud and alleviate their anxiety. This makes it possible to detect the risk of fraud in real time during a call and ensure the safety of the user.

[0468] A "call" is the act of transmitting voice information bidirectionally over a communication network, and is a means of communication through voice.

[0469] "Auditory information" refers to data that is recorded or transmitted as sound, including human speech and other sound signals.

[0470] "Noise reduction" is a process that makes audio clearer by removing unwanted background noise and interference sounds from audio data.

[0471] "Speech recognition technology" is a technology that converts speech information into a digital format, analyzes the content of the speech, and converts it into text information.

[0472] "Textual information" refers to string data extracted through speech recognition, and includes information expressed in human language.

[0473] "Emotion identification" is the process of analyzing textual information and audio data to identify the speaker's emotional state (for example, confusion, anxiety, joy, etc.).

[0474] "Fraud risk" is an indicator that numerically or qualitatively assesses the likelihood that the person on the other end of a call is likely to commit fraud.

[0475] A "threshold" is a boundary value that is set as a reference point, and when that value is exceeded, a specific action is performed.

[0476] "Speech synthesis" is a technology that converts text data into speech and generates new speech data.

[0477] "Changing the content of a conversation" refers to the act of manipulating or editing voice messages sent during a call to convey intended information.

[0478] A "user" is a person or group that makes a call using this system and is protected from fraud.

[0479] "Information that alleviates anxiety" refers to messages and notifications presented in a way that provides users with a sense of psychological reassurance.

[0480] This invention is a system for detecting call fraud via a communication network and protecting users. This system operates in cooperation with a server, a user's terminal, and the network.

[0481] The server acquires audio information from calls transmitted from the user's terminal in real time. The audio information is streamed to the server via a digital processing unit. This system uses speech recognition technology to convert the audio information into text. Specifically, it uses a processing library to remove noise and the Google Cloud Speech-to-Text API to transcribe the text.

[0482] Text information undergoes sentiment analysis on the server using natural language processing (NLP) techniques. The system utilizes Python's natural language processing libraries, NLTK and spaCy, to identify the speaker's emotions. This system specifically assesses fraud risk by detecting emotions such as confusion and anxiety. If the fraud risk assessment exceeds a threshold, speech synthesis technology is used to generate a deterrent warning message for fraudsters. Specifically, speech synthesis software is used to broadcast the message, "This call is being monitored by security."

[0483] The server also notifies the user's device of the suspected fraud and displays a reassuring message on the screen. For example, it might say, "This may be a suspicious call. Please rest assured, we are handling it."

[0484] One concrete example of this system's use is that when a user receives a suspicious phone call, they can input a prompt message into the AI ​​model such as, "I have been notified that there has been an unauthorized payment on my credit card. Please analyze the sentiment and assess the risk of fraud."

[0485] Without requiring any special action from the user, the system automatically detects potential fraud and takes appropriate action, preventing fraud and allowing users to continue their calls with peace of mind. This contributes to improving user safety and social trust.

[0486] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0487] Step 1:

[0488] The server acquires the audio information of the call transmitted from the user's terminal. This audio information is streamed via the VoIP protocol and input to the server in digital format. As output, the server prepares the audio data to pass to the next processing step while maintaining real-time performance.

[0489] Step 2:

[0490] The server processes the acquired audio information to remove noise. Specifically, it uses an audio processing library to apply filters that reduce background noise, processing the audio to make it clearer. The input is audio data containing noise, and the output is clean audio data with the noise minimized.

[0491] Step 3:

[0492] The server converts clean audio data into text information using speech recognition technology. Specifically, it uses speech recognition software to convert audio data into text format. The input is processed audio data, and the output is the corresponding text data. This conversion enables subsequent natural language processing.

[0493] Step 4:

[0494] The server applies natural language processing (NLP) techniques to the generated text information to analyze emotions. It uses a Python NLP library to analyze the text, focusing on specific emotions, particularly confusion and anxiety. The input is text data, and the output is the identification and intensity of the detected emotions.

[0495] Step 5:

[0496] The server assesses fraud risk using the results of sentiment analysis. A fraud risk assessment algorithm is used to quantify the risk level based on the identified emotions. The input is identified sentiment information, and the output is a numerical evaluation of fraud risk.

[0497] Step 6:

[0498] If the fraud risk assessment exceeds a certain threshold, the server uses speech synthesis technology to generate a warning message and insert it into the call. Specifically, it uses speech synthesis software to synthesize a message such as "This call is being monitored by security" to deter fraudsters. The input is the result of the fraud risk assessment, and the output is the generated warning message.

[0499] Step 7:

[0500] The server notifies the user's device of a suspected scam and displays a reassuring message on the screen. This notification allows the user to understand the potential for fraud and take prompt action. The input is the result of the fraud risk assessment, and the output is the reassuring message displayed on the device.

[0501] (Application Example 1)

[0502] 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."

[0503] In recent years, the methods used in telephone fraud have become more sophisticated, making it difficult for individual users to deal with them. Therefore, there is a need for effective means to prevent damage from telephone fraud. This invention aims to provide a new defense system that can quickly and accurately assess the possibility of telephone fraud and protect users.

[0504] 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.

[0505] In this invention, the server includes means for acquiring voice information in a call, means for analyzing the acquired voice information to identify emotions, means for evaluating the likelihood of fraud based on the identified emotions, means for modifying the conversation content using voice reconstruction if the likelihood of fraud exceeds a certain standard, means for displaying information to notify the user of the suspicion of fraud and alleviate anxiety, and means for sending a dynamic warning message to the fraudulent caller. This makes it possible to effectively deter special fraud calls and provide peace of mind to the user.

[0506] "Voice information in a call" refers to the entirety of voice data transmitted via communication means.

[0507] "Means of acquisition" refers to the technology for receiving and storing audio information in real time.

[0508] "Methods for analyzing and identifying emotions" refers to algorithms that process audio data to identify the speaker's psychological state.

[0509] "Methods for assessing the likelihood of fraud" refer to methods for quantifying the risk of fraudulent activity based on analyzed sentiment information.

[0510] "Methods for altering conversation content using voice reproduction" refers to technology that generates synthesized speech and automatically provides a new message to the other party on the call.

[0511] "A means of notifying users of suspected fraud and displaying information to alleviate their anxiety" refers to a method of providing reassurance by displaying a warning message on the user's device when fraud is suspected.

[0512] "Means of sending dynamic warning messages to fraudulent communicators" refers to technology that, when fraudulent activity is detected, sends a message in real time to warn those attempting to commit fraud.

[0513] To implement this invention, it is necessary to construct a system that acquires and analyzes voice information during a call in real time. This system consists of a user terminal including a microphone and communication devices for acquiring voice information, and a server for analyzing the voice data and evaluating the risk of fraud.

[0514] The server uses speech recognition software and natural language processing (NLP) algorithms to convert speech information into text and analyze its sentiment. This analysis assesses the likelihood of fraud, and if it exceeds a certain threshold, a warning message using voice reproduction is generated for the fraudster. This warning message is sent to the fraudster in real time using synthesized speech technology.

[0515] The user's device immediately receives a notification if fraud is suspected, and reassuring information is displayed. This information is displayed dynamically based on voice analysis results, so the user can always stay informed of the latest situation. Furthermore, dynamic warning messages are sent directly to fraudsters, acting to deter fraudulent activity.

[0516] As a concrete example, when a user receives a call from an unknown number, the server analyzes the conversation and determines that it may be a scam. As a result, the server issues a synthesized voice warning that "this call is being monitored," and simultaneously displays a notification to the user stating "this may be a scam." This kind of action allows users to prevent themselves from becoming victims of fraud.

[0517] An example of a prompt message for a generative AI model is: "Audio data has been input. Perform speech recognition and convert it to text data. Then, use sentiment analysis to assess the fraud risk, and if the risk is high, generate a warning message and notify the user." This prompt allows the server to perform appropriate analysis and evaluation.

[0518] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0519] Step 1:

[0520] The server receives in-call audio information transmitted from the user terminal in real time. The input audio information is transmitted in digital format and processed by the server's speech recognition software. At this stage, noise reduction is performed on the audio signal to obtain clear audio data.

[0521] Step 2:

[0522] The server converts the received clear audio data into text information using speech recognition technology. By converting the audio data to text, natural language processing algorithms become applicable. In this process, the entire content of the audio is recorded as text and used as input data for the next analysis step.

[0523] Step 3:

[0524] The server analyzes the transcribed audio information using natural language processing (NLP) algorithms to identify the speaker's emotions. Sentiment analysis evaluates keywords and context within the text, particularly extracting feelings of confusion and anxiety associated with fraud. This outputs the emotional data necessary for assessing fraud risk.

[0525] Step 4:

[0526] The server assesses the likelihood of fraud based on emotions. The assessment algorithm compares the sentiment analysis results with known fraud patterns to quantify the fraud risk. This output is used to determine whether the fraud risk exceeds a set threshold.

[0527] Step 5:

[0528] The server uses voice synthesis technology to generate and send a warning message to the fraudulent caller when the fraud risk exceeds a certain threshold. A synthesized voice engine creates messages such as "This call is being monitored" and transmits them to the fraudulent caller in real time.

[0529] Step 6:

[0530] The device displays a notification to the user regarding identified suspected fraud. This notification is displayed on the device screen as a visually reassuring message. By reviewing the information on the screen, the user can prepare for the possibility of a fraudulent call.

[0531] 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.

[0532] This invention is an AI agent system for preventing fraudulent phone calls, integrating voice data acquisition, emotion recognition, fraud risk assessment, voice emulation, user notification, and user emotion analysis using an emotion engine. The system consists of a server, a terminal, and a user interface.

[0533] The server first acquires audio data of calls made on the user's terminal in real time. This audio data is then subjected to noise reduction processing and converted into text data using speech recognition technology. This converted text data is then analyzed using natural language processing (NLP) technology to identify the speaker's emotions, particularly confusion or anxiety.

[0534] In addition, the emotion engine analyzes the user's voice and speech patterns, and also evaluates the user's emotional state. By identifying the user's stress and tension, it more accurately assesses the likelihood of fraud based on fraud risk and emotional state. If the fraud risk exceeds a certain threshold, the server dynamically changes the content of the call using AI-powered voice emulation. This sends a message that will cause scammers to hesitate.

[0535] As a concrete example, when a user receives a potentially fraudulent phone call, the server uses an emotion engine to detect signs of the user's anxiety and thoroughly reassess the fraud risk. If it determines that the fraud risk is high, it then uses voice emulation to broadcast a synthesized voice message such as "This call is being monitored" to psychologically unsettle the fraudster. Furthermore, a reassuring message is displayed on the user's device to alleviate their anxiety.

[0536] Thus, the present invention can prevent fraudulent activities and enhance psychological protection for users by recognizing the emotional state of the user and implementing dynamic fraud prevention measures based on that state.

[0537] The following describes the processing flow.

[0538] Step 1:

[0539] The server receives a call initiation notification from the user's device and begins acquiring audio data. It receives the audio stream in real time and prepares it for data processing.

[0540] Step 2:

[0541] The server performs noise reduction on the acquired audio data to generate a clear audio signal. This prepares the system for improving the accuracy of speech recognition.

[0542] Step 3:

[0543] The server converts clear audio data into text data using speech recognition technology. The converted text is then used for subsequent analysis.

[0544] Step 4:

[0545] The server analyzes text data using natural language processing (NLP) techniques to identify the speaker's emotions. In particular, it focuses on analyzing emotions such as confusion and anxiety.

[0546] Step 5:

[0547] The server uses an emotion engine to analyze the user's voice tone and speech patterns to recognize the user's emotional state. This allows it to determine whether the user is experiencing stress or tension.

[0548] Step 6:

[0549] The server assesses the fraud risk based on the analysis results. It compares the results against known fraudulent phrases and patterns to calculate a risk score. It also considers the user's emotional state to improve the accuracy of the assessment.

[0550] Step 7:

[0551] If the fraud risk exceeds a set threshold, the server modifies the conversation using AI-powered voice emulation. Specifically, it generates synthesized messages such as "This call is being monitored by security" to psychologically unsettle the fraudster.

[0552] Step 8:

[0553] The server notifies the user's device of the suspected fraud and displays a reassuring message. This allows the user to take appropriate action quickly.

[0554] Step 9:

[0555] The server logs call data and analysis results, preparing for future analysis and provision to the police. The data is stored securely with user privacy in mind.

[0556] (Example 2)

[0557] 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."

[0558] In modern times, fraudulent activities such as special fraud and wire transfer fraud are on the rise. These scams are often conducted via voice, and victims are frequently deceived and lose money. With current technology, it is difficult to detect the risk of fraud in real time during a call and take appropriate action. Furthermore, it is not easy for victims themselves to recognize the risk of fraud, which can cause psychological distress. Therefore, there is a need to prevent fraudulent activities and provide users with a sense of security.

[0559] 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.

[0560] In this invention, the server includes means for acquiring voice information in a call, means for applying noise reduction processing to the acquired voice information, and means for converting the noise-reduced voice information into text information using speech recognition technology. This enables real-time assessment of fraud risk and, if necessary, generates messages using synthesized speech to deter fraudulent activity, thereby providing psychological protection to the user.

[0561] "Audio information" refers to data based on the wavelengths of sounds contained during a call, and is used to represent the content of the conversation.

[0562] "Noise reduction processing" is a technique used to reduce or remove unnecessary background noise and interfering sounds from audio information in order to obtain a clear audio signal.

[0563] "Speech recognition technology" is a technology that analyzes speech information and converts its linguistic content into text format.

[0564] "Text information" refers to string data in which speech information converted by speech recognition technology is linguistically represented.

[0565] "Emotional analysis" is an analytical technique used to identify a speaker's emotional state from text or audio information.

[0566] "Vocal characteristics" refer to information that describes the tone, speed, and pitch of a speaker's voice, and are used to analyze their emotional state and intentions.

[0567] "Fraud risk" is an indicator that shows the likelihood of fraudulent activity based on the content of the call and the emotional state of the speaker.

[0568] "Synthesized speech" refers to artificially generated voice data, which is an acoustic signal used to convey a specific message.

[0569] "Psychological security" refers to the sense of safety and peace of mind provided to system users, and a state in which they feel protected from fraudulent activities.

[0570] This invention relates to a system for preventing special fraud, and includes a series of processes for acquiring and analyzing voice information and assessing fraud risk. It consists of three main elements: a server, a terminal, and a user.

[0571] Server Functions

[0572] The server focuses on acquiring, analyzing, and evaluating audio information. The server receives audio information transmitted in real time from terminals. Digital signal processing technology is used for audio processing, and a dedicated algorithm is implemented to remove noise from the audio information. For speech recognition technology, a cloud-based speech recognition service is used to convert speech into text. Specifically, Google Cloud Speech-to-Text API or other commercial speech recognition engines can be used.

[0573] The server further analyzes text information based on sentiment analysis technology to identify emotions from the user's statements. This process utilizes natural language processing libraries, with NLTK and other machine learning models proving effective. Additionally, open-source speech characteristic analysis tools are used to analyze the user's voice characteristics, enabling the detection of user stress, tension, and other factors. In assessing fraud risk, these analysis results are integrated to determine the likelihood of fraudulent activity.

[0574] If the risk of fraud is high, the server uses AI speech synthesis technology to generate and transmit a synthesized voice message, such as "This call is being monitored." Commercial speech synthesis engines such as Amazon Polly can be used for this purpose. The generated voice message aims to have a psychological effect to deter fraudulent activity.

[0575] Device functions

[0576] The terminal functions as an interface with the user, collecting and transmitting voice information. Voice information acquired during a call is filtered for noise before being sent to the server. It also plays a role in alleviating anxiety by displaying reassuring messages received from the server to the user.

[0577] User roles

[0578] Users are users of the system and are monitored by the system during calls. To minimize the burden on users while providing protection from fraudulent activities, the system is designed to make them feel safe through reassuring messages.

[0579] Specific example

[0580] For example, if a user receives a suspicious call, the server immediately analyzes the audio and text to detect signs of fraud. If it is deemed high-risk, a synthesized voice message such as "This call is being monitored" is sent to the caller. In addition, the user's device displays "You are currently on a secure call" to provide reassurance.

[0581] Examples of prompts for generative AI models

[0582] "What is needed to develop a prototype of a call content analysis system for evaluating the risk of fraudulent phone calls?"

[0583] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0584] Step 1: Obtain audio information

[0585] The device detects the start of a call and captures the call's audio information in real time using its internal microphone. The captured audio information is then sent to the server as a digital signal. The input is raw audio, and the output is digitized audio data.

[0586] Step 2: Noise Reduction and Speech Recognition

[0587] The server performs noise reduction processing on the received audio data. This reduces background noise and improves the accuracy of speech recognition. Next, the noise-reduced audio data is converted into text information using speech recognition technology. Specifically, a speech recognition API is used to convert sound waves into character data. The input is noise-reduced audio data, and the output is text data.

[0588] Step 3: Emotion Analysis

[0589] The server analyzes the generated text information to identify the speaker's emotions. Using natural language processing techniques, it calculates emotional indicators from words and phrases in the text. A sentiment analysis library is used for this analysis. The input is text data, and the output is an emotional score and an inferred emotional state.

[0590] Step 4: User Voice Characteristics Assessment

[0591] The server uses a voice analysis tool to evaluate the user's voice characteristics. It analyzes voice tone, tempo, pitch, etc., to identify signs of stress or tension. The input is voice data, and the output is the evaluation result regarding the user's voice characteristics.

[0592] Step 5: Assessing the risk of fraud

[0593] The server integrates the sentiment analysis results and voice characteristic evaluation results to assess fraud risk. This assessment uses an algorithm that compares the results with fraud patterns to calculate a risk score. The inputs are the sentiment score and voice characteristic evaluation results, and the output is the fraud risk score.

[0594] Step 6: Intervention using voice emulation

[0595] If the server determines that there is a high risk of fraud, it uses AI speech synthesis technology to generate a message such as "This call is being monitored" and plays the synthesized voice to the caller. The input is the fraud risk score, and the output is synthesized voice data.

[0596] Step 7: Displaying user notifications and safety messages

[0597] The terminal, based on information sent from the server, displays a reassuring message to the user on the screen, such as "You are currently on a secure call," providing a sense of psychological security. The input is instruction data from the server, and the output is the reassuring message displayed on the screen.

[0598] (Application Example 2)

[0599] 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."

[0600] Fraudulent use of electronic payments poses a significant threat to users. Preventing losses due to fraudulent activity and ensuring user confidence are crucial. However, current technology lacks the means to rapidly assess the possibility of fraud in real time and respond appropriately to users. As a result, users are not fully protected from fraudulent activity.

[0601] 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.

[0602] In this invention, the server includes a processing unit for acquiring voice data, a processing unit for analyzing the acquired voice data and identifying emotions, and a processing unit for evaluating the possibility of fraudulent activity based on the identified emotions. This makes it possible to detect changes in the user's emotions in real time and respond quickly if there are signs of fraudulent activity.

[0603] A "processing device for acquiring voice data" is a device that has the function of collecting voice information from phone calls and conversations and converting it into a digital data format that can be used within the system.

[0604] A "processing device for identifying emotions" is a device that has the function of identifying the emotional state of a speaker based on collected audio data and analyzing it as data.

[0605] A "processing device for evaluating the likelihood of fraudulent activity" is an electronic device that evaluates the likelihood of fraud in transactions or communications based on identified emotions and other relevant data, and makes decisions on actions based on that evaluation.

[0606] A "processing device that adjusts dialogue content using speech synthesis technology" is a device that operates technology used to generate voice data and send warnings and notifications to the user.

[0607] A "device that displays information to provide a sense of security" is a device that has the function of visually presenting messages and information to alleviate anxiety in the user.

[0608] This invention relates to an information processing system for preventing misuse. The invention mainly consists of a server and a user terminal, and performs a series of processes to analyze voice data, identify emotions to evaluate the possibility of misuse, and notify the user.

[0609] The server first acquires audio data transmitted from the user's terminal in real time. Mobile devices such as smartphones and tablets can be used for this process. The acquired audio data is then noise-reduced using FFT-based noise reduction technology. This audio data is then converted into text using the Google Cloud Speech-to-Text API.

[0610] Next, the server analyzes the converted text information using natural language processing techniques such as Hugging Face's Transformers library to identify the user's emotions. Based on the results of this emotion analysis, the server assesses the likelihood of fraudulent activity. This assessment includes comparing and matching the data against pre-registered, known fraudulent patterns.

[0611] If the server determines that there is a high probability of fraud, it will send a warning to the user's terminal via synthesized voice. The user will be notified of a message such as, "This transaction is being monitored. Do you really want to proceed?", which is generated using speech synthesis technology.

[0612] The user's terminal simultaneously displays visually reassuring information to reduce the likelihood of the user becoming involved in fraudulent activity. Furthermore, if fraudulent activity is suspected, the transaction is temporarily suspended, giving the user time to review the situation.

[0613] For example, when making a high-value online purchase, the system can issue a warning if the user is likely to be redirected to a fraudulent website, allowing them to reconfirm the transaction and prevent fraud.

[0614] An example of a prompt message is: "Create a program that analyzes the user's voice in real time, detects changes in emotion, and outputs a warning message in real time if there is a possibility of fraud."

[0615] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0616] Step 1:

[0617] The device acquires the user's voice data. This voice data is an analog signal and is collected in real time through the device's microphone. The input is the user's voice signal, and the output is digital voice data sent to the server. Noise reduction technology is used to reduce background noise and ensure clear voice acquisition.

[0618] Step 2:

[0619] The server converts the acquired audio data into text using the Google Cloud Speech-to-Text API. The input is digital audio data, and the output is text data generated as a result of analysis by the API. This conversion process makes it possible to accurately extract the user's speech from the audio.

[0620] Step 3:

[0621] The server performs emotion recognition processing based on the converted text data. Here, natural language processing technologies such as Hugging Face's Transformers are used to analyze and identify the user's emotional state (e.g., anxiety, confusion). The input is text data, and the output is emotion labels and emotion scores. This allows for an accurate understanding of the user's psychological state.

[0622] Step 4:

[0623] The server assesses the likelihood of fraud by comparing the user's emotional state with known fraud patterns. Inputs include emotional labels, emotional scores, and past fraud pattern data. Output is a risk score, which quantifies the likelihood of fraud. If this score exceeds a certain threshold, the server proceeds to the next step.

[0624] Step 5:

[0625] If the server determines that there is a high probability of fraudulent activity, it will use speech synthesis technology to create a warning message. For example, it might generate a warning such as "This transaction is being monitored." The input is a predefined message template and a fraud score, and the output is a synthesized voice file. This allows the user to recognize potential fraud.

[0626] Step 6:

[0627] The terminal receives warning messages from the server and notifies the user visually and audibly. Input is synthesized speech files and text messages, and output is the information presented through the terminal's speaker or display. This draws the user's attention to fraudulent activity and prompts them to review or cancel transactions as needed.

[0628] 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.

[0629] 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.

[0630] 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.

[0631] [Fourth Embodiment]

[0632] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[0633] 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.

[0634] 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).

[0635] 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.

[0636] 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.

[0637] 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).

[0638] 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.

[0639] 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.

[0640] 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.

[0641] 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.

[0642] 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.

[0643] 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.

[0644] 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".

[0645] This invention provides an AI agent system for effectively combating fraudulent phone calls. This system is server-centric and operates in conjunction with the user's terminal. The program's processing and its application are described in detail below.

[0646] The server acquires audio data of calls initiated from the user's terminal in real time. This audio data is converted into a clean signal using noise reduction technology and then converted into text data using speech recognition technology. This text data is analyzed using natural language processing (NLP) to identify the speaker's emotions. The emotion identification algorithm particularly identifies emotions such as confusion and anxiety, and this information is used to assess the risk of fraud.

[0647] When the fraud risk exceeds a certain threshold, the server dynamically adjusts the conversation using voice emulation. This voice emulation generates synthesized speech and sends messages to deter fraudsters. Specifically, it generates messages such as, "This call is being monitored by security." During this process, the user's device also displays reassuring messages notifying them that the caller may be fraudulent and to alleviate their anxiety.

[0648] For example, if a user receives a phone call that may be fraudulent, the server automatically executes the process described above and intervenes with voice emulation to thwart the fraudster's intentions. As a result, it is possible to prevent fraud and reduce the user's psychological burden.

[0649] In this way, the system proposed in the patent provides an effective defense against telephone fraud by integrating speech recognition, natural language processing, and speech synthesis technologies. This enhances user trust and contributes to improved social safety.

[0650] The following describes the processing flow.

[0651] Step 1:

[0652] The server acquires audio data for calls initiated on the user's terminal. The audio signal is sent to the server in real time and prepared for data processing.

[0653] Step 2:

[0654] The server performs noise reduction on the acquired audio data to clarify the audio signal. This generates clean audio data that improves the accuracy of speech recognition.

[0655] Step 3:

[0656] The server converts clean audio data into text data using speech recognition technology. This process extracts the content of the audio as a string of characters, preparing it for subsequent processing.

[0657] Step 4:

[0658] The server analyzes the generated text data using natural language processing (NLP) techniques to identify the speaker's emotions. In particular, it scrutinizes feelings of confusion and anxiety, and measures their intensity if present.

[0659] Step 5:

[0660] The server assesses fraud risk based on the analyzed sentiment data. It calculates a risk score, which also involves matching the data against existing fraud patterns to detect fraudulent phrases and unusual patterns.

[0661] Step 6:

[0662] If the server determines that the fraud risk has exceeded a set threshold, it dynamically alters the call content using AI-powered voice emulation. At this time, it sends a counter-message to the fraudster in synthesized speech.

[0663] Step 7:

[0664] The server displays a reassuring message on the user's device to alleviate their concerns and notify them of the potential for fraud. This allows the user to take appropriate action quickly.

[0665] Step 8:

[0666] The server ultimately logs the call and stores data for analyzing fraud risk. This allows for later review of the details and, if necessary, assist in legal proceedings.

[0667] (Example 1)

[0668] 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".

[0669] In modern society, fraudulent activities conducted via communication networks are on the rise, with phone-based fraud being a particular social problem. These scams are becoming more sophisticated, making it difficult for ordinary users to detect them on their own. Furthermore, fraudulent activities induce psychological confusion and anxiety in users, making it difficult for them to take prompt and appropriate action. Therefore, there is a need for a system that can detect potential fraud on behalf of users and respond quickly.

[0670] 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.

[0671] In this invention, the server includes means for acquiring voice information during a call, means for removing noise from the acquired voice information and converting it into text information using speech recognition technology, means for analyzing the text information to identify emotions, means for modifying the conversation content using speech synthesis when the risk of fraud exceeds a certain threshold, and means for displaying information to notify the user of the suspicion of fraud and alleviate their anxiety. This makes it possible to detect the risk of fraud in real time during a call and ensure the safety of the user.

[0672] A "call" is the act of transmitting voice information bidirectionally over a communication network, and is a means of communication through voice.

[0673] "Auditory information" refers to data that is recorded or transmitted as sound, including human speech and other sound signals.

[0674] "Noise reduction" is a process that makes audio clearer by removing unwanted background noise and interference sounds from audio data.

[0675] "Speech recognition technology" is a technology that converts speech information into a digital format, analyzes the content of the speech, and converts it into text information.

[0676] "Textual information" refers to string data extracted through speech recognition, and includes information expressed in human language.

[0677] "Emotion identification" is the process of analyzing textual information and audio data to identify the speaker's emotional state (for example, confusion, anxiety, joy, etc.).

[0678] "Fraud risk" is an indicator that numerically or qualitatively assesses the likelihood that the person on the other end of a call is likely to commit fraud.

[0679] A "threshold" is a boundary value that is set as a reference point, and when that value is exceeded, a specific action is performed.

[0680] "Speech synthesis" is a technology that converts text data into speech and generates new speech data.

[0681] "Changing the content of a conversation" refers to the act of manipulating or editing voice messages sent during a call to convey intended information.

[0682] A "user" is a person or group that makes a call using this system and is protected from fraud.

[0683] "Information that alleviates anxiety" refers to messages and notifications presented in a way that provides users with a sense of psychological reassurance.

[0684] This invention is a system for detecting call fraud via a communication network and protecting users. This system operates in cooperation with a server, a user's terminal, and the network.

[0685] The server acquires audio information from calls transmitted from the user's terminal in real time. The audio information is streamed to the server via a digital processing unit. This system uses speech recognition technology to convert the audio information into text. Specifically, it uses a processing library to remove noise and the Google Cloud Speech-to-Text API to transcribe the text.

[0686] Text information undergoes sentiment analysis on the server using natural language processing (NLP) techniques. The system utilizes Python's natural language processing libraries, NLTK and spaCy, to identify the speaker's emotions. This system specifically assesses fraud risk by detecting emotions such as confusion and anxiety. If the fraud risk assessment exceeds a threshold, speech synthesis technology is used to generate a deterrent warning message for fraudsters. Specifically, speech synthesis software is used to broadcast the message, "This call is being monitored by security."

[0687] The server also notifies the user's device of the suspected fraud and displays a reassuring message on the screen. For example, it might say, "This may be a suspicious call. Please rest assured, we are handling it."

[0688] One concrete example of this system's use is that when a user receives a suspicious phone call, they can input a prompt message into the AI ​​model such as, "I have been notified that there has been an unauthorized payment on my credit card. Please analyze the sentiment and assess the risk of fraud."

[0689] Without requiring any special action from the user, the system automatically detects potential fraud and takes appropriate action, preventing fraud and allowing users to continue their calls with peace of mind. This contributes to improving user safety and social trust.

[0690] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0691] Step 1:

[0692] The server acquires the audio information of the call transmitted from the user's terminal. This audio information is streamed via the VoIP protocol and input to the server in digital format. As output, the server prepares the audio data to pass to the next processing step while maintaining real-time performance.

[0693] Step 2:

[0694] The server processes the acquired audio information to remove noise. Specifically, it uses an audio processing library to apply filters that reduce background noise, processing the audio to make it clearer. The input is audio data containing noise, and the output is clean audio data with the noise minimized.

[0695] Step 3:

[0696] The server converts clean audio data into text information using speech recognition technology. Specifically, it uses speech recognition software to convert audio data into text format. The input is processed audio data, and the output is the corresponding text data. This conversion enables subsequent natural language processing.

[0697] Step 4:

[0698] The server applies natural language processing (NLP) techniques to the generated text information to analyze emotions. It uses a Python NLP library to analyze the text, focusing on specific emotions, particularly confusion and anxiety. The input is text data, and the output is the identification and intensity of the detected emotions.

[0699] Step 5:

[0700] The server assesses fraud risk using the results of sentiment analysis. A fraud risk assessment algorithm is used to quantify the risk level based on the identified emotions. The input is identified sentiment information, and the output is a numerical evaluation of fraud risk.

[0701] Step 6:

[0702] If the fraud risk assessment exceeds a certain threshold, the server uses speech synthesis technology to generate a warning message and insert it into the call. Specifically, it uses speech synthesis software to synthesize a message such as "This call is being monitored by security" to deter fraudsters. The input is the result of the fraud risk assessment, and the output is the generated warning message.

[0703] Step 7:

[0704] The server notifies the user's device of a suspected scam and displays a reassuring message on the screen. This notification allows the user to understand the potential for fraud and take prompt action. The input is the result of the fraud risk assessment, and the output is the reassuring message displayed on the device.

[0705] (Application Example 1)

[0706] 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".

[0707] In recent years, the methods used in telephone fraud have become more sophisticated, making it difficult for individual users to deal with them. Therefore, there is a need for effective means to prevent damage from telephone fraud. This invention aims to provide a new defense system that can quickly and accurately assess the possibility of telephone fraud and protect users.

[0708] 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.

[0709] In this invention, the server includes means for acquiring voice information in a call, means for analyzing the acquired voice information to identify emotions, means for evaluating the likelihood of fraud based on the identified emotions, means for modifying the conversation content using voice reconstruction if the likelihood of fraud exceeds a certain standard, means for displaying information to notify the user of the suspicion of fraud and alleviate anxiety, and means for sending a dynamic warning message to the fraudulent caller. This makes it possible to effectively deter special fraud calls and provide peace of mind to the user.

[0710] "Voice information in a call" refers to the entirety of voice data transmitted via communication means.

[0711] "Means of acquisition" refers to the technology for receiving and storing audio information in real time.

[0712] "Methods for analyzing and identifying emotions" refers to algorithms that process audio data to identify the speaker's psychological state.

[0713] "Methods for assessing the likelihood of fraud" refer to methods for quantifying the risk of fraudulent activity based on analyzed sentiment information.

[0714] "Methods for altering conversation content using voice reproduction" refers to technology that generates synthesized speech and automatically provides a new message to the other party on the call.

[0715] "A means of notifying users of suspected fraud and displaying information to alleviate their anxiety" refers to a method of providing reassurance by displaying a warning message on the user's device when fraud is suspected.

[0716] "Means of sending dynamic warning messages to fraudulent communicators" refers to technology that, when fraudulent activity is detected, sends a message in real time to warn those attempting to commit fraud.

[0717] To implement this invention, it is necessary to construct a system that acquires and analyzes voice information during a call in real time. This system consists of a user terminal including a microphone and communication devices for acquiring voice information, and a server for analyzing the voice data and evaluating the risk of fraud.

[0718] The server uses speech recognition software and natural language processing (NLP) algorithms to convert speech information into text and analyze its sentiment. This analysis assesses the likelihood of fraud, and if it exceeds a certain threshold, a warning message using voice reproduction is generated for the fraudster. This warning message is sent to the fraudster in real time using synthesized speech technology.

[0719] The user's device immediately receives a notification if fraud is suspected, and reassuring information is displayed. This information is displayed dynamically based on voice analysis results, so the user can always stay informed of the latest situation. Furthermore, dynamic warning messages are sent directly to fraudsters, acting to deter fraudulent activity.

[0720] As a concrete example, when a user receives a call from an unknown number, the server analyzes the conversation and determines that it may be a scam. As a result, the server issues a synthesized voice warning that "this call is being monitored," and simultaneously displays a notification to the user stating "this may be a scam." This kind of action allows users to prevent themselves from becoming victims of fraud.

[0721] An example of a prompt message for a generative AI model is: "Audio data has been input. Perform speech recognition and convert it to text data. Then, use sentiment analysis to assess the fraud risk, and if the risk is high, generate a warning message and notify the user." This prompt allows the server to perform appropriate analysis and evaluation.

[0722] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0723] Step 1:

[0724] The server receives in-call audio information transmitted from the user terminal in real time. The input audio information is transmitted in digital format and processed by the server's speech recognition software. At this stage, noise reduction is performed on the audio signal to obtain clear audio data.

[0725] Step 2:

[0726] The server converts the received clear audio data into text information using speech recognition technology. By converting the audio data to text, natural language processing algorithms become applicable. In this process, the entire content of the audio is recorded as text and used as input data for the next analysis step.

[0727] Step 3:

[0728] The server analyzes the transcribed audio information using natural language processing (NLP) algorithms to identify the speaker's emotions. Sentiment analysis evaluates keywords and context within the text, particularly extracting feelings of confusion and anxiety associated with fraud. This outputs the emotional data necessary for assessing fraud risk.

[0729] Step 4:

[0730] The server assesses the likelihood of fraud based on emotions. The assessment algorithm compares the sentiment analysis results with known fraud patterns to quantify the fraud risk. This output is used to determine whether the fraud risk exceeds a set threshold.

[0731] Step 5:

[0732] The server uses voice synthesis technology to generate and send a warning message to the fraudulent caller when the fraud risk exceeds a certain threshold. A synthesized voice engine creates messages such as "This call is being monitored" and transmits them to the fraudulent caller in real time.

[0733] Step 6:

[0734] The device displays a notification to the user regarding identified suspected fraud. This notification is displayed on the device screen as a visually reassuring message. By reviewing the information on the screen, the user can prepare for the possibility of a fraudulent call.

[0735] 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.

[0736] This invention is an AI agent system for preventing fraudulent phone calls, integrating voice data acquisition, emotion recognition, fraud risk assessment, voice emulation, user notification, and user emotion analysis using an emotion engine. The system consists of a server, a terminal, and a user interface.

[0737] The server first acquires audio data of calls made on the user's terminal in real time. This audio data is then subjected to noise reduction processing and converted into text data using speech recognition technology. This converted text data is then analyzed using natural language processing (NLP) technology to identify the speaker's emotions, particularly confusion or anxiety.

[0738] In addition, the emotion engine analyzes the user's voice and speech patterns, and also evaluates the user's emotional state. By identifying the user's stress and tension, it more accurately assesses the likelihood of fraud based on fraud risk and emotional state. If the fraud risk exceeds a certain threshold, the server dynamically changes the content of the call using AI-powered voice emulation. This sends a message that will cause scammers to hesitate.

[0739] As a concrete example, when a user receives a potentially fraudulent phone call, the server uses an emotion engine to detect signs of the user's anxiety and thoroughly reassess the fraud risk. If it determines that the fraud risk is high, it then uses voice emulation to broadcast a synthesized voice message such as "This call is being monitored" to psychologically unsettle the fraudster. Furthermore, a reassuring message is displayed on the user's device to alleviate their anxiety.

[0740] Thus, the present invention can prevent fraudulent activities and enhance psychological protection for users by recognizing the emotional state of the user and implementing dynamic fraud prevention measures based on that state.

[0741] The following describes the processing flow.

[0742] Step 1:

[0743] The server receives a call initiation notification from the user's device and begins acquiring audio data. It receives the audio stream in real time and prepares it for data processing.

[0744] Step 2:

[0745] The server performs noise reduction on the acquired audio data to generate a clear audio signal. This prepares the system for improving the accuracy of speech recognition.

[0746] Step 3:

[0747] The server converts clear audio data into text data using speech recognition technology. The converted text is then used for subsequent analysis.

[0748] Step 4:

[0749] The server analyzes text data using natural language processing (NLP) techniques to identify the speaker's emotions. In particular, it focuses on analyzing emotions such as confusion and anxiety.

[0750] Step 5:

[0751] The server uses an emotion engine to analyze the user's voice tone and speech patterns to recognize the user's emotional state. This allows it to determine whether the user is experiencing stress or tension.

[0752] Step 6:

[0753] The server assesses the fraud risk based on the analysis results. It compares the results against known fraudulent phrases and patterns to calculate a risk score. It also considers the user's emotional state to improve the accuracy of the assessment.

[0754] Step 7:

[0755] If the fraud risk exceeds a set threshold, the server modifies the conversation using AI-powered voice emulation. Specifically, it generates synthesized messages such as "This call is being monitored by security" to psychologically unsettle the fraudster.

[0756] Step 8:

[0757] The server notifies the user's device of the suspected fraud and displays a reassuring message. This allows the user to take appropriate action quickly.

[0758] Step 9:

[0759] The server logs call data and analysis results, preparing for future analysis and provision to the police. The data is stored securely with user privacy in mind.

[0760] (Example 2)

[0761] 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".

[0762] In modern times, fraudulent activities such as special fraud and wire transfer fraud are on the rise. These scams are often conducted via voice, and victims are frequently deceived and lose money. With current technology, it is difficult to detect the risk of fraud in real time during a call and take appropriate action. Furthermore, it is not easy for victims themselves to recognize the risk of fraud, which can cause psychological distress. Therefore, there is a need to prevent fraudulent activities and provide users with a sense of security.

[0763] 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.

[0764] In this invention, the server includes means for acquiring voice information in a call, means for applying noise reduction processing to the acquired voice information, and means for converting the noise-reduced voice information into text information using speech recognition technology. This enables real-time assessment of fraud risk and, if necessary, generates messages using synthesized speech to deter fraudulent activity, thereby providing psychological protection to the user.

[0765] "Audio information" refers to data based on the wavelengths of sounds contained during a call, and is used to represent the content of the conversation.

[0766] "Noise reduction processing" is a technique used to reduce or remove unnecessary background noise and interfering sounds from audio information in order to obtain a clear audio signal.

[0767] "Speech recognition technology" is a technology that analyzes speech information and converts its linguistic content into text format.

[0768] "Text information" refers to string data in which speech information converted by speech recognition technology is linguistically represented.

[0769] "Emotional analysis" is an analytical technique used to identify a speaker's emotional state from text or audio information.

[0770] "Vocal characteristics" refer to information that describes the tone, speed, and pitch of a speaker's voice, and are used to analyze their emotional state and intentions.

[0771] "Fraud risk" is an indicator that shows the likelihood of fraudulent activity based on the content of the call and the emotional state of the speaker.

[0772] "Synthesized speech" refers to artificially generated voice data, which is an acoustic signal used to convey a specific message.

[0773] "Psychological security" refers to the sense of safety and peace of mind provided to system users, and a state in which they feel protected from fraudulent activities.

[0774] This invention relates to a system for preventing special fraud, and includes a series of processes for acquiring and analyzing voice information and assessing fraud risk. It consists of three main elements: a server, a terminal, and a user.

[0775] Server Functions

[0776] The server focuses on acquiring, analyzing, and evaluating audio information. The server receives audio information transmitted in real time from terminals. Digital signal processing technology is used for audio processing, and a dedicated algorithm is implemented to remove noise from the audio information. For speech recognition technology, a cloud-based speech recognition service is used to convert speech into text. Specifically, Google Cloud Speech-to-Text API or other commercial speech recognition engines can be used.

[0777] The server further analyzes text information based on sentiment analysis technology to identify emotions from the user's statements. This process utilizes natural language processing libraries, with NLTK and other machine learning models proving effective. Additionally, open-source speech characteristic analysis tools are used to analyze the user's voice characteristics, enabling the detection of user stress, tension, and other factors. In assessing fraud risk, these analysis results are integrated to determine the likelihood of fraudulent activity.

[0778] If the risk of fraud is high, the server uses AI speech synthesis technology to generate and transmit a synthesized voice message, such as "This call is being monitored." Commercial speech synthesis engines such as Amazon Polly can be used for this purpose. The generated voice message aims to have a psychological effect to deter fraudulent activity.

[0779] Device functions

[0780] The terminal functions as an interface with the user, collecting and transmitting voice information. Voice information acquired during a call is filtered for noise before being sent to the server. It also plays a role in alleviating anxiety by displaying reassuring messages received from the server to the user.

[0781] User roles

[0782] Users are users of the system and are monitored by the system during calls. To minimize the burden on users while providing protection from fraudulent activities, the system is designed to make them feel safe through reassuring messages.

[0783] Specific example

[0784] For example, if a user receives a suspicious call, the server immediately analyzes the audio and text to detect signs of fraud. If it is deemed high-risk, a synthesized voice message such as "This call is being monitored" is sent to the caller. In addition, the user's device displays "You are currently on a secure call" to provide reassurance.

[0785] Examples of prompts for generative AI models

[0786] "What is needed to develop a prototype of a call content analysis system for evaluating the risk of fraudulent phone calls?"

[0787] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0788] Step 1: Obtain audio information

[0789] The device detects the start of a call and captures the call's audio information in real time using its internal microphone. The captured audio information is then sent to the server as a digital signal. The input is raw audio, and the output is digitized audio data.

[0790] Step 2: Noise Reduction and Speech Recognition

[0791] The server performs noise reduction processing on the received audio data. This reduces background noise and improves the accuracy of speech recognition. Next, the noise-reduced audio data is converted into text information using speech recognition technology. Specifically, a speech recognition API is used to convert sound waves into character data. The input is noise-reduced audio data, and the output is text data.

[0792] Step 3: Emotion Analysis

[0793] The server analyzes the generated text information to identify the speaker's emotions. Using natural language processing techniques, it calculates emotional indicators from words and phrases in the text. A sentiment analysis library is used for this analysis. The input is text data, and the output is an emotional score and an inferred emotional state.

[0794] Step 4: User Voice Characteristics Assessment

[0795] The server uses a voice analysis tool to evaluate the user's voice characteristics. It analyzes voice tone, tempo, pitch, etc., to identify signs of stress or tension. The input is voice data, and the output is the evaluation result regarding the user's voice characteristics.

[0796] Step 5: Assessing the risk of fraud

[0797] The server integrates the sentiment analysis results and voice characteristic evaluation results to assess fraud risk. This assessment uses an algorithm that compares the results with fraud patterns to calculate a risk score. The inputs are the sentiment score and voice characteristic evaluation results, and the output is the fraud risk score.

[0798] Step 6: Intervention using voice emulation

[0799] If the server determines that there is a high risk of fraud, it uses AI speech synthesis technology to generate a message such as "This call is being monitored" and plays the synthesized voice to the caller. The input is the fraud risk score, and the output is synthesized voice data.

[0800] Step 7: Displaying user notifications and safety messages

[0801] The terminal, based on information sent from the server, displays a reassuring message to the user on the screen, such as "You are currently on a secure call," providing a sense of psychological security. The input is instruction data from the server, and the output is the reassuring message displayed on the screen.

[0802] (Application Example 2)

[0803] 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".

[0804] Fraudulent use of electronic payments poses a significant threat to users. Preventing losses due to fraudulent activity and ensuring user confidence are crucial. However, current technology lacks the means to rapidly assess the possibility of fraud in real time and respond appropriately to users. As a result, users are not fully protected from fraudulent activity.

[0805] 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.

[0806] In this invention, the server includes a processing unit for acquiring voice data, a processing unit for analyzing the acquired voice data and identifying emotions, and a processing unit for evaluating the possibility of fraudulent activity based on the identified emotions. This makes it possible to detect changes in the user's emotions in real time and respond quickly if there are signs of fraudulent activity.

[0807] A "processing device for acquiring voice data" is a device that has the function of collecting voice information from phone calls and conversations and converting it into a digital data format that can be used within the system.

[0808] A "processing device for identifying emotions" is a device that has the function of identifying the emotional state of a speaker based on collected audio data and analyzing it as data.

[0809] A "processing device for evaluating the likelihood of fraudulent activity" is an electronic device that evaluates the likelihood of fraud in transactions or communications based on identified emotions and other relevant data, and makes decisions on actions based on that evaluation.

[0810] A "processing device that adjusts dialogue content using speech synthesis technology" is a device that operates technology used to generate voice data and send warnings and notifications to the user.

[0811] A "device that displays information to provide a sense of security" is a device that has the function of visually presenting messages and information to alleviate anxiety in the user.

[0812] This invention relates to an information processing system for preventing misuse. The invention mainly consists of a server and a user terminal, and performs a series of processes to analyze voice data, identify emotions to evaluate the possibility of misuse, and notify the user.

[0813] The server first acquires audio data transmitted from the user's terminal in real time. Mobile devices such as smartphones and tablets can be used for this process. The acquired audio data is then noise-reduced using FFT-based noise reduction technology. This audio data is then converted into text using the Google Cloud Speech-to-Text API.

[0814] Next, the server analyzes the converted text information using natural language processing techniques such as Hugging Face's Transformers library to identify the user's emotions. Based on the results of this emotion analysis, the server assesses the likelihood of fraudulent activity. This assessment includes comparing and matching the data against pre-registered, known fraudulent patterns.

[0815] If the server determines that there is a high probability of fraud, it will send a warning to the user's terminal via synthesized voice. The user will be notified of a message such as, "This transaction is being monitored. Do you really want to proceed?", which is generated using speech synthesis technology.

[0816] The user's terminal simultaneously displays visually reassuring information to reduce the likelihood of the user becoming involved in fraudulent activity. Furthermore, if fraudulent activity is suspected, the transaction is temporarily suspended, giving the user time to review the situation.

[0817] For example, when making a high-value online purchase, the system can issue a warning if the user is likely to be redirected to a fraudulent website, allowing them to reconfirm the transaction and prevent fraud.

[0818] An example of a prompt message is: "Create a program that analyzes the user's voice in real time, detects changes in emotion, and outputs a warning message in real time if there is a possibility of fraud."

[0819] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0820] Step 1:

[0821] The device acquires the user's voice data. This voice data is an analog signal and is collected in real time through the device's microphone. The input is the user's voice signal, and the output is digital voice data sent to the server. Noise reduction technology is used to reduce background noise and ensure clear voice acquisition.

[0822] Step 2:

[0823] The server converts the acquired audio data into text using the Google Cloud Speech-to-Text API. The input is digital audio data, and the output is text data generated as a result of analysis by the API. This conversion process makes it possible to accurately extract the user's speech from the audio.

[0824] Step 3:

[0825] The server performs emotion recognition processing based on the converted text data. Here, natural language processing technologies such as Hugging Face's Transformers are used to analyze and identify the user's emotional state (e.g., anxiety, confusion). The input is text data, and the output is emotion labels and emotion scores. This allows for an accurate understanding of the user's psychological state.

[0826] Step 4:

[0827] The server assesses the likelihood of fraud by comparing the user's emotional state with known fraud patterns. Inputs include emotional labels, emotional scores, and past fraud pattern data. Output is a risk score, which quantifies the likelihood of fraud. If this score exceeds a certain threshold, the server proceeds to the next step.

[0828] Step 5:

[0829] If the server determines that there is a high probability of fraudulent activity, it will use speech synthesis technology to create a warning message. For example, it might generate a warning such as "This transaction is being monitored." The input is a predefined message template and a fraud score, and the output is a synthesized voice file. This allows the user to recognize potential fraud.

[0830] Step 6:

[0831] The terminal receives warning messages from the server and notifies the user visually and audibly. Input is synthesized speech files and text messages, and output is the information presented through the terminal's speaker or display. This draws the user's attention to fraudulent activity and prompts them to review or cancel transactions as needed.

[0832] 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.

[0833] 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.

[0834] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[0835] 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.

[0836] 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.

[0837] 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.

[0838] 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.

[0839] 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.

[0840] 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."

[0841] 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.

[0842] 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.

[0843] 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.

[0844] 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.

[0845] 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.

[0846] 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.

[0847] 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.

[0848] 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.

[0849] 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.

[0850] 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.

[0851] 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.

[0852] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

[0853] The following is further disclosed regarding the embodiments described above.

[0854] (Claim 1)

[0855] Means for acquiring voice data during a call,

[0856] A method for identifying emotions by analyzing acquired audio data,

[0857] A means of assessing fraud risk based on identified emotions,

[0858] A means of altering the content of a conversation using voice emulation when the risk of fraud exceeds a certain threshold,

[0859] A means of displaying a message to notify the user of suspected fraud and alleviate their anxiety,

[0860] A system that includes this.

[0861] (Claim 2)

[0862] The system according to claim 1, further comprising means for converting the aforementioned audio data from an audio signal to text data.

[0863] (Claim 3)

[0864] The system according to claim 1, further comprising means for matching against known fraud patterns in assessing fraud risk.

[0865] "Example 1"

[0866] (Claim 1)

[0867] Means for acquiring voice information during a call,

[0868] A means for removing noise from acquired audio information and converting it into text information using speech recognition technology,

[0869] A method for analyzing textual information to identify emotions,

[0870] A means of assessing fraud risk based on identified emotions,

[0871] A means of altering the content of a conversation using speech synthesis when the risk of fraud exceeds a certain threshold,

[0872] A means of notifying users of suspected fraud and displaying information to alleviate their anxiety,

[0873] A system that includes this.

[0874] (Claim 2)

[0875] The system according to claim 1, further comprising means of using natural language processing techniques in identifying the aforementioned emotion.

[0876] (Claim 3)

[0877] The system according to claim 1, further comprising means for matching existing fraud patterns in the assessment of fraud risk.

[0878] "Application Example 1"

[0879] (Claim 1)

[0880] Means for acquiring voice information during a call,

[0881] A means of identifying emotions by analyzing acquired audio information,

[0882] A means of assessing the likelihood of fraud based on identified emotions,

[0883] A means of altering the content of a conversation using voice reproduction when the likelihood of fraud exceeds a certain standard,

[0884] A means of notifying users of suspected fraud and displaying information to alleviate their anxiety,

[0885] A means of sending dynamic warning messages to fraudulent communicators,

[0886] A system that includes this.

[0887] (Claim 2)

[0888] The system according to claim 1, further comprising means for converting the aforementioned audio information from an audio signal to text information.

[0889] (Claim 3)

[0890] The system according to claim 1, further comprising means for matching against known fraud types in the assessment of fraud potential.

[0891] "Example 2 of combining an emotion engine"

[0892] (Claim 1)

[0893] Means for acquiring voice information during a call,

[0894] A means for applying noise reduction processing to acquired audio information,

[0895] A means for converting noise-reduced audio information into text information using speech recognition technology,

[0896] A means of analyzing text information to identify the speaker's emotions,

[0897] A method for evaluating fraud risk based on sentiment analysis and user voice characteristics analysis,

[0898] A means of generating synthesized speech and altering the content of a call when the risk of fraud exceeds a certain threshold,

[0899] A means of displaying information that provides users with a sense of psychological security,

[0900] A system that includes this.

[0901] (Claim 2)

[0902] The system according to claim 1, which utilizes emotional state and user voice characteristics in assessing fraud risk.

[0903] (Claim 3)

[0904] The system according to claim 1, comprising means for generating a specific message using AI speech synthesis technology.

[0905] "Application example 2 when combining with an emotional engine"

[0906] (Claim 1)

[0907] A processing unit for acquiring audio data,

[0908] A processing device that analyzes acquired audio data and identifies emotions,

[0909] A processing device that evaluates the likelihood of misconduct based on identified emotions,

[0910] A processing device that adjusts the content of a conversation using speech synthesis technology when the possibility of fraudulent activity exceeds a predetermined standard,

[0911] A processing device that displays information to notify users of the possibility of fraud and provide them with a sense of security,

[0912] An information processing system that includes this.

[0913] (Claim 2)

[0914] The information processing system according to claim 1, further comprising a processing device that converts acquired audio data from audio signals into text information.

[0915] (Claim 3)

[0916] The information processing system according to claim 1, further comprising a processing device for comparing and matching with known fraud patterns when evaluating the possibility of fraudulent activity. [Explanation of symbols]

[0917] 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. Means for acquiring voice information during a call, A means of identifying emotions by analyzing acquired audio information, A means of assessing the likelihood of fraud based on identified emotions, A means of altering the content of a conversation using voice reproduction when the likelihood of fraud exceeds a certain standard, A means of notifying users of suspected fraud and displaying information to alleviate their anxiety, A means of sending dynamic warning messages to fraudulent communicators, A system that includes this.

2. The system according to claim 1, further comprising means for converting the aforementioned audio information from an audio signal into text information.

3. The system according to claim 1, further comprising means for matching against known fraud types in the assessment of fraud potential.

Images