System

Abstract

A system is provided.SOLUTION: A system comprising: means for acquiring conversation voice; means for transmitting the acquired voice data to a voice recognition engine in real time and converting the voice data into character data; means for analyzing the character data and detecting a set keyword; means for, when the keyword is detected, retroactively analyzing conversation contents before and after the keyword and extracting a main word or phrase; means for generating a search query based on the extracted word or phrase and making an inquiry to a search engine; and means for receiving and displaying a search result from the search engine.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

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

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

[0004] When field staff look up necessary information while serving customers, conventional methods require time to access the navigation system or other information sources, resulting in longer customer service times. This problem can lead to a decrease in store capacity, an increased burden on counter staff, and lower customer satisfaction. To solve these problems, there is a need to reduce research time and provide more efficient service. [Means for solving the problem]

[0005] A system including a means for acquiring conversational voice, a means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, a means for analyzing the text data and detecting set keywords, a means for, when a keyword is detected, retroactively analyzing the content of the conversation before and after the keyword and extracting key words and phrases, a means for generating a search query based on the extracted words and phrases and making an inquiry to a search engine, and a means for receiving and displaying search results from the search engine, can significantly reduce the time spent on research and improve customer service efficiency.

[0006] "Conversational audio" refers to the audio information exchanged between customers and on-site staff during service.

[0007] "Capture" refers to the process of collecting audio and text data into the system.

[0008] A "voice recognition engine" refers to a technology or system that analyzes voice data and converts it into text data.

[0009] "Character data" refers to information in text format converted from audio data.

[0010] "Analysis" refers to the process of extracting and understanding specific elements and meanings from the data obtained.

[0011] "Keywords" refer to specific words or phrases that the system considers to be particularly important and is intended to detect.

[0012] "Detect" refers to recognizing and identifying specific information or data based on set conditions.

[0013] "Preceding and following conversation content" refers to the audio and text data that exists before and after the detected keyword.

[0014] A "natural language processing engine" refers to technology or a system that allows a computer to understand and analyze natural language.

[0015] "Key words and phrases" refer to words and short phrases that are deemed to be important or relevant in the conversation.

[0016] A "search query" refers to a word or phrase entered into a search engine to search for specific information.

[0017] "Search engine" refers to a technology or system that searches for information and provides results based on an input query.

[0018] "Making a query" refers to sending a query to another system or database to obtain specific information.

[0019] "Search Results" means the information or data provided in response to a Search Query. [Brief explanation of the drawings]

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

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

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

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

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

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

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

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

[0028] [First embodiment]

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

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

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

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

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

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

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

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

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

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

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

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

[0041] Overview of the voice recognition system

[0042] This invention is a speech recognition system that enables on-site staff to quickly respond to customer questions while serving customers. This system captures conversational speech, performs real-time speech recognition, and automatically searches for and displays the necessary information.

[0043] Overview of program processing

[0044] 1. Acquiring audio data

[0045] Terminal: A microphone installed at the customer service counter constantly records conversations and collects real-time audio data, which is temporarily stored in a buffer within the system.

[0046] 2. Real-time speech recognition

[0047] Device: The acquired voice data is sent from the buffer to the speech recognition engine, which converts the voice data into text data using, for example, the Google Speech-to-Text API or a similar service.

[0048] Server: Receives the converted character data and saves it in a database with a timestamp.

[0049] 3. Keyword detection

[0050] Server: Analyzes the character data and detects the set keyword (e.g., "confirm") and stores the partial string up to the appearance of the keyword.

[0051] 4. Analysis of conversation content

[0052] Server: When a keyword is detected, the server extracts the surrounding conversation (usually within one minute) and sends the extracted text data to a natural language processing (NLP) engine to extract key words and phrases.

[0053] 5. Search Query Generation

[0054] Server: Generates a search query based on the extracted words and phrases. For example, a specific query such as "new account opening procedure" is generated.

[0055] 6. Contacting Search Engines

[0056] Server: Sends the generated query as an HTTP request to the search engine.

[0057] 7. Retrieving and Displaying Search Results

[0058] Server: Receives search results from the search engine and displays them on the terminals of field staff.

[0059] Terminal: The received search results are displayed on the on-site staff's display.

[0060] 8. Feedback Loops

[0061] User (field staff): Answers the customer based on the displayed information. If more specific information is needed, the system asks additional questions. The system receives feedback and searches for information again.

[0062] Specific examples

[0063] Scenario 1: Consultation regarding opening a bank account

[0064] 1. Audio Acquisition

[0065] Customer: "I'd like to open a new account and would like to confirm the procedure."

[0066] Device: Record this conversation and store the audio data in a buffer.

[0067] 2. Voice Recognition

[0068] On the device: Sends the audio data in the buffer to the Google Speech-to-Text API in real time.

[0069] Server: Receives the returned text data "I would like to open a new account and would like to confirm the procedure." and saves it in the database with a timestamp.

[0070] 3. Keyword detection

[0071] Server: Analyzes the text data and detects the keyword "confirm." At this point, the part "I would like to open a new account. What are the procedures?" is retained as the analysis target.

[0072] 4. Analysis of conversation content

[0073] Server: Uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[0074] 5. Search Query Generation

[0075] Server: Generate a specific search query: "new account opening procedure."

[0076] 6. Contacting Search Engines

[0077] Server: Sends the generated query as an HTTP request to the search engine's API.

[0078] 7. Retrieving and Displaying Search Results

[0079] Server: Sends search results received from the search engine to the terminals of field staff.

[0080] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[0081] 8. Feedback Loops

[0082] User (on-site staff): Answers the customer based on the displayed information. For example, "To open a new account, you need the following documents." If the customer has a more detailed question, the system detects the keywords again and searches for the necessary information.

[0083] This system allows field staff to efficiently obtain information and quickly answer customer questions, which is expected to reduce customer service time and improve customer satisfaction.

[0084] The processing flow will be explained below.

[0085] Step 1:

[0086] Terminal: A microphone installed at the customer service counter continuously records conversations and collects voice data in real time. The voice data is temporarily stored in a buffer within the system.

[0087] Step 2:

[0088] Terminal: The acquired voice data is sent from the buffer to the voice recognition engine, which converts the voice data into text data using, for example, the Google Speech-to-Text API.

[0089] Step 3:

[0090] Server: Receives converted text data from the speech recognition engine and stores it in a database with a timestamp.

[0091] Step 4:

[0092] Server: Analyzes the character data and detects the set keyword (e.g., "confirm") and stores the partial string up to the appearance of the keyword.

[0093] Step 5:

[0094] Server: When a keyword is detected, the conversation content before and after it (usually within one minute) is extracted.

[0095] Step 6:

[0096] Server: The extracted text data is sent to a natural language processing engine to extract key words and phrases.

[0097] Step 7:

[0098] Server: Generates a search query based on the extracted words and phrases. For example, it generates a specific query such as "open a new account."

[0099] Step 8:

[0100] Server: Sends the generated query as an HTTP request to the search engine's API.

[0101] Step 9:

[0102] Server: Receives search results from the search engine, formats them, and sends them to the terminals of field staff.

[0103] Step 10:

[0104] Terminal: The received search results are displayed on the on-site staff's display.

[0105] Step 11:

[0106] User (field staff): Responds to the customer based on the displayed information. If more specific information is required, the user asks additional questions. The system receives feedback and searches for information again based on this.

[0107] Example 1

[0108] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0109] In customer service, staff need a lot of information to respond to customer questions quickly and accurately. However, manually searching for information is time-consuming and can lead to a decrease in customer satisfaction. Furthermore, a noisy and multitasking environment on-site makes it difficult to respond quickly. Therefore, there is a need for a system that can recognize voice in real time and automatically present relevant information.

[0110] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0111] In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, means for analyzing the text data and detecting set keywords, means for, when a keyword is detected, retroactively analyzing the conversation content before and after the keyword and extracting key words and phrases, means for generating a search query based on the extracted words and phrases and making an inquiry to a search engine, means for receiving and displaying search results from the search engine, and means for searching for additional information based on the feedback. This enables on-site staff to respond to customer questions quickly and accurately and improve customer satisfaction.

[0112] "Conversational audio" refers to the verbal exchanges that take place between customers and on-site staff during customer service operations.

[0113] "Audio data" means a digital representation of speech collected using a device such as a microphone.

[0114] A "voice recognition engine" is software or a service that receives voice data as input and converts it into text-formatted character data.

[0115] "Character data" refers to text-format data converted by a voice recognition engine, and primarily refers to information expressed using characters (such as alphabets or kanji).

[0116] "Keywords" are words or phrases that are considered particularly important in the conversation and serve as triggers for the system to perform specific operations.

[0117] The "conversation content" is a series of words exchanged between the customer and the on-site staff, and is a group of text data containing specific information.

[0118] A "natural language processing engine" is an algorithm or software that analyzes text data and extracts key words and phrases.

[0119] A "search query" is a specific phrase or combination of words generated to ask a search engine.

[0120] A "search engine" is a system or software that searches the Internet or databases for relevant information based on an input search query.

[0121] "Search results" refers to the list or content of information that a search engine provides based on a search query.

[0122] "Feedback" refers to the process and results of field staff conducting additional information searches based on the information and results they receive from the system.

[0123] This invention is a speech recognition system that enables on-site staff to respond quickly and accurately to customer questions during customer service duties. This system captures conversational voices, performs real-time speech recognition, and automatically searches for and displays necessary information, significantly improving on-site efficiency.

[0124] 1. Hardware and Software Used

[0125] The system hardware configuration is as follows:

[0126] Microphone (to capture speech)

[0127] Terminal (to acquire audio data and perform initial processing)

[0128] Server (for data analysis, generating search queries, and communicating with search engines)

[0129] Display (to show results to field staff)

[0130] The system software configuration is as follows:

[0131] Google Speech-to-Text API (voice recognition engine)

[0132] NLP engines (for natural language processing, e.g. SpaCy)

[0133] Search engine APIs (for information searches, e.g., Google and Bing APIs)

[0134] 2. System processing overview

[0135] The process of this speech recognition system involves the following steps:

[0136] Acquiring audio data:

[0137] The device collects conversational audio in real time through a microphone and temporarily stores it in a buffer. This audio data is stored in PCM format.

[0138] Real-time speech recognition:

[0139] The device splits the audio data in the buffer into regular batches (e.g., every 5 seconds) and sends them to the Google Speech-to-Text API, which converts the audio into text data.

[0140] The server stores the received character data in a database with a timestamp.

[0141] Keyword detection and conversation analysis:

[0142] The server periodically scans new text data for set keywords, and if a keyword is found, sends the surrounding conversational content to an NLP engine to extract key words and phrases.

[0143] Search query generation and query:

[0144] The server then combines the extracted words and phrases to generate a search query, which it then sends as an HTTP request to the search engine's API, which then returns the search results.

[0145] Displaying search results:

[0146] The server displays the search results received from the search engine on the terminal's display, and the field staff responds to the customer based on the displayed information.

[0147] Feedback Loop:

[0148] The user (field staff) responds to the customer based on the displayed information. If more details are needed, the user provides feedback to the system and searches for additional information.

[0149] 3. Explanation of specific examples

[0150] Scenario 1: Consultation regarding opening a bank account

[0151] 1. Audio capture:

[0152] Customer: "I'd like to open a new account and would like to confirm the procedure."

[0153] The device records this conversation and stores the audio data in a buffer.

[0154] 2. Speech Recognition:

[0155] The device sends the buffered audio data to the Google Speech-to-Text API in real time.

[0156] The server receives the returned text data, "I would like to open a new account and would like to confirm the procedure," and stores it in the database with a timestamp.

[0157] 3. Keyword detection:

[0158] The server analyzes the text data and detects the keyword "confirmation." At this point, it retains the part "I would like to open a new account. What are the procedures?" as the analysis target.

[0159] 4. Conversation Analysis:

[0160] The server uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[0161] 5. Generating search queries:

[0162] The server generates a specific search query, "new account opening procedure," and sends it as an HTTP request to the search engine's API.

[0163] 6. Retrieving and displaying search results:

[0164] The server sends the search results received from the search engine to the terminal of the on-site staff and displays them on the screen.

[0165] The terminal then displays the search results on the on-site staff's screen, such as information about the documents and conditions required to open a new account.

[0166] 7. Feedback Loop:

[0167] The user (field staff) responds to the customer based on the displayed information. If the customer has a more detailed question, the system will again detect the keywords and search for the necessary information.

[0168] This system allows field staff to efficiently obtain information and quickly answer customer questions, which is expected to reduce customer service time and improve customer satisfaction.

[0169] Example prompt sentence:

[0170] I would like to open a new account and would like to confirm the procedure.

[0171] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0172] Step 1: Acquire audio data

[0173] The terminal captures conversational audio in real time from a microphone installed at the customer service counter. The input is the conversational audio between the customer and the front-line staff, which is converted into digital format and stored in a buffer in the terminal's memory. Specifically, the audio signal is converted from analog to digital format and continuously written to the buffer in PCM format.

[0174] Step 2: Real-time speech recognition

[0175] The device buffers the audio data in regular batches (for example, every 5 seconds). The input is audio data, which is sent to the Google Speech-to-Text API to be converted into text. At this stage, the audio data is analyzed by a speech recognition engine, and text data is generated as output. This text data is then sent from the device to the server.

[0176] Step 3: Storing Character Data

[0177] The server stores the text data received from the device in a database with a timestamp. The input is text data, and a timestamp is added during the process of saving it in the database. Specifically, the text data is stored in an SQLite database, and synchronization between voice and text is maintained.

[0178] Step 4: Keyword detection

[0179] The server periodically scans new text data to detect configured keywords. The input is character data, and it searches for specific keywords using regular expressions, etc. This detection process outputs the part containing the configured keyword. Specifically, when a keyword is detected, the conversation content before and after it is temporarily stored in memory.

[0180] Step 5: Analyzing the conversation

[0181] The server sends the conversational content, including one minute before and after the text where the keyword was detected, to a natural language processing (NLP) engine. The input is text data, which is analyzed by an NLP engine (e.g., SpaCy) to extract key words and phrases. The output is a list of extracted words and phrases. Specifically, the NLP engine tokenizes the text and identifies and extracts key words such as nouns and verbs.

[0182] Step 6: Generating a search query

[0183] The server generates a query based on the extracted words and phrases in a format that is easy for users to search. The input is a list of key words and phrases, and a specific search query is generated by string concatenation. The output is the search query. For example, a phrase such as "new account opening procedure" is generated.

[0184] Step 7: Contact the search engines

[0185] The server then sends the generated search query to the search engine's API. The input is the search query, which is queried to the search engine (e.g., Google or Bing) using an HTTP GET request. The output is the search results. Specifically, the query is appended to a URL and sent to the search engine, which returns the results in JSON format.

[0186] Step 8: Retrieving and Displaying Search Results

[0187] The server sends the search results received from the search engine to the terminal and displays them on the on-site staff's display. The input is the search results from the search engine, which are formatted for display on the terminal's display. The output is the information displayed on the on-site staff's display. Specifically, information such as "Documents and conditions required for opening a new account" is displayed.

[0188] Step 9: Feedback Loop

[0189] The user (on-site staff) responds to the customer based on the displayed information. If more specific information is required, they ask an additional question. The input is the customer's additional question, and the content of this question is picked up again from the microphone, and processing returns to step 1 for reprocessing. The output is updated search results and a new answer. Specifically, more detailed information is searched again based on the customer's additional question and displayed.

[0190] (Application example 1)

[0191] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0192] Conventional voice recognition systems were not sufficient for quickly obtaining necessary information during conversations with customers. Furthermore, search functions based on specific keywords alone sometimes made it difficult for on-site staff to accurately answer customer questions. As a result, customer satisfaction and response efficiency declined.

[0193] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0194] In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, means for analyzing the text data and detecting set keywords, means for retroactively analyzing the conversation content before and after a keyword is detected and extracting key words and phrases, means for generating a search query based on the extracted words and phrases and making an inquiry to a search engine, means for receiving search results from the search engine and displaying them on a user terminal, and means for performing another information search based on the feedback. This enables on-site staff to quickly and accurately answer customer questions.

[0195] "Conversational voice" refers to the voice exchanged between the user and the person answering the call.

[0196] "Means for acquiring" refers to a device or system for collecting audio data.

[0197] "Send to the speech recognition engine in real time" means that collected speech data is immediately transferred to the speech recognition engine.

[0198] "Means for converting to text data" refers to a process or device that converts audio data into text data.

[0199] "Analyzing character data" refers to the process of interpreting the converted text data and extracting necessary information.

[0200] "Means for detecting set keywords" refers to a system that identifies specific predefined words or phrases within text data.

[0201] "When a keyword is detected, the conversation content before and after it is analyzed" refers to analyzing the text before and after the detected keyword and extracting related content.

[0202] "Means for extracting key words and phrases" refers to a system that finds important words and phrases from the analyzed text data.

[0203] "Means for generating a search query and sending it to a search engine" refers to creating a search query based on the extracted words and phrases and sending that query to a search engine.

[0204] "Means for receiving search results from a search engine and displaying them on a user's device" refers to a system that receives responses from a search engine and displays that information on a user's device.

[0205] "Means of searching for information again based on feedback" refers to the process of generating a search query again based on additional questions or requests from the user to obtain new information.

[0206] "User devices" refer to devices used by customers and field staff, such as smartphones and tablets.

[0207] A "natural language processing engine" is a system that analyzes text data and extracts meaning and related information.

[0208] "General-purpose device" refers to a device that has a wide range of functions and is not limited to a specific use, such as a general-purpose smartphone or tablet.

[0209] The system for implementing this invention uses the following hardware and software. First, the hardware used includes a smartphone with a microphone for recording voice and a server for data processing. The software includes a Python program, a voice recognition library (e.g., the SpeechRecognition library), the Google Speech-to-Text API, and a natural language processing engine (e.g., an NLP engine for natural language processing).

[0210] The server uses a speech recognition engine to convert voice data into text data in real time. The converted text data is analyzed to detect set keywords. When a keyword is detected, the conversation content before and after it is analyzed to extract key words and phrases. A search query is generated based on these extracted words and phrases, and this search query is used as data to query a search engine.

[0211] The search results returned by the search engine are displayed on the user's smartphone. Here, "user device" refers to the device used by the customer or field staff, such as a smartphone or tablet. Furthermore, the system also incorporates a mechanism for re-searching information based on user feedback, enabling more accurate responses to questions.

[0212] For example, when a store clerk asks, "Is this item in stock?", the system instantly recognizes the keyword "in stock" and searches for and displays inventory information. This information is detailed information that customers want, and helps field staff respond quickly.

[0213] Example prompt sentence:

[0214] Input Sentence: Customer: "Is this item in stock?"

[0215] Example output: System: In response to the question "Is this item in stock?", the system recognizes the keyword "in stock" and searches for and displays stock information.

[0216] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0217] Step 1:

[0218] The user uses the microphone on their smartphone to record the conversation. This voice data is temporarily stored in a buffer within the system. The input is the conversation between the user and the customer, and the output is voice data.

[0219] Step 2:

[0220] The device sends the captured voice data in real time to a voice recognition engine (for example, Google Speech-to-Text API). This engine converts the voice data into text data. The input is voice data, and the output is text data.

[0221] Step 3:

[0222] The server receives the converted character data and stores it in a database with a timestamp. The input is character data, and the output is character data with a timestamp. This timestamp is used for later analysis.

[0223] Step 4:

[0224] The server analyzes the text data and detects the set keywords. For example, keywords such as "stock" and "price" are detected. The input is the text data, and the output is the detected keywords.

[0225] Step 5:

[0226] When a keyword is detected, the server analyzes the conversation content before and after the keyword and extracts key words and phrases. For example, key phrases such as "this product" and "is it in stock" are extracted. The input is the keyword and text data, and the output is the extracted key words and phrases.

[0227] Step 6:

[0228] The server generates a search query based on the extracted words and phrases. This query is intended to retrieve specific information. For example, a search query might be generated for "this product in stock." The input is the main word or phrase, and the output is the search query.

[0229] Step 7:

[0230] The server sends the generated search query to the search engine as an HTTP request and waits for a response from the search engine (e.g., Google). The input is the search query, and the output is the search results from the search engine.

[0231] Step 8:

[0232] The server receives search results from the search engine and displays them on the user's terminal. These results are information that the user provides to their customers. The input is the search results from the search engine, and the output is the information displayed on the user's terminal.

[0233] Step 9:

[0234] The user responds to the customer based on the displayed information. If the customer requires more detailed information, the user asks an additional question, and the system searches for information again based on the feedback. The input is the additional question from the customer, and the output is the information obtained again.

[0235] This process allows the system to provide quick and accurate answers to customer questions.

[0236] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0237] Overview of speech and emotion recognition systems

[0238] This invention is a speech and emotion recognition system that aims to provide more accurate service by not only enabling on-site staff to respond quickly to customer questions while serving customers, but also by recognizing customer emotions and utilizing that information. This system captures conversational voice, performs speech and emotion recognition in real time, and automatically searches for and displays the necessary information.

[0239] Overview of program processing

[0240] 1. Acquiring audio data

[0241] Terminal: A microphone installed at the customer service counter constantly records conversations, collecting voice data in real time. This voice data is temporarily stored in a buffer within the system.

[0242] 2. Real-time speech recognition

[0243] Terminal: The acquired voice data is sent from the buffer to the speech recognition engine, which converts the voice data into text data. For example, the Google Speech-to-Text API is used to convert voice to text.

[0244] Server: The character data received from the speech recognition engine is saved in a database with a timestamp.

[0245] 3. Keyword detection

[0246] Server: Analyzes the character data and detects the set keyword (e.g., "confirm") and stores the partial string up to the appearance of the keyword.

[0247] 4. Analysis of conversation content

[0248] Server: When a keyword is detected, the server extracts the surrounding conversation (usually within one minute) and sends the extracted text data to a natural language processing (NLP) engine to extract key words and phrases.

[0249] 5. Emotional Recognition

[0250] Device: Voice data is input into an emotion recognition engine to analyze the user's emotions. For example, it determines whether the user is angry or happy based on the tone and speed of the voice.

[0251] Server: Stores the user's emotional data based on the analyzed emotional information.

[0252] 6. Search Query Generation

[0253] Server: Generates a search query based on the extracted words and phrases and sentiment information. For example, "Open a new account" and if the user is angry, a quick response is required.

[0254] 7. Contacting Search Engines

[0255] Server: Sends the generated query as an HTTP request to the search engine's API.

[0256] 8. Retrieving and Displaying Search Results

[0257] Server: Receives search results from the search engine and sends them to the terminals of field staff. Search results also reflect emotional information, for example, prioritizing content that requires a quick response.

[0258] Terminal: The received search results are displayed on the on-site staff's display.

[0259] 9. Feedback Loops

[0260] User (on-site staff): Responds to the customer based on the displayed information. If more specific information is required, the system asks additional questions. The system receives feedback and searches for information again. During this process, the user's emotions are continuously monitored.

[0261] Specific examples

[0262] Scenario 1: Consultation regarding opening a bank account

[0263] 1. Audio Acquisition

[0264] Customer: "I'd like to open a new account and would like to confirm the procedure."

[0265] Device: Record this conversation and store the audio data in a buffer.

[0266] 2. Voice Recognition

[0267] On the device: Sends the audio data in the buffer to the Google Speech-to-Text API in real time.

[0268] Server: Receives the returned text data "I would like to open a new account and would like to confirm the procedure." and saves it in the database with a timestamp.

[0269] 3. Keyword detection

[0270] Server: Analyzes the text data and detects the keyword "confirm." At this point, the part "I would like to open a new account. What are the procedures?" is retained as the analysis target.

[0271] 4. Analysis of conversation content

[0272] Server: Uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[0273] 5. Emotional Recognition

[0274] On the device: An emotion recognition engine is used to determine the user's emotion. In this case, it is determined that the customer has a question.

[0275] 6. Search Query Generation

[0276] Server: Generate a specific search query, such as "open a new account," based on sentiment information. If the customer is in a hurry, a quick response is required.

[0277] 7. Contacting Search Engines

[0278] Server: Sends the generated query as an HTTP request to the search engine's API.

[0279] 8. Retrieving and Displaying Search Results

[0280] Server: Sends search results received from the search engine to the terminals of field staff. The results are displayed with priority given to items that require immediate attention.

[0281] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[0282] 9. Feedback Loops

[0283] User (on-site staff): Responds to the customer based on the displayed information. For example, "To open a new account, you need the following documents." If the customer has a more detailed question, the system will again detect keywords, search for the necessary information, and continuously monitor the customer's emotions.

[0284] This system allows on-site staff to efficiently obtain information and respond to customers with consideration for their feelings, which is expected to reduce customer service time and improve customer satisfaction.

[0285] The processing flow will be explained below.

[0286] Step 1:

[0287] Terminal: A microphone installed at the customer service counter continuously records conversations and collects voice data in real time. The voice data is temporarily stored in a buffer within the system.

[0288] Step 2:

[0289] Terminal: The acquired voice data is sent from the buffer to a voice recognition engine, for example, using the Google Speech-to-Text API to convert the voice data into text data.

[0290] Step 3:

[0291] Server: Receives converted text data from the speech recognition engine and stores it in a database with a timestamp.

[0292] Step 4:

[0293] Server: Analyzes the character data and detects the set keyword (e.g., "confirm"), and stores the substring up to the detected keyword.

[0294] Step 5:

[0295] Server: When a keyword is detected, the conversation content before and after it (usually within one minute) is extracted.

[0296] Step 6:

[0297] Server: The extracted text data is sent to a natural language processing (NLP) engine to extract key words and phrases.

[0298] Step 7:

[0299] Device: Voice data is input into an emotion recognition engine to analyze the user's emotions. For example, the emotional state is determined from the tone, speed, and pitch of the voice, and emotions such as joy, anger, and sadness are identified.

[0300] Step 8:

[0301] Server: Based on the analyzed emotional information, the user's emotional data is stored in a database.

[0302] Step 9:

[0303] Server: Generates search queries based on the extracted words and phrases and emotion information. For example, it generates a specific query such as "new account opening procedure" and sets a high priority if the user is angry.

[0304] Step 10:

[0305] Server: Sends the generated search query as an HTTP request to the search engine's API.

[0306] Step 11:

[0307] Server: Receives search results from the search engine, formats them, and sends them to the terminals of field staff. Search results also reflect emotional information, and priority is given to content that requires a quick response.

[0308] Step 12:

[0309] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[0310] Step 13:

[0311] User (on-site staff): Responds to the customer based on the displayed information. If more specific information is required, the system asks additional questions. The system receives feedback and searches for information again. During this process, the system also continuously monitors the user's emotions.

[0312] Specific examples

[0313] Scenario 1: Consultation regarding opening a bank account

[0314] 1. Audio Acquisition

[0315] Customer: "I'd like to open a new account and would like to confirm the procedure."

[0316] Device: Record this conversation and store the audio data in a buffer.

[0317] 2. Voice Recognition

[0318] On the device: Sends the audio data in the buffer to the Google Speech-to-Text API in real time.

[0319] Server: Receives the returned text data "I would like to open a new account and would like to confirm the procedure." and saves it in the database with a timestamp.

[0320] 3. Keyword detection

[0321] Server: Analyzes the text data and detects the keyword "confirm." At this point, the part "I would like to open a new account. What are the procedures?" is retained as the analysis target.

[0322] 4. Analysis of conversation content

[0323] Server: Uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[0324] 5. Emotional Recognition

[0325] On the device: An emotion recognition engine is used to determine the user's emotion. In this case, it is determined that the customer has a question.

[0326] 6. Search Query Generation

[0327] Server: Generate a specific search query, such as "open a new account," based on sentiment information. If the customer is in a hurry, a quick response is required.

[0328] 7. Contacting Search Engines

[0329] Server: Sends the generated query as an HTTP request to the search engine's API.

[0330] 8. Retrieving and Displaying Search Results

[0331] Server: Sends search results received from the search engine to the terminals of field staff. The results are displayed with priority given to items that require immediate attention.

[0332] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[0333] 9. Feedback Loops

[0334] User (on-site staff): Responds to the customer based on the displayed information. For example, "To open a new account, you need the following documents." If the customer has a more detailed question, the system will again detect keywords, search for the necessary information, and continuously monitor the customer's emotions.

[0335] This system allows on-site staff to efficiently obtain information and respond to customers with consideration for their feelings, which is expected to reduce customer service time and improve customer satisfaction.

[0336] Example 2

[0337] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0338] While serving customers, on-site staff need to be able to respond quickly to customer questions and accurately recognize their emotions to provide more appropriate service. However, conventional systems only use voice recognition and do not analyze emotions, making it difficult to improve customer satisfaction. In addition, the need to process large amounts of data in real time requires specialized hardware and advanced software, making on-site implementation and operation complicated.

[0339] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0340] In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, means for analyzing the text data and detecting set keywords, means for, when a keyword is detected, retroactively analyzing the conversation content before and after the keyword and extracting key words and phrases, means for analyzing the tone, pitch, and speed of the voice data and recognizing emotions, means for generating search queries based on the emotional information and the extracted words and phrases and making inquiries to a search engine, and means for receiving and displaying search results from the search engine. This enables quick and accurate responses to customer questions and the provision of services that take emotions into consideration.

[0341] "Conversational audio" refers to the audio information exchanged between customers and on-site staff during customer service.

[0342] "Audio data" refers to information that is a digital recording of conversational voice.

[0343] A "voice recognition engine" is software that analyzes voice data and converts it into text data.

[0344] "Character data" refers to text information generated from voice data by a voice recognition engine.

[0345] A "keyword" is a particularly important word or phrase in the character data to be analyzed.

[0346] A "natural language processing engine" is software that analyzes text data and extracts key words and phrases.

[0347] A "search query" is a string of characters or phrases generated to search for specific information.

[0348] A "search engine" is a system or software that searches for information based on a search query.

[0349] "Tone" is an attribute that represents the pitch and quality of sound in audio data.

[0350] "Pitch" is an attribute that represents the frequency of sound in audio data.

[0351] "Speed" is an attribute that indicates the speaking speed of the voice in the voice data.

[0352] "Emotion recognition" is the process of identifying a speaker's emotions by analyzing the tone, pitch, and rate of speech data.

[0353] "Field staff" refers to employees who handle customer service.

[0354] "Customer" means the person receiving the service.

[0355] "Display" means a device that visually displays search results or other information.

[0356] A "buffer" is a memory space for temporarily storing audio data.

[0357] This invention relates to a system that enables on-site staff to provide more appropriate service by quickly and accurately answering customer questions while serving customers and recognizing their emotions. The system collects conversational voices, performs real-time speech and emotion recognition, and automatically searches for and displays the necessary information.

[0358] Specifically, the device uses a microphone installed at the customer service counter to continuously record the conversation between the customer and the on-site staff, temporarily storing the audio data in a buffer, and then sends this audio data in real time to a speech recognition engine (for example, Google Speech-to-Text API) and converts it into text data.

[0359] The server then stores the acquired text data in a database with a timestamp. The stored text data is analyzed to detect set keywords (e.g., "confirmation"). If a keyword is detected, the server extracts the surrounding conversation content and uses a natural language processing (NLP) engine (e.g., spaCy or NLTK) to extract key words and phrases.

[0360] Furthermore, the device analyzes the acoustic characteristics of the voice data, such as tone, pitch, and speed, and identifies the user's emotions using an emotion recognition engine (e.g., IBM Watson Tone Analyzer). This emotion information is analyzed together with the text data, and the server performs processing according to the emotion.

[0361] The server generates a search query based on the extracted key words and phrases and the identified sentiment information. The generated search query is sent as an HTTP request to a search engine (e.g., a general-purpose search engine API). The search results from the search engine are received by the server, appropriately filtered, and then sent to the device.

[0362] The terminal displays the received search results on the on-site staff's screen, allowing them to obtain the necessary information in a timely manner and respond to customers quickly and efficiently.If additional questions arise, a feedback loop is created in which the keywords are detected again and the necessary information is searched for.

[0363] As a concrete example, consider a consultation about opening a bank account. When a customer says, "I'd like to open a new account, but I'd like to confirm the procedure," the system records this conversation and saves the voice data in a buffer. A speech recognition engine converts this voice data into text data, generating the text data, "I'd like to open a new account, but I'd like to confirm the procedure."

[0364] Next, the server detects the keyword "confirmation" and analyzes and extracts the conversation content before and after it, obtaining key words and phrases such as "new account" and "opening procedure." At the same time, the device performs emotion recognition to identify the customer's question. The server generates a search query for "new account opening procedure" and displays appropriate search results if a quick response is needed.

[0365] An example of a prompt sentence is, "I would like to know the procedure for opening a new account at the bank. What documents do I need?" By using this, the system can quickly search for the relevant information and provide it to the field staff, allowing the customer to be served appropriately.

[0366] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0367] Step 1: Acquire audio data

[0368] Input: Voice conversation between customer and on-site staff

[0369] Specific operation: The terminal uses a microphone installed at the customer service counter to continuously record the conversation between the customer and the on-site staff. The recorded audio data is temporarily stored in a buffer within the terminal.

[0370] Output: Buffered audio data

[0371] Step 2: Real-time speech recognition

[0372] Input: Buffered audio data

[0373] Specific operation: The device sends the acquired voice data to a voice recognition engine (e.g., Google Speech-to-Text API) in real time, and converts the voice data into text data.

[0374] Output: Text data returned from the speech recognition engine

[0375] Step 3: Storing Character Data

[0376] Input: Character data returned from the speech recognition engine

[0377] Specific operation: The server adds a timestamp to the received character data and saves it in the database.

[0378] Output: Character data stored in the database with a timestamp

[0379] Step 4: Keyword detection

[0380] Input: Character data stored in a database

[0381] Specific operation: The server analyzes the text data and detects the set keyword (e.g., "confirm"), triggering the detection of the keyword and saving the conversation content before and after the keyword.

[0382] Output: The part where the keyword was detected and the conversation content before and after it

[0383] Step 5: Analyzing the conversation

[0384] Input: Conversation content before and after the keyword

[0385] What it does: The server uses a natural language processing engine (e.g., spaCy or NLTK) to extract key words and phrases, including tokenization, part-of-speech tagging, and named entity recognition.

[0386] Output: Extracted key words and phrases

[0387] Step 6: Recognize emotions

[0388] Input: Buffered audio data

[0389] Specific operation: The device analyzes acoustic features such as tone, pitch, and speed of the voice data and uses an emotion recognition engine (e.g., IBM Watson Tone Analyzer) to identify the user's emotion.

[0390] Output: Identified emotion information

[0391] Step 7: Generating a search query

[0392] Input: extracted key words and phrases, sentiment information

[0393] Specific operation: The server fuses the extracted words and phrases with sentiment information to generate search queries.

[0394] Output: Generated search query

[0395] Step 8: Contact the search engines

[0396] Input: Generated search query

[0397] Specific operation: The server creates and sends an HTTP request to send the generated query to the search engine's API.

[0398] Output: HTTP request to the search engine's API

[0399] Step 9: Retrieving and Displaying Search Results

[0400] Input: Search results from a search engine

[0401] Specific operation: The server analyzes search results from the search engine, filters and prioritizes them appropriately, and then sends them to the on-site staff's terminal. The terminal then displays the received search results on the on-site staff's display.

[0402] Output: Search results displayed on the device display

[0403] Step 10: Feedback Loop

[0404] Input: Search results displayed on the device display, additional questions from the customer

[0405] Specific operation: The user (on-site staff) answers the customer based on the displayed information. If the customer has a more detailed question, the system will detect keywords again, search for the necessary information, and continuously monitor the customer's emotions.

[0406] Output: Updated search results and continuously monitored sentiment information

[0407] (Application example 2)

[0408] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0409] When on-site staff deal with customers, they are required to respond to their questions quickly and accurately. Furthermore, understanding customer emotions and providing appropriate responses leads to improved customer satisfaction. However, with conventional systems, it has been difficult to accurately recognize customer emotions and provide information in real time. Therefore, there is a need for a system that integrates emotion recognition and voice recognition to provide appropriate services in real time that take customer emotions into consideration.

[0410] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into character data, means for analyzing the character data and detecting set keywords, means for recognizing emotions based on the voice data, means for generating a search query based on the extracted words, phrases, and emotion information and making an inquiry to a search engine, and means for receiving and displaying search results from the search engine. This enables field staff to understand customer emotions and quickly provide information that is appropriate for those emotions.

[0411] "Means for acquiring conversational audio" refers to devices or technologies for recording conversations between customers and wait staff in real time and collecting audio data.

[0412] "Means for transmitting voice data to a voice recognition engine in real time and converting it into text data" refers to a device or technology that immediately sends acquired voice data to a voice recognition engine and converts it from voice to text data.

[0413] "Means for analyzing character data and detecting set keywords" refers to a device or technology that analyzes the converted character data and finds specific keywords that have been set in advance.

[0414] "Means for analyzing the conversational content before and after a keyword is detected and extracting key words and phrases" refers to devices or technology that, when a keyword is found, analyzes the conversational content before and after the keyword and identifies key words and phrases.

[0415] "Means for recognizing emotions based on voice data" refers to devices or technologies that analyze characteristics of voice, such as tone and speed, to determine the speaker's emotions.

[0416] "Means for generating a search query based on the extracted words, phrases, and emotional information, and sending a query to a search engine" refers to a device or technology that creates a search query based on the words and phrases obtained through the analysis and the recognized emotional information, and performs an information search.

[0417] "Means for receiving and displaying search results from a search engine" refers to a device or technology that receives search results from a search engine and displays them in an easy-to-view format for the user.

[0418] This invention builds a system that integrates voice recognition and emotion recognition, centered around smart glasses worn by customer service staff. The system aims to respond quickly and accurately to customer requests and provide appropriate service according to the customer's emotions.

[0419] The server receives real-time voice data from the smart glasses' microphone and converts it into text using the Google Speech-to-Text API. This text data is then sent to the server and analyzed for keyword detection. If a keyword is detected, the server analyzes the surrounding conversation using a natural language processing engine to extract key words and phrases.

[0420] At the same time, the server uses the voice data to perform emotion recognition and obtain the customer's emotional information. Based on this emotional information and the extracted words and phrases, a search query is generated and sent to a search engine. The information received from the search engine is then displayed on the smart glasses' display, allowing the customer service staff to provide the customer with a quick and appropriate answer.

[0421] As a concrete example, consider a scenario in a consumer electronics retail store. If a customer asks, "Is this TV on sale?", the server detects this conversation, finds the keyword "sale" and analyzes the conversation before and after. At the same time, it recognizes the customer's emotions from the tone of their voice to determine whether they are anxious or angry about the question. Based on this, the server generates a search query, "TV sale now," and uses it to query the inventory management system, immediately displaying the results to the staff. The staff can quickly respond through the smart glasses, "Yes, it is currently on sale."

[0422] An example of a prompt sentence is, "By using the Google Speech-to-Text API to convert voice into text, and when specific keywords are detected, the system searches for information related to those keywords and displays it in real time on the staff's smart glasses," the system can generate appropriate search queries and provide accurate information in real time.

[0423] As a result, the on-site staff can understand the customer's emotions and quickly provide information that is in line with those emotions, thereby improving customer satisfaction.

[0424] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0425] Step 1:

[0426] The server acquires conversational voice from the microphone of the smart glasses. The acquired voice data is temporarily stored in a buffer in the smart glasses. The input is conversational voice and the output is voice data. This voice data is kept in the buffer for subsequent processing.

[0427] Step 2:

[0428] The device sends the audio data in the buffer to the speech recognition engine in real time, which uses the Google Speech-to-Text API to convert the audio data into text data. The input is audio data, and the output is text data converted from the audio.

[0429] Step 3:

[0430] The server analyzes the text data received from the speech recognition engine. Here, it detects set keywords (e.g., "special price" or "returns"). The input is the text data, and the output is the analysis results including the detected keywords.

[0431] Step 4:

[0432] When a keyword is detected, the server analyzes the conversation content before and after it, and uses a natural language processing engine to extract key words and phrases. The input is the text data and the detected keyword, and the output is the extracted words and phrases.

[0433] Step 5:

[0434] The server performs emotion recognition based on the voice data. It analyzes the tone and speed of the voice data to determine the customer's emotion (e.g., anger, joy, confusion). The input is the voice data, and the output is the analyzed emotional information.

[0435] Step 6:

[0436] The server generates a search query based on the extracted words and phrases and emotional information. Here, the search query is adjusted taking into account emotional information, such as when the customer is in a hurry. The input is the extracted words and phrases and emotional information, and the output is the generated search query.

[0437] Step 7:

[0438] The server queries the search engine using the generated search query. It sends the query as an HTTP request to the search engine's API. The input is the search query, and the output is the search results from the search engine.

[0439] Step 8:

[0440] The server receives search results from the search engine and sends them to the on-site staff's terminal, which then displays the search results on the smart glasses' display. The input is the search results from the search engine, and the output is the search results converted into a displayable information format.

[0441] Step 9:

[0442] The user (on-site staff) responds to the customer based on the displayed information. If more specific information is required, an additional question is asked, and the system receives feedback again to search for the required information. The input is the customer's voice asking the additional question and the search results from the system, and the output is an appropriate answer to the customer.

[0443] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0444] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0445] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0446] [Second embodiment]

[0447] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0448] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0449] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0450] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0451] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0452] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0453] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0454] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0455] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0456] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0457] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0458] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0459] Overview of the voice recognition system

[0460] This invention is a speech recognition system that enables on-site staff to quickly respond to customer questions while serving customers. This system captures conversational speech, performs real-time speech recognition, and automatically searches for and displays the necessary information.

[0461] Overview of program processing

[0462] 1. Acquiring audio data

[0463] Terminal: A microphone installed at the customer service counter constantly records conversations and collects real-time audio data, which is temporarily stored in a buffer within the system.

[0464] 2. Real-time speech recognition

[0465] Device: The acquired voice data is sent from the buffer to the speech recognition engine, which converts the voice data into text data using, for example, the Google Speech-to-Text API or a similar service.

[0466] Server: Receives the converted character data and saves it in a database with a timestamp.

[0467] 3. Keyword detection

[0468] Server: Analyzes the character data and detects the set keyword (e.g., "confirm") and stores the partial string up to the appearance of the keyword.

[0469] 4. Analysis of conversation content

[0470] Server: When a keyword is detected, the server extracts the surrounding conversation (usually within one minute) and sends the extracted text data to a natural language processing (NLP) engine to extract key words and phrases.

[0471] 5. Search Query Generation

[0472] Server: Generates a search query based on the extracted words and phrases. For example, a specific query such as "new account opening procedure" is generated.

[0473] 6. Contacting Search Engines

[0474] Server: Sends the generated query as an HTTP request to the search engine.

[0475] 7. Retrieving and Displaying Search Results

[0476] Server: Receives search results from the search engine and displays them on the terminals of field staff.

[0477] Terminal: The received search results are displayed on the on-site staff's display.

[0478] 8. Feedback Loops

[0479] User (field staff): Answers the customer based on the displayed information. If more specific information is needed, the system asks additional questions. The system receives feedback and searches for information again.

[0480] Specific examples

[0481] Scenario 1: Consultation regarding opening a bank account

[0482] 1. Audio Acquisition

[0483] Customer: "I'd like to open a new account and would like to confirm the procedure."

[0484] Device: Record this conversation and store the audio data in a buffer.

[0485] 2. Voice Recognition

[0486] On the device: Sends the audio data in the buffer to the Google Speech-to-Text API in real time.

[0487] Server: Receives the returned text data "I would like to open a new account and would like to confirm the procedure." and saves it in the database with a timestamp.

[0488] 3. Keyword detection

[0489] Server: Analyzes the text data and detects the keyword "confirm." At this point, the part "I would like to open a new account. What are the procedures?" is retained as the analysis target.

[0490] 4. Analysis of conversation content

[0491] Server: Uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[0492] 5. Search Query Generation

[0493] Server: Generate a specific search query: "new account opening procedure."

[0494] 6. Contacting Search Engines

[0495] Server: Sends the generated query as an HTTP request to the search engine's API.

[0496] 7. Retrieving and Displaying Search Results

[0497] Server: Sends search results received from the search engine to the terminals of field staff.

[0498] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[0499] 8. Feedback Loops

[0500] User (on-site staff): Answers the customer based on the displayed information. For example, "To open a new account, you need the following documents." If the customer has a more detailed question, the system detects the keywords again and searches for the necessary information.

[0501] This system allows field staff to efficiently obtain information and quickly answer customer questions, which is expected to reduce customer service time and improve customer satisfaction.

[0502] The processing flow will be explained below.

[0503] Step 1:

[0504] Terminal: A microphone installed at the customer service counter continuously records conversations and collects voice data in real time. The voice data is temporarily stored in a buffer within the system.

[0505] Step 2:

[0506] Terminal: The acquired voice data is sent from the buffer to the voice recognition engine, which converts the voice data into text data using, for example, the Google Speech-to-Text API.

[0507] Step 3:

[0508] Server: Receives converted text data from the speech recognition engine and stores it in a database with a timestamp.

[0509] Step 4:

[0510] Server: Analyzes the character data and detects the set keyword (e.g., "confirm") and stores the partial string up to the appearance of the keyword.

[0511] Step 5:

[0512] Server: When a keyword is detected, the conversation content before and after it (usually within one minute) is extracted.

[0513] Step 6:

[0514] Server: The extracted text data is sent to a natural language processing engine to extract key words and phrases.

[0515] Step 7:

[0516] Server: Generates a search query based on the extracted words and phrases. For example, it generates a specific query such as "open a new account."

[0517] Step 8:

[0518] Server: Sends the generated query as an HTTP request to the search engine's API.

[0519] Step 9:

[0520] Server: Receives search results from the search engine, formats them, and sends them to the terminals of field staff.

[0521] Step 10:

[0522] Terminal: The received search results are displayed on the on-site staff's display.

[0523] Step 11:

[0524] User (field staff): Responds to the customer based on the displayed information. If more specific information is required, the user asks additional questions. The system receives feedback and searches for information again based on this.

[0525] Example 1

[0526] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0527] In customer service, staff need a lot of information to respond to customer questions quickly and accurately. However, manually searching for information is time-consuming and can lead to a decrease in customer satisfaction. Furthermore, a noisy and multitasking environment on-site makes it difficult to respond quickly. Therefore, there is a need for a system that can recognize voice in real time and automatically present relevant information.

[0528] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0529] In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, means for analyzing the text data and detecting set keywords, means for, when a keyword is detected, retroactively analyzing the conversation content before and after the keyword and extracting key words and phrases, means for generating a search query based on the extracted words and phrases and making an inquiry to a search engine, means for receiving and displaying search results from the search engine, and means for searching for additional information based on the feedback. This enables on-site staff to respond to customer questions quickly and accurately and improve customer satisfaction.

[0530] "Conversational audio" refers to the verbal exchanges that take place between customers and on-site staff during customer service operations.

[0531] "Audio data" means a digital representation of speech collected using a device such as a microphone.

[0532] A "voice recognition engine" is software or a service that receives voice data as input and converts it into text-formatted character data.

[0533] "Character data" refers to text-format data converted by a voice recognition engine, and primarily refers to information expressed using characters (such as alphabets or kanji).

[0534] "Keywords" are words or phrases that are considered particularly important in the conversation and serve as triggers for the system to perform specific operations.

[0535] The "conversation content" is a series of words exchanged between the customer and the on-site staff, and is a group of text data containing specific information.

[0536] A "natural language processing engine" is an algorithm or software that analyzes text data and extracts key words and phrases.

[0537] A "search query" is a specific phrase or combination of words generated to ask a search engine.

[0538] A "search engine" is a system or software that searches the Internet or databases for relevant information based on an input search query.

[0539] "Search results" refers to the list or content of information that a search engine provides based on a search query.

[0540] "Feedback" refers to the process and results of field staff conducting additional information searches based on the information and results they receive from the system.

[0541] This invention is a speech recognition system that enables on-site staff to respond quickly and accurately to customer questions during customer service duties. This system captures conversational voices, performs real-time speech recognition, and automatically searches for and displays necessary information, significantly improving on-site efficiency.

[0542] 1. Hardware and Software Used

[0543] The system hardware configuration is as follows:

[0544] Microphone (to capture speech)

[0545] Terminal (to acquire audio data and perform initial processing)

[0546] Server (for data analysis, generating search queries, and communicating with search engines)

[0547] Display (to show results to field staff)

[0548] The system software configuration is as follows:

[0549] Google Speech-to-Text API (voice recognition engine)

[0550] NLP engines (for natural language processing, e.g. SpaCy)

[0551] Search engine APIs (for information searches, e.g., Google and Bing APIs)

[0552] 2. System processing overview

[0553] The process of this speech recognition system involves the following steps:

[0554] Acquiring audio data:

[0555] The device collects conversational audio in real time through a microphone and temporarily stores it in a buffer. This audio data is stored in PCM format.

[0556] Real-time speech recognition:

[0557] The device splits the audio data in the buffer into regular batches (e.g., every 5 seconds) and sends them to the Google Speech-to-Text API, which converts the audio into text data.

[0558] The server stores the received character data in a database with a timestamp.

[0559] Keyword detection and conversation analysis:

[0560] The server periodically scans new text data for set keywords, and if a keyword is found, sends the surrounding conversational content to an NLP engine to extract key words and phrases.

[0561] Search query generation and query:

[0562] The server then combines the extracted words and phrases to generate a search query, which it then sends as an HTTP request to the search engine's API, which then returns the search results.

[0563] Displaying search results:

[0564] The server displays the search results received from the search engine on the terminal's display, and the field staff responds to the customer based on the displayed information.

[0565] Feedback Loop:

[0566] The user (field staff) responds to the customer based on the displayed information. If more details are needed, the user provides feedback to the system and searches for additional information.

[0567] 3. Explanation of specific examples

[0568] Scenario 1: Consultation regarding opening a bank account

[0569] 1. Audio capture:

[0570] Customer: "I'd like to open a new account and would like to confirm the procedure."

[0571] The device records this conversation and stores the audio data in a buffer.

[0572] 2. Speech Recognition:

[0573] The device sends the buffered audio data to the Google Speech-to-Text API in real time.

[0574] The server receives the returned text data, "I would like to open a new account and would like to confirm the procedure," and stores it in the database with a timestamp.

[0575] 3. Keyword detection:

[0576] The server analyzes the text data and detects the keyword "confirmation." At this point, it retains the part "I would like to open a new account. What are the procedures?" as the analysis target.

[0577] 4. Conversation Analysis:

[0578] The server uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[0579] 5. Generating search queries:

[0580] The server generates a specific search query, "new account opening procedure," and sends it as an HTTP request to the search engine's API.

[0581] 6. Retrieving and displaying search results:

[0582] The server sends the search results received from the search engine to the terminal of the on-site staff and displays them on the screen.

[0583] The terminal then displays the search results on the on-site staff's screen, such as information about the documents and conditions required to open a new account.

[0584] 7. Feedback Loop:

[0585] The user (field staff) responds to the customer based on the displayed information. If the customer has a more detailed question, the system will again detect the keywords and search for the necessary information.

[0586] This system allows field staff to efficiently obtain information and quickly answer customer questions, which is expected to reduce customer service time and improve customer satisfaction.

[0587] Example prompt sentence:

[0588] I would like to open a new account and would like to confirm the procedure.

[0589] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0590] Step 1: Acquire audio data

[0591] The terminal captures conversational audio in real time from a microphone installed at the customer service counter. The input is the conversational audio between the customer and the front-line staff, which is converted into digital format and stored in a buffer in the terminal's memory. Specifically, the audio signal is converted from analog to digital format and continuously written to the buffer in PCM format.

[0592] Step 2: Real-time speech recognition

[0593] The device buffers the audio data in regular batches (for example, every 5 seconds). The input is audio data, which is sent to the Google Speech-to-Text API to be converted into text. At this stage, the audio data is analyzed by a speech recognition engine, and text data is generated as output. This text data is then sent from the device to the server.

[0594] Step 3: Storing Character Data

[0595] The server stores the text data received from the device in a database with a timestamp. The input is text data, and a timestamp is added during the process of saving it in the database. Specifically, the text data is stored in an SQLite database, and synchronization between voice and text is maintained.

[0596] Step 4: Keyword detection

[0597] The server periodically scans new text data to detect configured keywords. The input is character data, and it searches for specific keywords using regular expressions, etc. This detection process outputs the part containing the configured keyword. Specifically, when a keyword is detected, the conversation content before and after it is temporarily stored in memory.

[0598] Step 5: Analyzing the conversation

[0599] The server sends the conversational content, including one minute before and after the text where the keyword was detected, to a natural language processing (NLP) engine. The input is text data, which is analyzed by an NLP engine (e.g., SpaCy) to extract key words and phrases. The output is a list of extracted words and phrases. Specifically, the NLP engine tokenizes the text and identifies and extracts key words such as nouns and verbs.

[0600] Step 6: Generating a search query

[0601] The server generates a query based on the extracted words and phrases in a format that is easy for users to search. The input is a list of key words and phrases, and a specific search query is generated by string concatenation. The output is the search query. For example, a phrase such as "new account opening procedure" is generated.

[0602] Step 7: Contact the search engines

[0603] The server then sends the generated search query to the search engine's API. The input is the search query, which is queried to the search engine (e.g., Google or Bing) using an HTTP GET request. The output is the search results. Specifically, the query is appended to a URL and sent to the search engine, which returns the results in JSON format.

[0604] Step 8: Retrieving and Displaying Search Results

[0605] The server sends the search results received from the search engine to the terminal and displays them on the on-site staff's display. The input is the search results from the search engine, which are formatted for display on the terminal's display. The output is the information displayed on the on-site staff's display. Specifically, information such as "Documents and conditions required for opening a new account" is displayed.

[0606] Step 9: Feedback Loop

[0607] The user (on-site staff) responds to the customer based on the displayed information. If more specific information is required, they ask an additional question. The input is the customer's additional question, and the content of this question is picked up again from the microphone, and processing returns to step 1 for reprocessing. The output is updated search results and a new answer. Specifically, more detailed information is searched again based on the customer's additional question and displayed.

[0608] (Application example 1)

[0609] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0610] Conventional voice recognition systems were not sufficient for quickly obtaining necessary information during conversations with customers. Furthermore, search functions based on specific keywords alone sometimes made it difficult for on-site staff to accurately answer customer questions. As a result, customer satisfaction and response efficiency declined.

[0611] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0612] In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, means for analyzing the text data and detecting set keywords, means for retroactively analyzing the conversation content before and after a keyword is detected and extracting key words and phrases, means for generating a search query based on the extracted words and phrases and making an inquiry to a search engine, means for receiving search results from the search engine and displaying them on a user terminal, and means for performing another information search based on the feedback. This enables on-site staff to quickly and accurately answer customer questions.

[0613] "Conversational voice" refers to the voice exchanged between the user and the person answering the call.

[0614] "Means for acquiring" refers to a device or system for collecting audio data.

[0615] "Send to the speech recognition engine in real time" means that collected speech data is immediately transferred to the speech recognition engine.

[0616] "Means for converting to text data" refers to a process or device that converts audio data into text data.

[0617] "Analyzing character data" refers to the process of interpreting the converted text data and extracting necessary information.

[0618] "Means for detecting set keywords" refers to a system that identifies specific predefined words or phrases within text data.

[0619] "When a keyword is detected, the conversation content before and after it is analyzed" refers to analyzing the text before and after the detected keyword and extracting related content.

[0620] "Means for extracting key words and phrases" refers to a system that finds important words and phrases from the analyzed text data.

[0621] "Means for generating a search query and sending it to a search engine" refers to creating a search query based on the extracted words and phrases and sending that query to a search engine.

[0622] "Means for receiving search results from a search engine and displaying them on a user's device" refers to a system that receives responses from a search engine and displays that information on a user's device.

[0623] "Means of searching for information again based on feedback" refers to the process of generating a search query again based on additional questions or requests from the user to obtain new information.

[0624] "User devices" refer to devices used by customers and field staff, such as smartphones and tablets.

[0625] A "natural language processing engine" is a system that analyzes text data and extracts meaning and related information.

[0626] "General-purpose device" refers to a device that has a wide range of functions and is not limited to a specific use, such as a general-purpose smartphone or tablet.

[0627] The system for implementing this invention uses the following hardware and software. First, the hardware used includes a smartphone with a microphone for recording voice and a server for data processing. The software includes a Python program, a voice recognition library (e.g., the SpeechRecognition library), the Google Speech-to-Text API, and a natural language processing engine (e.g., an NLP engine for natural language processing).

[0628] The server uses a speech recognition engine to convert voice data into text data in real time. The converted text data is analyzed to detect set keywords. When a keyword is detected, the conversation content before and after it is analyzed to extract key words and phrases. A search query is generated based on these extracted words and phrases, and this search query is used as data to query a search engine.

[0629] The search results returned by the search engine are displayed on the user's smartphone. Here, "user device" refers to the device used by the customer or field staff, such as a smartphone or tablet. Furthermore, the system also incorporates a mechanism for re-searching information based on user feedback, enabling more accurate responses to questions.

[0630] For example, when a store clerk asks, "Is this item in stock?", the system instantly recognizes the keyword "in stock" and searches for and displays inventory information. This information is detailed information that customers want, and helps field staff respond quickly.

[0631] Example prompt sentence:

[0632] Input Sentence: Customer: "Is this item in stock?"

[0633] Example output: System: In response to the question "Is this item in stock?", the system recognizes the keyword "in stock" and searches for and displays stock information.

[0634] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0635] Step 1:

[0636] The user uses the microphone on their smartphone to record the conversation. This voice data is temporarily stored in a buffer within the system. The input is the conversation between the user and the customer, and the output is voice data.

[0637] Step 2:

[0638] The device sends the captured voice data in real time to a voice recognition engine (for example, Google Speech-to-Text API). This engine converts the voice data into text data. The input is voice data, and the output is text data.

[0639] Step 3:

[0640] The server receives the converted character data and stores it in a database with a timestamp. The input is character data, and the output is character data with a timestamp. This timestamp is used for later analysis.

[0641] Step 4:

[0642] The server analyzes the text data and detects the set keywords. For example, keywords such as "stock" and "price" are detected. The input is the text data, and the output is the detected keywords.

[0643] Step 5:

[0644] When a keyword is detected, the server analyzes the conversation content before and after the keyword and extracts key words and phrases. For example, key phrases such as "this product" and "is it in stock" are extracted. The input is the keyword and text data, and the output is the extracted key words and phrases.

[0645] Step 6:

[0646] The server generates a search query based on the extracted words and phrases. This query is intended to retrieve specific information. For example, a search query might be generated for "this product in stock." The input is the main word or phrase, and the output is the search query.

[0647] Step 7:

[0648] The server sends the generated search query to the search engine as an HTTP request and waits for a response from the search engine (e.g., Google). The input is the search query, and the output is the search results from the search engine.

[0649] Step 8:

[0650] The server receives search results from the search engine and displays them on the user's terminal. These results are information that the user provides to their customers. The input is the search results from the search engine, and the output is the information displayed on the user's terminal.

[0651] Step 9:

[0652] The user responds to the customer based on the displayed information. If the customer requires more detailed information, the user asks an additional question, and the system searches for information again based on the feedback. The input is the additional question from the customer, and the output is the information obtained again.

[0653] This process allows the system to provide quick and accurate answers to customer questions.

[0654] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0655] Overview of speech and emotion recognition systems

[0656] This invention is a speech and emotion recognition system that aims to provide more accurate service by not only enabling on-site staff to respond quickly to customer questions while serving customers, but also by recognizing customer emotions and utilizing that information. This system captures conversational voice, performs speech and emotion recognition in real time, and automatically searches for and displays the necessary information.

[0657] Overview of program processing

[0658] 1. Acquiring audio data

[0659] Terminal: A microphone installed at the customer service counter constantly records conversations, collecting voice data in real time. This voice data is temporarily stored in a buffer within the system.

[0660] 2. Real-time speech recognition

[0661] Terminal: The acquired voice data is sent from the buffer to the speech recognition engine, which converts the voice data into text data. For example, the Google Speech-to-Text API is used to convert voice to text.

[0662] Server: The character data received from the speech recognition engine is saved in a database with a timestamp.

[0663] 3. Keyword detection

[0664] Server: Analyzes the character data and detects the set keyword (e.g., "confirm") and stores the partial string up to the appearance of the keyword.

[0665] 4. Analysis of conversation content

[0666] Server: When a keyword is detected, the server extracts the surrounding conversation (usually within one minute) and sends the extracted text data to a natural language processing (NLP) engine to extract key words and phrases.

[0667] 5. Emotional Recognition

[0668] Device: Voice data is input into an emotion recognition engine to analyze the user's emotions. For example, it determines whether the user is angry or happy based on the tone and speed of the voice.

[0669] Server: Stores the user's emotional data based on the analyzed emotional information.

[0670] 6. Search Query Generation

[0671] Server: Generates a search query based on the extracted words and phrases and sentiment information. For example, "Open a new account" and if the user is angry, a quick response is required.

[0672] 7. Contacting Search Engines

[0673] Server: Sends the generated query as an HTTP request to the search engine's API.

[0674] 8. Retrieving and Displaying Search Results

[0675] Server: Receives search results from the search engine and sends them to the terminals of field staff. Search results also reflect emotional information, for example, prioritizing content that requires a quick response.

[0676] Terminal: The received search results are displayed on the on-site staff's display.

[0677] 9. Feedback Loops

[0678] User (on-site staff): Responds to the customer based on the displayed information. If more specific information is required, the system asks additional questions. The system receives feedback and searches for information again. During this process, the user's emotions are continuously monitored.

[0679] Specific examples

[0680] Scenario 1: Consultation regarding opening a bank account

[0681] 1. Audio Acquisition

[0682] Customer: "I'd like to open a new account and would like to confirm the procedure."

[0683] Device: Record this conversation and store the audio data in a buffer.

[0684] 2. Voice Recognition

[0685] On the device: Sends the audio data in the buffer to the Google Speech-to-Text API in real time.

[0686] Server: Receives the returned text data "I would like to open a new account and would like to confirm the procedure." and saves it in the database with a timestamp.

[0687] 3. Keyword detection

[0688] Server: Analyzes the text data and detects the keyword "confirm." At this point, the part "I would like to open a new account. What are the procedures?" is retained as the analysis target.

[0689] 4. Analysis of conversation content

[0690] Server: Uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[0691] 5. Emotional Recognition

[0692] On the device: An emotion recognition engine is used to determine the user's emotion. In this case, it is determined that the customer has a question.

[0693] 6. Search Query Generation

[0694] Server: Generate a specific search query, such as "open a new account," based on sentiment information. If the customer is in a hurry, a quick response is required.

[0695] 7. Contacting Search Engines

[0696] Server: Sends the generated query as an HTTP request to the search engine's API.

[0697] 8. Retrieving and Displaying Search Results

[0698] Server: Sends search results received from the search engine to the terminals of field staff. The results are displayed with priority given to items that require immediate attention.

[0699] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[0700] 9. Feedback Loops

[0701] User (on-site staff): Responds to the customer based on the displayed information. For example, "To open a new account, you need the following documents." If the customer has a more detailed question, the system will again detect keywords, search for the necessary information, and continuously monitor the customer's emotions.

[0702] This system allows on-site staff to efficiently obtain information and respond to customers with consideration for their feelings, which is expected to reduce customer service time and improve customer satisfaction.

[0703] The processing flow will be explained below.

[0704] Step 1:

[0705] Terminal: A microphone installed at the customer service counter continuously records conversations and collects voice data in real time. The voice data is temporarily stored in a buffer within the system.

[0706] Step 2:

[0707] Terminal: The acquired voice data is sent from the buffer to a voice recognition engine, for example, using the Google Speech-to-Text API to convert the voice data into text data.

[0708] Step 3:

[0709] Server: Receives converted text data from the speech recognition engine and stores it in a database with a timestamp.

[0710] Step 4:

[0711] Server: Analyzes the character data and detects the set keyword (e.g., "confirm"), and stores the substring up to the detected keyword.

[0712] Step 5:

[0713] Server: When a keyword is detected, the conversation content before and after it (usually within one minute) is extracted.

[0714] Step 6:

[0715] Server: The extracted text data is sent to a natural language processing (NLP) engine to extract key words and phrases.

[0716] Step 7:

[0717] Device: Voice data is input into an emotion recognition engine to analyze the user's emotions. For example, the emotional state is determined from the tone, speed, and pitch of the voice, and emotions such as joy, anger, and sadness are identified.

[0718] Step 8:

[0719] Server: Based on the analyzed emotional information, the user's emotional data is stored in a database.

[0720] Step 9:

[0721] Server: Generates search queries based on the extracted words and phrases and emotion information. For example, it generates a specific query such as "new account opening procedure" and sets a high priority if the user is angry.

[0722] Step 10:

[0723] Server: Sends the generated search query as an HTTP request to the search engine's API.

[0724] Step 11:

[0725] Server: Receives search results from the search engine, formats them, and sends them to the terminals of field staff. Search results also reflect emotional information, and priority is given to content that requires a quick response.

[0726] Step 12:

[0727] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[0728] Step 13:

[0729] User (on-site staff): Responds to the customer based on the displayed information. If more specific information is required, the system asks additional questions. The system receives feedback and searches for information again. During this process, the system also continuously monitors the user's emotions.

[0730] Specific examples

[0731] Scenario 1: Consultation regarding opening a bank account

[0732] 1. Audio Acquisition

[0733] Customer: "I'd like to open a new account and would like to confirm the procedure."

[0734] Device: Record this conversation and store the audio data in a buffer.

[0735] 2. Voice Recognition

[0736] On the device: Sends the audio data in the buffer to the Google Speech-to-Text API in real time.

[0737] Server: Receives the returned text data "I would like to open a new account and would like to confirm the procedure." and saves it in the database with a timestamp.

[0738] 3. Keyword detection

[0739] Server: Analyzes the text data and detects the keyword "confirm." At this point, the part "I would like to open a new account. What are the procedures?" is retained as the analysis target.

[0740] 4. Analysis of conversation content

[0741] Server: Uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[0742] 5. Emotional Recognition

[0743] On the device: An emotion recognition engine is used to determine the user's emotion. In this case, it is determined that the customer has a question.

[0744] 6. Search Query Generation

[0745] Server: Generate a specific search query, such as "open a new account," based on sentiment information. If the customer is in a hurry, a quick response is required.

[0746] 7. Contacting Search Engines

[0747] Server: Sends the generated query as an HTTP request to the search engine's API.

[0748] 8. Retrieving and Displaying Search Results

[0749] Server: Sends search results received from the search engine to the terminals of field staff. The results are displayed with priority given to items that require immediate attention.

[0750] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[0751] 9. Feedback Loops

[0752] User (on-site staff): Responds to the customer based on the displayed information. For example, "To open a new account, you need the following documents." If the customer has a more detailed question, the system will again detect keywords, search for the necessary information, and continuously monitor the customer's emotions.

[0753] This system allows on-site staff to efficiently obtain information and respond to customers with consideration for their feelings, which is expected to reduce customer service time and improve customer satisfaction.

[0754] Example 2

[0755] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0756] While serving customers, on-site staff need to be able to respond quickly to customer questions and accurately recognize their emotions to provide more appropriate service. However, conventional systems only use voice recognition and do not analyze emotions, making it difficult to improve customer satisfaction. In addition, the need to process large amounts of data in real time requires specialized hardware and advanced software, making on-site implementation and operation complicated.

[0757] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0758] In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, means for analyzing the text data and detecting set keywords, means for, when a keyword is detected, retroactively analyzing the conversation content before and after the keyword and extracting key words and phrases, means for analyzing the tone, pitch, and speed of the voice data and recognizing emotions, means for generating search queries based on the emotional information and the extracted words and phrases and making inquiries to a search engine, and means for receiving and displaying search results from the search engine. This enables quick and accurate responses to customer questions and the provision of services that take emotions into consideration.

[0759] "Conversational audio" refers to the audio information exchanged between customers and on-site staff during customer service.

[0760] "Audio data" refers to information that is a digital recording of conversational voice.

[0761] A "voice recognition engine" is software that analyzes voice data and converts it into text data.

[0762] "Character data" refers to text information generated from voice data by a voice recognition engine.

[0763] A "keyword" is a particularly important word or phrase in the character data to be analyzed.

[0764] A "natural language processing engine" is software that analyzes text data and extracts key words and phrases.

[0765] A "search query" is a string of characters or phrases generated to search for specific information.

[0766] A "search engine" is a system or software that searches for information based on a search query.

[0767] "Tone" is an attribute that represents the pitch and quality of sound in audio data.

[0768] "Pitch" is an attribute that represents the frequency of sound in audio data.

[0769] "Speed" is an attribute that indicates the speaking speed of the voice in the voice data.

[0770] "Emotion recognition" is the process of identifying a speaker's emotions by analyzing the tone, pitch, and rate of speech data.

[0771] "Field staff" refers to employees who handle customer service.

[0772] "Customer" means the person receiving the service.

[0773] "Display" means a device that visually displays search results or other information.

[0774] A "buffer" is a memory space for temporarily storing audio data.

[0775] This invention relates to a system that enables on-site staff to provide more appropriate service by quickly and accurately answering customer questions while serving customers and recognizing their emotions. The system collects conversational voices, performs real-time speech and emotion recognition, and automatically searches for and displays the necessary information.

[0776] Specifically, the device uses a microphone installed at the customer service counter to continuously record the conversation between the customer and the on-site staff, temporarily storing the audio data in a buffer, and then sends this audio data in real time to a speech recognition engine (for example, Google Speech-to-Text API) and converts it into text data.

[0777] The server then stores the acquired text data in a database with a timestamp. The stored text data is analyzed to detect set keywords (e.g., "confirmation"). If a keyword is detected, the server extracts the surrounding conversation content and uses a natural language processing (NLP) engine (e.g., spaCy or NLTK) to extract key words and phrases.

[0778] Furthermore, the device analyzes the acoustic characteristics of the voice data, such as tone, pitch, and speed, and identifies the user's emotions using an emotion recognition engine (e.g., IBM Watson Tone Analyzer). This emotion information is analyzed together with the text data, and the server performs processing according to the emotion.

[0779] The server generates a search query based on the extracted key words and phrases and the identified sentiment information. The generated search query is sent as an HTTP request to a search engine (e.g., a general-purpose search engine API). The search results from the search engine are received by the server, appropriately filtered, and then sent to the device.

[0780] The terminal displays the received search results on the on-site staff's screen, allowing them to obtain the necessary information in a timely manner and respond to customers quickly and efficiently.If additional questions arise, a feedback loop is created in which the keywords are detected again and the necessary information is searched for.

[0781] As a concrete example, consider a consultation about opening a bank account. When a customer says, "I'd like to open a new account, but I'd like to confirm the procedure," the system records this conversation and saves the voice data in a buffer. A speech recognition engine converts this voice data into text data, generating the text data, "I'd like to open a new account, but I'd like to confirm the procedure."

[0782] Next, the server detects the keyword "confirmation" and analyzes and extracts the conversation content before and after it, obtaining key words and phrases such as "new account" and "opening procedure." At the same time, the device performs emotion recognition to identify the customer's question. The server generates a search query for "new account opening procedure" and displays appropriate search results if a quick response is needed.

[0783] An example of a prompt sentence is, "I would like to know the procedure for opening a new account at the bank. What documents do I need?" By using this, the system can quickly search for the relevant information and provide it to the field staff, allowing the customer to be served appropriately.

[0784] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0785] Step 1: Acquire audio data

[0786] Input: Voice conversation between customer and on-site staff

[0787] Specific operation: The terminal uses a microphone installed at the customer service counter to continuously record the conversation between the customer and the on-site staff. The recorded audio data is temporarily stored in a buffer within the terminal.

[0788] Output: Buffered audio data

[0789] Step 2: Real-time speech recognition

[0790] Input: Buffered audio data

[0791] Specific operation: The device sends the acquired voice data to a voice recognition engine (e.g., Google Speech-to-Text API) in real time, and converts the voice data into text data.

[0792] Output: Text data returned from the speech recognition engine

[0793] Step 3: Storing Character Data

[0794] Input: Character data returned from the speech recognition engine

[0795] Specific operation: The server adds a timestamp to the received character data and saves it in the database.

[0796] Output: Character data stored in the database with a timestamp

[0797] Step 4: Keyword detection

[0798] Input: Character data stored in a database

[0799] Specific operation: The server analyzes the text data and detects the set keyword (e.g., "confirm"), triggering the detection of the keyword and saving the conversation content before and after the keyword.

[0800] Output: The part where the keyword was detected and the conversation content before and after it

[0801] Step 5: Analyzing the conversation

[0802] Input: Conversation content before and after the keyword

[0803] What it does: The server uses a natural language processing engine (e.g., spaCy or NLTK) to extract key words and phrases, including tokenization, part-of-speech tagging, and named entity recognition.

[0804] Output: Extracted key words and phrases

[0805] Step 6: Recognize emotions

[0806] Input: Buffered audio data

[0807] Specific operation: The device analyzes acoustic features such as tone, pitch, and speed of the voice data and uses an emotion recognition engine (e.g., IBM Watson Tone Analyzer) to identify the user's emotion.

[0808] Output: Identified emotion information

[0809] Step 7: Generating a search query

[0810] Input: extracted key words and phrases, sentiment information

[0811] Specific operation: The server fuses the extracted words and phrases with sentiment information to generate search queries.

[0812] Output: Generated search query

[0813] Step 8: Contact the search engines

[0814] Input: Generated search query

[0815] Specific operation: The server creates and sends an HTTP request to send the generated query to the search engine's API.

[0816] Output: HTTP request to the search engine's API

[0817] Step 9: Retrieving and Displaying Search Results

[0818] Input: Search results from a search engine

[0819] Specific operation: The server analyzes search results from the search engine, filters and prioritizes them appropriately, and then sends them to the on-site staff's terminal. The terminal then displays the received search results on the on-site staff's display.

[0820] Output: Search results displayed on the device display

[0821] Step 10: Feedback Loop

[0822] Input: Search results displayed on the device display, additional questions from the customer

[0823] Specific operation: The user (on-site staff) answers the customer based on the displayed information. If the customer has a more detailed question, the system will detect keywords again, search for the necessary information, and continuously monitor the customer's emotions.

[0824] Output: Updated search results and continuously monitored sentiment information

[0825] (Application example 2)

[0826] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0827] When on-site staff deal with customers, they are required to respond to their questions quickly and accurately. Furthermore, understanding customer emotions and providing appropriate responses leads to improved customer satisfaction. However, with conventional systems, it has been difficult to accurately recognize customer emotions and provide information in real time. Therefore, there is a need for a system that integrates emotion recognition and voice recognition to provide appropriate services in real time that take customer emotions into consideration.

[0828] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into character data, means for analyzing the character data and detecting set keywords, means for recognizing emotions based on the voice data, means for generating a search query based on the extracted words, phrases, and emotion information and making an inquiry to a search engine, and means for receiving and displaying search results from the search engine. This enables field staff to understand customer emotions and quickly provide information that is appropriate for those emotions.

[0829] "Means for acquiring conversational audio" refers to devices or technologies for recording conversations between customers and wait staff in real time and collecting audio data.

[0830] "Means for transmitting voice data to a voice recognition engine in real time and converting it into text data" refers to a device or technology that immediately sends acquired voice data to a voice recognition engine and converts it from voice to text data.

[0831] "Means for analyzing character data and detecting set keywords" refers to a device or technology that analyzes the converted character data and finds specific keywords that have been set in advance.

[0832] "Means for analyzing the conversational content before and after a keyword is detected and extracting key words and phrases" refers to devices or technology that, when a keyword is found, analyzes the conversational content before and after the keyword and identifies key words and phrases.

[0833] "Means for recognizing emotions based on voice data" refers to devices or technologies that analyze characteristics of voice, such as tone and speed, to determine the speaker's emotions.

[0834] "Means for generating a search query based on the extracted words, phrases, and emotional information, and sending a query to a search engine" refers to a device or technology that creates a search query based on the words and phrases obtained through the analysis and the recognized emotional information, and performs an information search.

[0835] "Means for receiving and displaying search results from a search engine" refers to a device or technology that receives search results from a search engine and displays them in an easy-to-view format for the user.

[0836] This invention builds a system that integrates voice recognition and emotion recognition, centered around smart glasses worn by customer service staff. The system aims to respond quickly and accurately to customer requests and provide appropriate service according to the customer's emotions.

[0837] The server receives real-time voice data from the smart glasses' microphone and converts it into text using the Google Speech-to-Text API. This text data is then sent to the server and analyzed for keyword detection. If a keyword is detected, the server analyzes the surrounding conversation using a natural language processing engine to extract key words and phrases.

[0838] At the same time, the server uses the voice data to perform emotion recognition and obtain the customer's emotional information. Based on this emotional information and the extracted words and phrases, a search query is generated and sent to a search engine. The information received from the search engine is then displayed on the smart glasses' display, allowing the customer service staff to provide the customer with a quick and appropriate answer.

[0839] As a concrete example, consider a scenario in a consumer electronics retail store. If a customer asks, "Is this TV on sale?", the server detects this conversation, finds the keyword "sale" and analyzes the conversation before and after. At the same time, it recognizes the customer's emotions from the tone of their voice to determine whether they are anxious or angry about the question. Based on this, the server generates a search query, "TV sale now," and uses it to query the inventory management system, immediately displaying the results to the staff. The staff can quickly respond through the smart glasses, "Yes, it is currently on sale."

[0840] An example of a prompt sentence is, "By using the Google Speech-to-Text API to convert voice into text, and when specific keywords are detected, the system searches for information related to those keywords and displays it in real time on the staff's smart glasses," the system can generate appropriate search queries and provide accurate information in real time.

[0841] As a result, the on-site staff can understand the customer's emotions and quickly provide information that is in line with those emotions, thereby improving customer satisfaction.

[0842] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0843] Step 1:

[0844] The server acquires conversational voice from the microphone of the smart glasses. The acquired voice data is temporarily stored in a buffer in the smart glasses. The input is conversational voice and the output is voice data. This voice data is kept in the buffer for subsequent processing.

[0845] Step 2:

[0846] The device sends the audio data in the buffer to the speech recognition engine in real time, which uses the Google Speech-to-Text API to convert the audio data into text data. The input is audio data, and the output is text data converted from the audio.

[0847] Step 3:

[0848] The server analyzes the text data received from the speech recognition engine. Here, it detects set keywords (e.g., "special price" or "returns"). The input is the text data, and the output is the analysis results including the detected keywords.

[0849] Step 4:

[0850] When a keyword is detected, the server analyzes the conversation content before and after it, and uses a natural language processing engine to extract key words and phrases. The input is the text data and the detected keyword, and the output is the extracted words and phrases.

[0851] Step 5:

[0852] The server performs emotion recognition based on the voice data. It analyzes the tone and speed of the voice data to determine the customer's emotion (e.g., anger, joy, confusion). The input is the voice data, and the output is the analyzed emotional information.

[0853] Step 6:

[0854] The server generates a search query based on the extracted words and phrases and emotional information. Here, the search query is adjusted taking into account emotional information, such as when the customer is in a hurry. The input is the extracted words and phrases and emotional information, and the output is the generated search query.

[0855] Step 7:

[0856] The server queries the search engine using the generated search query. It sends the query as an HTTP request to the search engine's API. The input is the search query, and the output is the search results from the search engine.

[0857] Step 8:

[0858] The server receives search results from the search engine and sends them to the on-site staff's terminal, which then displays the search results on the smart glasses' display. The input is the search results from the search engine, and the output is the search results converted into a displayable information format.

[0859] Step 9:

[0860] The user (on-site staff) responds to the customer based on the displayed information. If more specific information is required, an additional question is asked, and the system receives feedback again to search for the required information. The input is the customer's voice asking the additional question and the search results from the system, and the output is an appropriate answer to the customer.

[0861] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0862] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0863] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0864] [Third embodiment]

[0865] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0866] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0867] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0868] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0869] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0870] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0871] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0872] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0873] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0874] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0875] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0876] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0877] Overview of the voice recognition system

[0878] This invention is a speech recognition system that enables on-site staff to quickly respond to customer questions while serving customers. This system captures conversational speech, performs real-time speech recognition, and automatically searches for and displays the necessary information.

[0879] Overview of program processing

[0880] 1. Acquiring audio data

[0881] Terminal: A microphone installed at the customer service counter constantly records conversations and collects real-time audio data, which is temporarily stored in a buffer within the system.

[0882] 2. Real-time speech recognition

[0883] Device: The acquired voice data is sent from the buffer to the speech recognition engine, which converts the voice data into text data using, for example, the Google Speech-to-Text API or a similar service.

[0884] Server: Receives the converted character data and saves it in a database with a timestamp.

[0885] 3. Keyword detection

[0886] Server: Analyzes the character data and detects the set keyword (e.g., "confirm") and stores the partial string up to the appearance of the keyword.

[0887] 4. Analysis of conversation content

[0888] Server: When a keyword is detected, the server extracts the surrounding conversation (usually within one minute) and sends the extracted text data to a natural language processing (NLP) engine to extract key words and phrases.

[0889] 5. Generating Search Queries

[0890] Server: Generates a search query based on the extracted words and phrases. For example, a specific query such as "new account opening procedure" is generated.

[0891] 6. Contacting Search Engines

[0892] Server: Sends the generated query as an HTTP request to the search engine.

[0893] 7. Retrieving and Displaying Search Results

[0894] Server: Receives search results from the search engine and displays them on the terminals of field staff.

[0895] Terminal: The received search results are displayed on the on-site staff's display.

[0896] 8. Feedback Loops

[0897] User (field staff): Answers the customer based on the displayed information. If more specific information is needed, the system asks additional questions. The system receives feedback and searches for information again.

[0898] Specific examples

[0899] Scenario 1: Consultation regarding opening a bank account

[0900] 1. Audio Acquisition

[0901] Customer: "I'd like to open a new account and would like to confirm the procedure."

[0902] Device: Record this conversation and store the audio data in a buffer.

[0903] 2. Voice Recognition

[0904] On the device: Sends the audio data in the buffer to the Google Speech-to-Text API in real time.

[0905] Server: Receives the returned text data "I would like to open a new account and would like to confirm the procedure." and saves it in the database with a timestamp.

[0906] 3. Keyword detection

[0907] Server: Analyzes the text data and detects the keyword "confirm." At this point, the part "I would like to open a new account. What are the procedures?" is retained as the analysis target.

[0908] 4. Analysis of conversation content

[0909] Server: Uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[0910] 5. Generating Search Queries

[0911] Server: Generate a specific search query: "new account opening procedure."

[0912] 6. Contacting Search Engines

[0913] Server: Sends the generated query as an HTTP request to the search engine's API.

[0914] 7. Retrieving and Displaying Search Results

[0915] Server: Sends search results received from the search engine to the terminals of field staff.

[0916] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[0917] 8. Feedback Loops

[0918] User (on-site staff): Answers the customer based on the displayed information. For example, "To open a new account, you need the following documents." If the customer has a more detailed question, the system detects the keywords again and searches for the necessary information.

[0919] This system allows field staff to efficiently obtain information and quickly answer customer questions, which is expected to reduce customer service time and improve customer satisfaction.

[0920] The processing flow will be explained below.

[0921] Step 1:

[0922] Terminal: A microphone installed at the customer service counter continuously records conversations and collects voice data in real time. The voice data is temporarily stored in a buffer within the system.

[0923] Step 2:

[0924] Terminal: The acquired voice data is sent from the buffer to the voice recognition engine, which converts the voice data into text data using, for example, the Google Speech-to-Text API.

[0925] Step 3:

[0926] Server: Receives converted text data from the speech recognition engine and stores it in a database with a timestamp.

[0927] Step 4:

[0928] Server: Analyzes the character data and detects the set keyword (e.g., "confirm") and stores the partial string up to the appearance of the keyword.

[0929] Step 5:

[0930] Server: When a keyword is detected, the conversation content before and after it (usually within one minute) is extracted.

[0931] Step 6:

[0932] Server: The extracted text data is sent to a natural language processing engine to extract key words and phrases.

[0933] Step 7:

[0934] Server: Generates a search query based on the extracted words and phrases. For example, it generates a specific query such as "open a new account."

[0935] Step 8:

[0936] Server: Sends the generated query as an HTTP request to the search engine's API.

[0937] Step 9:

[0938] Server: Receives search results from the search engine, formats them, and sends them to the terminals of field staff.

[0939] Step 10:

[0940] Terminal: The received search results are displayed on the on-site staff's display.

[0941] Step 11:

[0942] User (field staff): Responds to the customer based on the displayed information. If more specific information is required, the user asks additional questions. The system receives feedback and searches for information again based on this.

[0943] Example 1

[0944] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0945] In customer service, staff need a lot of information to respond to customer questions quickly and accurately. However, manually searching for information is time-consuming and can lead to a decrease in customer satisfaction. Furthermore, a noisy and multitasking environment on-site makes it difficult to respond quickly. Therefore, there is a need for a system that can recognize voice in real time and automatically present relevant information.

[0946] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0947] In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, means for analyzing the text data and detecting set keywords, means for retroactively analyzing the conversation content before and after a keyword is detected and extracting key words and phrases, means for generating a search query based on the extracted words and phrases and making an inquiry to a search engine, means for receiving and displaying search results from the search engine, and means for searching for additional information based on the feedback. This enables on-site staff to respond to customer questions quickly and accurately and improve customer satisfaction.

[0948] "Conversational audio" refers to the verbal exchanges that take place between customers and on-site staff during customer service.

[0949] "Audio data" means a digital representation of speech collected using a device such as a microphone.

[0950] A "voice recognition engine" is software or a service that receives voice data as input and converts it into text-based character data.

[0951] "Character data" refers to text-format data converted by a voice recognition engine, and primarily refers to information expressed using characters (such as alphabets or kanji).

[0952] "Keywords" are words or phrases that are considered particularly important in the conversation and serve as triggers for the system to perform specific operations.

[0953] The "conversation content" is a series of words exchanged between the customer and the on-site staff, and is a group of text data containing specific information.

[0954] A "natural language processing engine" is an algorithm or software that analyzes text data and extracts key words and phrases.

[0955] A "search query" is a specific phrase or combination of words generated to ask a search engine.

[0956] A "search engine" is a system or software that searches the Internet and databases for relevant information based on an input search query.

[0957] "Search results" refers to the list or content of information that a search engine provides based on a search query.

[0958] "Feedback" refers to the process and results of field staff conducting additional information searches based on the information and results they receive from the system.

[0959] This invention is a speech recognition system that enables on-site staff to respond quickly and accurately to customer questions during customer service duties. This system captures conversational voices, performs real-time speech recognition, and automatically searches for and displays necessary information, significantly improving on-site efficiency.

[0960] 1. Hardware and Software Used

[0961] The system hardware configuration is as follows:

[0962] Microphone (to capture speech)

[0963] Terminal (to acquire audio data and perform initial processing)

[0964] Server (for data analysis, generating search queries, and communicating with search engines)

[0965] Display (to show results to field staff)

[0966] The system software configuration is as follows:

[0967] Google Speech-to-Text API (voice recognition engine)

[0968] NLP engines (for natural language processing, e.g. SpaCy)

[0969] Search engine APIs (for information searches, e.g., Google and Bing APIs)

[0970] 2. System processing overview

[0971] The process of this speech recognition system involves the following steps:

[0972] Acquiring audio data:

[0973] The device collects conversational audio in real time through a microphone and temporarily stores it in a buffer. This audio data is stored in PCM format.

[0974] Real-time speech recognition:

[0975] The device splits the audio data in the buffer into regular batches (e.g., every 5 seconds) and sends them to the Google Speech-to-Text API, which converts the audio into text data.

[0976] The server stores the received character data in a database with a timestamp.

[0977] Keyword detection and conversation analysis:

[0978] The server periodically scans new text data for set keywords, and if a keyword is found, sends the surrounding conversational content to an NLP engine to extract key words and phrases.

[0979] Search query generation and query:

[0980] The server then combines the extracted words and phrases to generate a search query, which it then sends as an HTTP request to the search engine's API, which then returns the search results.

[0981] Displaying search results:

[0982] The server displays the search results received from the search engine on the terminal's display, and the field staff responds to the customer based on the displayed information.

[0983] Feedback Loop:

[0984] The user (field staff) responds to the customer based on the displayed information. If more details are needed, the user provides feedback to the system and searches for additional information.

[0985] 3. Explanation of specific examples

[0986] Scenario 1: Consultation regarding opening a bank account

[0987] 1. Audio capture:

[0988] Customer: "I'd like to open a new account and would like to confirm the procedure."

[0989] The device records this conversation and stores the audio data in a buffer.

[0990] 2. Speech Recognition:

[0991] The device sends the buffered audio data to the Google Speech-to-Text API in real time.

[0992] The server receives the returned text data, "I would like to open a new account and would like to confirm the procedure," and stores it in the database with a timestamp.

[0993] 3. Keyword detection:

[0994] The server analyzes the text data and detects the keyword "confirmation." At this point, it retains the part "I would like to open a new account. What are the procedures?" as the analysis target.

[0995] 4. Conversation Analysis:

[0996] The server uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[0997] 5. Generating search queries:

[0998] The server generates a specific search query, "new account opening procedure," and sends it as an HTTP request to the search engine's API.

[0999] 6. Retrieving and displaying search results:

[1000] The server sends the search results received from the search engine to the terminal of the on-site staff and displays them on the screen.

[1001] The terminal then displays the search results on the on-site staff's screen, such as information about the documents and conditions required to open a new account.

[1002] 7. Feedback Loop:

[1003] The user (field staff) responds to the customer based on the displayed information. If the customer has a more detailed question, the system will again detect the keywords and search for the necessary information.

[1004] This system allows field staff to efficiently obtain information and quickly answer customer questions, which is expected to reduce customer service time and improve customer satisfaction.

[1005] Example prompt sentence:

[1006] I would like to open a new account and would like to confirm the procedure.

[1007] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1008] Step 1: Acquire audio data

[1009] The terminal captures conversational audio in real time from a microphone installed at the customer service counter. The input is the conversational audio between the customer and the front-line staff, which is converted into digital format and stored in a buffer in the terminal's memory. Specifically, the audio signal is converted from analog to digital format and continuously written to the buffer in PCM format.

[1010] Step 2: Real-time speech recognition

[1011] The device buffers the audio data in regular batches (for example, every 5 seconds). The input is audio data, which is sent to the Google Speech-to-Text API to be converted into text. At this stage, the audio data is analyzed by a speech recognition engine, and text data is generated as output. This text data is then sent from the device to the server.

[1012] Step 3: Storing Character Data

[1013] The server stores the text data received from the device in a database with a timestamp. The input is text data, and a timestamp is added during the process of saving it in the database. Specifically, the text data is stored in an SQLite database, and synchronization between voice and text is maintained.

[1014] Step 4: Keyword detection

[1015] The server periodically scans new text data to detect configured keywords. The input is character data, and it searches for specific keywords using regular expressions, etc. This detection process outputs the part containing the configured keyword. Specifically, when a keyword is detected, the conversation content before and after it is temporarily stored in memory.

[1016] Step 5: Analyzing the conversation

[1017] The server sends the conversational content, including one minute before and after the text where the keyword was detected, to a natural language processing (NLP) engine. The input is text data, which is analyzed by an NLP engine (e.g., SpaCy) to extract key words and phrases. The output is a list of extracted words and phrases. Specifically, the NLP engine tokenizes the text and identifies and extracts key words such as nouns and verbs.

[1018] Step 6: Generating a search query

[1019] The server generates a query based on the extracted words and phrases in a format that is easy for users to search. The input is a list of key words and phrases, and a specific search query is generated by string concatenation. The output is the search query. For example, a phrase such as "new account opening procedure" is generated.

[1020] Step 7: Contact the search engines

[1021] The server then sends the generated search query to the search engine's API. The input is the search query, which is queried to the search engine (e.g., Google or Bing) using an HTTP GET request. The output is the search results. Specifically, the query is appended to a URL and sent to the search engine, which returns the results in JSON format.

[1022] Step 8: Retrieving and Displaying Search Results

[1023] The server sends the search results received from the search engine to the terminal and displays them on the on-site staff's display. The input is the search results from the search engine, which are formatted for display on the terminal's display. The output is the information displayed on the on-site staff's display. Specifically, information such as "Documents and conditions required for opening a new account" is displayed.

[1024] Step 9: Feedback Loop

[1025] The user (on-site staff) responds to the customer based on the displayed information. If more specific information is required, they ask an additional question. The input is the customer's additional question, and the content of this question is picked up again from the microphone, and processing returns to step 1 for reprocessing. The output is updated search results and a new answer. Specifically, more detailed information is searched again based on the customer's additional question and displayed.

[1026] (Application example 1)

[1027] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1028] Conventional voice recognition systems were not sufficient for quickly obtaining necessary information during conversations with customers. Furthermore, search functions based on specific keywords alone sometimes made it difficult for on-site staff to accurately answer customer questions. As a result, customer satisfaction and response efficiency declined.

[1029] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1030] In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, means for analyzing the text data and detecting set keywords, means for retroactively analyzing the conversation content before and after a keyword is detected and extracting key words and phrases, means for generating a search query based on the extracted words and phrases and making an inquiry to a search engine, means for receiving search results from the search engine and displaying them on a user terminal, and means for performing another information search based on the feedback. This enables on-site staff to quickly and accurately answer customer questions.

[1031] "Conversational voice" refers to the voice exchanged between the user and the person answering the call.

[1032] "Means for acquiring" refers to a device or system for collecting audio data.

[1033] "Send to the speech recognition engine in real time" means that collected speech data is immediately transferred to the speech recognition engine.

[1034] "Means for converting to text data" refers to a process or device that converts audio data into text data.

[1035] "Analyzing character data" refers to the process of interpreting the converted text data and extracting necessary information.

[1036] "Means for detecting set keywords" refers to a system that identifies specific predefined words or phrases within text data.

[1037] "When a keyword is detected, the conversation content before and after it is analyzed" refers to analyzing the text before and after the detected keyword and extracting related content.

[1038] "Means for extracting key words and phrases" refers to a system that finds important words and phrases from the analyzed text data.

[1039] "Means for generating a search query and sending it to a search engine" refers to creating a search query based on the extracted words and phrases and sending that query to a search engine.

[1040] "Means for receiving search results from a search engine and displaying them on a user's device" refers to a system that receives responses from a search engine and displays that information on a user's device.

[1041] "Means of searching for information again based on feedback" refers to the process of generating a search query again based on additional questions or requests from the user to obtain new information.

[1042] "User devices" refer to devices used by customers and field staff, such as smartphones and tablets.

[1043] A "natural language processing engine" is a system that analyzes text data and extracts meaning and related information.

[1044] "General-purpose device" refers to a device that has a wide range of functions and is not limited to a specific use, such as a general-purpose smartphone or tablet.

[1045] The system for implementing this invention uses the following hardware and software. First, the hardware used includes a smartphone with a microphone for recording voice and a server for data processing. The software includes a Python program, a voice recognition library (e.g., the SpeechRecognition library), the Google Speech-to-Text API, and a natural language processing engine (e.g., an NLP engine for natural language processing).

[1046] The server uses a speech recognition engine to convert voice data into text data in real time. The converted text data is analyzed to detect set keywords. When a keyword is detected, the conversation content before and after it is analyzed to extract key words and phrases. A search query is generated based on these extracted words and phrases, and this search query is used as data to query a search engine.

[1047] The search results returned by the search engine are displayed on the user's smartphone. Here, "user device" refers to the device used by the customer or field staff, such as a smartphone or tablet. Furthermore, the system also incorporates a mechanism for re-searching information based on user feedback, enabling more accurate responses to questions.

[1048] For example, when a store clerk asks, "Is this item in stock?", the system instantly recognizes the keyword "in stock" and searches for and displays inventory information. This information is detailed information that customers want, and helps field staff respond quickly.

[1049] Example prompt sentence:

[1050] Input Sentence: Customer: "Is this item in stock?"

[1051] Example output: System: In response to the question "Is this item in stock?", the system recognizes the keyword "in stock" and searches for and displays stock information.

[1052] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1053] Step 1:

[1054] The user uses the microphone on their smartphone to record the conversation. This voice data is temporarily stored in a buffer within the system. The input is the conversation between the user and the customer, and the output is voice data.

[1055] Step 2:

[1056] The device sends the captured voice data in real time to a voice recognition engine (for example, Google Speech-to-Text API). This engine converts the voice data into text data. The input is voice data, and the output is text data.

[1057] Step 3:

[1058] The server receives the converted character data and stores it in a database with a timestamp. The input is character data, and the output is character data with a timestamp. This timestamp is used for later analysis.

[1059] Step 4:

[1060] The server analyzes the text data and detects the set keywords. For example, keywords such as "stock" and "price" are detected. The input is the text data, and the output is the detected keywords.

[1061] Step 5:

[1062] When a keyword is detected, the server analyzes the conversation content before and after the keyword and extracts key words and phrases. For example, key phrases such as "this product" and "is it in stock" are extracted. The input is the keyword and text data, and the output is the extracted key words and phrases.

[1063] Step 6:

[1064] The server generates a search query based on the extracted words and phrases. This query is intended to retrieve specific information. For example, a search query might be generated for "this product in stock." The input is the main word or phrase, and the output is the search query.

[1065] Step 7:

[1066] The server sends the generated search query to the search engine as an HTTP request and waits for a response from the search engine (e.g., Google). The input is the search query, and the output is the search results from the search engine.

[1067] Step 8:

[1068] The server receives search results from the search engine and displays them on the user's terminal. These results are information that the user provides to their customers. The input is the search results from the search engine, and the output is the information displayed on the user's terminal.

[1069] Step 9:

[1070] The user responds to the customer based on the displayed information. If the customer requires more detailed information, the user asks an additional question, and the system searches for information again based on the feedback. The input is the additional question from the customer, and the output is the information obtained again.

[1071] This process allows the system to provide quick and accurate answers to customer questions.

[1072] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1073] Overview of speech and emotion recognition systems

[1074] This invention is a speech and emotion recognition system that aims to provide more accurate service by not only enabling on-site staff to respond quickly to customer questions while serving customers, but also by recognizing customer emotions and utilizing that information. This system captures conversational voice, performs speech and emotion recognition in real time, and automatically searches for and displays the necessary information.

[1075] Overview of program processing

[1076] 1. Acquiring audio data

[1077] Terminal: A microphone installed at the customer service counter constantly records conversations, collecting voice data in real time. This voice data is temporarily stored in a buffer within the system.

[1078] 2. Real-time speech recognition

[1079] Terminal: The acquired voice data is sent from the buffer to the speech recognition engine, which converts the voice data into text data. For example, the Google Speech-to-Text API is used to convert voice to text.

[1080] Server: The character data received from the speech recognition engine is saved in a database with a timestamp.

[1081] 3. Keyword detection

[1082] Server: Analyzes the character data and detects the set keyword (e.g., "confirm") and stores the partial string up to the appearance of the keyword.

[1083] 4. Analysis of conversation content

[1084] Server: When a keyword is detected, the server extracts the surrounding conversation (usually within one minute) and sends the extracted text data to a natural language processing (NLP) engine to extract key words and phrases.

[1085] 5. Emotional Recognition

[1086] Device: Voice data is input into an emotion recognition engine to analyze the user's emotions. For example, it determines whether the user is angry or happy based on the tone and speed of the voice.

[1087] Server: Stores the user's emotional data based on the analyzed emotional information.

[1088] 6. Search Query Generation

[1089] Server: Generates a search query based on the extracted words and phrases and sentiment information. For example, "Open a new account" and if the user is angry, a quick response is required.

[1090] 7. Contacting Search Engines

[1091] Server: Sends the generated query as an HTTP request to the search engine's API.

[1092] 8. Retrieving and Displaying Search Results

[1093] Server: Receives search results from the search engine and sends them to the terminals of field staff. Search results also reflect emotional information, for example, prioritizing content that requires a quick response.

[1094] Terminal: The received search results are displayed on the on-site staff's display.

[1095] 9. Feedback Loops

[1096] User (on-site staff): Responds to the customer based on the displayed information. If more specific information is required, the system asks additional questions. The system receives feedback and searches for information again. During this process, the user's emotions are continuously monitored.

[1097] Specific examples

[1098] Scenario 1: Consultation regarding opening a bank account

[1099] 1. Audio Acquisition

[1100] Customer: "I'd like to open a new account and would like to confirm the procedure."

[1101] Device: Record this conversation and store the audio data in a buffer.

[1102] 2. Voice Recognition

[1103] On the device: Sends the audio data in the buffer to the Google Speech-to-Text API in real time.

[1104] Server: Receives the returned text data "I would like to open a new account and would like to confirm the procedure." and saves it in the database with a timestamp.

[1105] 3. Keyword detection

[1106] Server: Analyzes the text data and detects the keyword "confirm." At this point, the part "I would like to open a new account. What are the procedures?" is retained as the analysis target.

[1107] 4. Analysis of conversation content

[1108] Server: Uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[1109] 5. Emotional Recognition

[1110] On the device: An emotion recognition engine is used to determine the user's emotion. In this case, it is determined that the customer has a question.

[1111] 6. Search Query Generation

[1112] Server: Generate a specific search query, such as "open a new account," based on sentiment information. If the customer is in a hurry, a quick response is required.

[1113] 7. Contacting Search Engines

[1114] Server: Sends the generated query as an HTTP request to the search engine's API.

[1115] 8. Retrieving and Displaying Search Results

[1116] Server: Sends search results received from the search engine to the terminals of field staff. The results are displayed with priority given to items that require immediate attention.

[1117] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[1118] 9. Feedback Loops

[1119] User (on-site staff): Responds to the customer based on the displayed information. For example, "To open a new account, you need the following documents." If the customer has a more detailed question, the system again detects keywords, searches for the necessary information, and continuously monitors the customer's emotions.

[1120] This system allows on-site staff to efficiently obtain information and respond to customers with consideration for their feelings, which is expected to reduce customer service time and improve customer satisfaction.

[1121] The processing flow will be explained below.

[1122] Step 1:

[1123] Terminal: A microphone installed at the customer service counter continuously records conversations and collects voice data in real time. The voice data is temporarily stored in a buffer within the system.

[1124] Step 2:

[1125] Terminal: The acquired voice data is sent from the buffer to a voice recognition engine, for example, using the Google Speech-to-Text API to convert the voice data into text data.

[1126] Step 3:

[1127] Server: Receives converted text data from the speech recognition engine and stores it in a database with a timestamp.

[1128] Step 4:

[1129] Server: Analyzes the character data and detects the set keyword (e.g., "confirm"), and stores the substring up to the detected keyword.

[1130] Step 5:

[1131] Server: When a keyword is detected, the conversation content before and after it (usually within one minute) is extracted.

[1132] Step 6:

[1133] Server: The extracted text data is sent to a natural language processing (NLP) engine to extract key words and phrases.

[1134] Step 7:

[1135] Device: Voice data is input into an emotion recognition engine to analyze the user's emotions. For example, the emotional state is determined from the tone, speed, and pitch of the voice, and emotions such as joy, anger, and sadness are identified.

[1136] Step 8:

[1137] Server: Based on the analyzed emotional information, the user's emotional data is stored in a database.

[1138] Step 9:

[1139] Server: Generates search queries based on the extracted words and phrases and emotion information. For example, it generates a specific query such as "new account opening procedure" and sets a high priority if the user is angry.

[1140] Step 10:

[1141] Server: Sends the generated search query as an HTTP request to the search engine's API.

[1142] Step 11:

[1143] Server: Receives search results from the search engine, formats them, and sends them to the terminals of field staff. Search results also reflect emotional information, and priority is given to content that requires a quick response.

[1144] Step 12:

[1145] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[1146] Step 13:

[1147] User (on-site staff): Responds to the customer based on the displayed information. If more specific information is required, the system asks additional questions. The system receives feedback and searches for information again. During this process, the system also continuously monitors the user's emotions.

[1148] Specific examples

[1149] Scenario 1: Consultation regarding opening a bank account

[1150] 1. Audio Acquisition

[1151] Customer: "I'd like to open a new account and would like to confirm the procedure."

[1152] Device: Record this conversation and store the audio data in a buffer.

[1153] 2. Voice Recognition

[1154] On the device: Sends the audio data in the buffer to the Google Speech-to-Text API in real time.

[1155] Server: Receives the returned text data "I would like to open a new account and would like to confirm the procedure." and saves it in the database with a timestamp.

[1156] 3. Keyword detection

[1157] Server: Analyzes the text data and detects the keyword "confirm." At this point, the part "I would like to open a new account. What are the procedures?" is retained as the analysis target.

[1158] 4. Analysis of conversation content

[1159] Server: Uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[1160] 5. Emotional Recognition

[1161] On the device: An emotion recognition engine is used to determine the user's emotion. In this case, it is determined that the customer has a question.

[1162] 6. Search Query Generation

[1163] Server: Generate a specific search query, such as "open a new account," based on sentiment information. If the customer is in a hurry, a quick response is required.

[1164] 7. Contacting Search Engines

[1165] Server: Sends the generated query as an HTTP request to the search engine's API.

[1166] 8. Retrieving and Displaying Search Results

[1167] Server: Sends search results received from the search engine to the terminals of field staff. The results are displayed with priority given to items that require immediate attention.

[1168] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[1169] 9. Feedback Loops

[1170] User (on-site staff): Responds to the customer based on the displayed information. For example, "To open a new account, you need the following documents." If the customer has a more detailed question, the system again detects keywords, searches for the necessary information, and continuously monitors the customer's emotions.

[1171] This system allows on-site staff to efficiently obtain information and respond to customers with consideration for their feelings, which is expected to reduce customer service time and improve customer satisfaction.

[1172] Example 2

[1173] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1174] While serving customers, on-site staff need to be able to respond quickly to customer questions and accurately recognize their emotions to provide more appropriate service. However, conventional systems only use voice recognition and do not analyze emotions, making it difficult to improve customer satisfaction. In addition, the need to process large amounts of data in real time requires specialized hardware and advanced software, making on-site implementation and operation complicated.

[1175] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1176] In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, means for analyzing the text data and detecting set keywords, means for, when a keyword is detected, retroactively analyzing the conversation content before and after the keyword and extracting key words and phrases, means for analyzing the tone, pitch, and speed of the voice data and recognizing emotions, means for generating search queries based on the emotional information and the extracted words and phrases and making inquiries to a search engine, and means for receiving and displaying search results from the search engine. This enables quick and accurate responses to customer questions and the provision of services that take emotions into consideration.

[1177] "Conversational audio" refers to the audio information exchanged between customers and on-site staff during customer service.

[1178] "Audio data" refers to information that is a digital recording of conversational voice.

[1179] A "voice recognition engine" is software that analyzes voice data and converts it into text data.

[1180] "Character data" refers to text information generated from voice data by a voice recognition engine.

[1181] A "keyword" is a particularly important word or phrase in the character data to be analyzed.

[1182] A "natural language processing engine" is software that analyzes text data and extracts key words and phrases.

[1183] A "search query" is a string of characters or phrases generated to search for specific information.

[1184] A "search engine" is a system or software that searches for information based on a search query.

[1185] "Tone" is an attribute that represents the pitch and quality of sound in audio data.

[1186] "Pitch" is an attribute that represents the frequency of sound in audio data.

[1187] "Speed" is an attribute that indicates the speaking speed of the voice in the voice data.

[1188] "Emotion recognition" is the process of identifying a speaker's emotions by analyzing the tone, pitch, and rate of speech data.

[1189] "Field staff" refers to employees who handle customer service.

[1190] "Customer" means the person receiving the service.

[1191] "Display" means a device that visually displays search results or other information.

[1192] A "buffer" is a memory space for temporarily storing audio data.

[1193] This invention relates to a system that enables on-site staff to provide more appropriate service by quickly and accurately answering customer questions while serving customers and recognizing their emotions. The system collects conversational voices, performs real-time speech and emotion recognition, and automatically searches for and displays the necessary information.

[1194] Specifically, the device uses a microphone installed at the customer service counter to continuously record the conversation between the customer and the on-site staff, temporarily storing the audio data in a buffer, and then sends this audio data in real time to a speech recognition engine (for example, Google Speech-to-Text API) and converts it into text data.

[1195] The server then stores the acquired text data in a database with a timestamp. The stored text data is analyzed to detect set keywords (e.g., "confirmation"). If a keyword is detected, the server extracts the surrounding conversation content and uses a natural language processing (NLP) engine (e.g., spaCy or NLTK) to extract key words and phrases.

[1196] Furthermore, the device analyzes the acoustic characteristics of the voice data, such as tone, pitch, and speed, and identifies the user's emotions using an emotion recognition engine (e.g., IBM Watson Tone Analyzer). This emotion information is analyzed together with the text data, and the server performs processing according to the emotion.

[1197] The server generates a search query based on the extracted key words and phrases and the identified sentiment information. The generated search query is sent as an HTTP request to a search engine (e.g., a general-purpose search engine API). The search results from the search engine are received by the server, appropriately filtered, and then sent to the device.

[1198] The terminal displays the received search results on the on-site staff's screen, allowing them to obtain the necessary information in a timely manner and respond to customers quickly and efficiently.If additional questions arise, a feedback loop is created in which the keywords are detected again and the necessary information is searched for.

[1199] As a concrete example, consider a consultation about opening a bank account. When a customer says, "I'd like to open a new account, but I'd like to confirm the procedure," the system records this conversation and saves the voice data in a buffer. A speech recognition engine converts this voice data into text data, generating the text data, "I'd like to open a new account, but I'd like to confirm the procedure."

[1200] Next, the server detects the keyword "confirmation" and analyzes and extracts the conversation content before and after it, obtaining key words and phrases such as "new account" and "opening procedure." At the same time, the device performs emotion recognition to identify the customer's question. The server generates a search query for "new account opening procedure" and displays appropriate search results if a quick response is needed.

[1201] An example of a prompt sentence is, "I would like to know the procedure for opening a new account at the bank. What documents do I need?" By using this, the system can quickly search for the relevant information and provide it to the field staff, allowing the customer to be served appropriately.

[1202] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1203] Step 1: Acquire audio data

[1204] Input: Voice conversation between customer and on-site staff

[1205] Specific operation: The terminal uses a microphone installed at the customer service counter to continuously record the conversation between the customer and the on-site staff. The recorded audio data is temporarily stored in a buffer within the terminal.

[1206] Output: Buffered audio data

[1207] Step 2: Real-time speech recognition

[1208] Input: Buffered audio data

[1209] Specific operation: The device sends the acquired voice data to a voice recognition engine (e.g., Google Speech-to-Text API) in real time, and converts the voice data into text data.

[1210] Output: Text data returned from the speech recognition engine

[1211] Step 3: Storing Character Data

[1212] Input: Character data returned from the speech recognition engine

[1213] Specific operation: The server adds a timestamp to the received character data and saves it in the database.

[1214] Output: Character data stored in the database with a timestamp

[1215] Step 4: Keyword detection

[1216] Input: Character data stored in a database

[1217] Specific operation: The server analyzes the text data and detects the set keyword (e.g., "confirm"), triggering the detection of the keyword and saving the conversation content before and after the keyword.

[1218] Output: The part where the keyword was detected and the conversation content before and after it

[1219] Step 5: Analyzing the conversation

[1220] Input: Conversation content before and after the keyword

[1221] What it does: The server uses a natural language processing engine (e.g., spaCy or NLTK) to extract key words and phrases, including tokenization, part-of-speech tagging, and named entity recognition.

[1222] Output: Extracted key words and phrases

[1223] Step 6: Recognize emotions

[1224] Input: Buffered audio data

[1225] Specific operation: The device analyzes acoustic features such as tone, pitch, and speed of the voice data and uses an emotion recognition engine (e.g., IBM Watson Tone Analyzer) to identify the user's emotion.

[1226] Output: Identified emotion information

[1227] Step 7: Generating a search query

[1228] Input: extracted key words and phrases, sentiment information

[1229] Specific operation: The server fuses the extracted words and phrases with sentiment information to generate search queries.

[1230] Output: Generated search query

[1231] Step 8: Contact the search engines

[1232] Input: Generated search query

[1233] Specific operation: The server creates and sends an HTTP request to send the generated query to the search engine's API.

[1234] Output: HTTP request to the search engine's API

[1235] Step 9: Retrieving and Displaying Search Results

[1236] Input: Search results from a search engine

[1237] Specific operation: The server analyzes search results from the search engine, filters and prioritizes them appropriately, and then sends them to the on-site staff's terminal. The terminal then displays the received search results on the on-site staff's display.

[1238] Output: Search results displayed on the device display

[1239] Step 10: Feedback Loop

[1240] Input: Search results displayed on the device display, additional questions from the customer

[1241] Specific operation: The user (on-site staff) answers the customer based on the displayed information. If the customer has a more detailed question, the system will detect keywords again, search for the necessary information, and continuously monitor the customer's emotions.

[1242] Output: Updated search results and continuously monitored sentiment information

[1243] (Application example 2)

[1244] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1245] When on-site staff deal with customers, they are required to respond to their questions quickly and accurately. Furthermore, understanding customer emotions and providing appropriate responses leads to improved customer satisfaction. However, with conventional systems, it has been difficult to accurately recognize customer emotions and provide information in real time. Therefore, there is a need for a system that integrates emotion recognition and voice recognition to provide appropriate services in real time that take customer emotions into consideration.

[1246] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into character data, means for analyzing the character data and detecting set keywords, means for recognizing emotions based on the voice data, means for generating a search query based on the extracted words, phrases, and emotion information and making an inquiry to a search engine, and means for receiving and displaying search results from the search engine. This enables field staff to understand customer emotions and quickly provide information that is appropriate for those emotions.

[1247] "Means for acquiring conversational audio" refers to devices or technologies for recording conversations between customers and wait staff in real time and collecting audio data.

[1248] "Means for transmitting voice data to a voice recognition engine in real time and converting it into text data" refers to a device or technology that immediately sends acquired voice data to a voice recognition engine and converts it from voice to text data.

[1249] "Means for analyzing character data and detecting set keywords" refers to a device or technology that analyzes the converted character data and finds specific keywords that have been set in advance.

[1250] "Means for analyzing the conversational content before and after a keyword is detected and extracting key words and phrases" refers to devices or technology that, when a keyword is found, analyzes the conversational content before and after the keyword and identifies key words and phrases.

[1251] "Means for recognizing emotions based on voice data" refers to devices or technologies that analyze characteristics of voice, such as tone and speed, to determine the speaker's emotions.

[1252] "Means for generating a search query based on the extracted words, phrases, and emotional information, and sending a query to a search engine" refers to a device or technology that creates a search query based on the words and phrases obtained through the analysis and the recognized emotional information, and performs an information search.

[1253] "Means for receiving and displaying search results from a search engine" refers to a device or technology that receives search results from a search engine and displays them in an easy-to-view format for the user.

[1254] This invention builds a system that integrates voice recognition and emotion recognition, centered around smart glasses worn by customer service staff. The system aims to respond quickly and accurately to customer requests and provide appropriate service according to the customer's emotions.

[1255] The server receives real-time voice data from the smart glasses' microphone and converts it into text using the Google Speech-to-Text API. This text data is then sent to the server and analyzed for keyword detection. If a keyword is detected, the server analyzes the surrounding conversation using a natural language processing engine to extract key words and phrases.

[1256] At the same time, the server uses the voice data to perform emotion recognition and obtain the customer's emotional information. Based on this emotional information and the extracted words and phrases, a search query is generated and sent to a search engine. The information received from the search engine is then displayed on the smart glasses' display, allowing the customer service staff to provide the customer with a quick and appropriate answer.

[1257] As a concrete example, consider a scenario in a consumer electronics retail store. If a customer asks, "Is this TV on sale?", the server detects this conversation, finds the keyword "sale" and analyzes the conversation before and after. At the same time, it recognizes the customer's emotions from the tone of their voice to determine whether they are anxious or angry about the question. Based on this, the server generates a search query, "TV sale now," and uses it to query the inventory management system, immediately displaying the results to the staff. The staff can quickly respond through the smart glasses, "Yes, it is currently on sale."

[1258] An example of a prompt sentence is, "By using the Google Speech-to-Text API to convert voice into text, and when specific keywords are detected, the system searches for information related to those keywords and displays it in real time on the staff's smart glasses," the system can generate appropriate search queries and provide accurate information in real time.

[1259] As a result, the on-site staff can understand the customer's emotions and quickly provide information that is in line with those emotions, thereby improving customer satisfaction.

[1260] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1261] Step 1:

[1262] The server acquires conversational voice from the microphone of the smart glasses. The acquired voice data is temporarily stored in a buffer in the smart glasses. The input is conversational voice and the output is voice data. This voice data is kept in the buffer for subsequent processing.

[1263] Step 2:

[1264] The device sends the audio data in the buffer to the speech recognition engine in real time, which uses the Google Speech-to-Text API to convert the audio data into text data. The input is audio data, and the output is text data converted from the audio.

[1265] Step 3:

[1266] The server analyzes the text data received from the speech recognition engine. Here, it detects set keywords (e.g., "special price" or "returns"). The input is the text data, and the output is the analysis results including the detected keywords.

[1267] Step 4:

[1268] When a keyword is detected, the server analyzes the conversation content before and after it, and uses a natural language processing engine to extract key words and phrases. The input is the text data and the detected keyword, and the output is the extracted words and phrases.

[1269] Step 5:

[1270] The server performs emotion recognition based on the voice data. It analyzes the tone and speed of the voice data to determine the customer's emotion (e.g., anger, joy, confusion). The input is the voice data, and the output is the analyzed emotional information.

[1271] Step 6:

[1272] The server generates a search query based on the extracted words and phrases and emotional information. Here, the search query is adjusted taking into account emotional information, such as when the customer is in a hurry. The input is the extracted words and phrases and emotional information, and the output is the generated search query.

[1273] Step 7:

[1274] The server queries the search engine using the generated search query. It sends the query as an HTTP request to the search engine's API. The input is the search query, and the output is the search results from the search engine.

[1275] Step 8:

[1276] The server receives search results from the search engine and sends them to the on-site staff's terminal, which then displays the search results on the display of the smart glasses. The input is the search results from the search engine, and the output is the search results converted into a displayable information format.

[1277] Step 9:

[1278] The user (on-site staff) responds to the customer based on the displayed information. If more specific information is required, an additional question is asked, and the system receives feedback again to search for the required information. The input is the customer's voice asking the additional question and the search results from the system, and the output is an appropriate answer to the customer.

[1279] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1280] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1281] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1282] [Fourth embodiment]

[1283] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1284] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1285] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1286] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1287] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1288] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1289] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1290] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1291] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1292] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1293] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1294] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1295] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1296] Overview of the voice recognition system

[1297] This invention is a speech recognition system that enables on-site staff to quickly respond to customer questions while serving customers. This system captures conversational speech, performs real-time speech recognition, and automatically searches for and displays the necessary information.

[1298] Overview of program processing

[1299] 1. Acquiring audio data

[1300] Terminal: A microphone installed at the customer service counter constantly records conversations and collects real-time audio data, which is temporarily stored in a buffer within the system.

[1301] 2. Real-time speech recognition

[1302] Device: The acquired voice data is sent from the buffer to the speech recognition engine, which converts the voice data into text data using, for example, the Google Speech-to-Text API or a similar service.

[1303] Server: Receives the converted character data and saves it in a database with a timestamp.

[1304] 3. Keyword detection

[1305] Server: Analyzes the character data and detects the set keyword (e.g., "confirm") and stores the partial string up to the appearance of the keyword.

[1306] 4. Analysis of conversation content

[1307] Server: When a keyword is detected, the server extracts the surrounding conversation (usually within one minute) and sends the extracted text data to a natural language processing (NLP) engine to extract key words and phrases.

[1308] 5. Generating Search Queries

[1309] Server: Generates a search query based on the extracted words and phrases. For example, a specific query such as "new account opening procedure" is generated.

[1310] 6. Contacting Search Engines

[1311] Server: Sends the generated query as an HTTP request to the search engine.

[1312] 7. Retrieving and Displaying Search Results

[1313] Server: Receives search results from the search engine and displays them on the terminals of field staff.

[1314] Terminal: The received search results are displayed on the on-site staff's display.

[1315] 8. Feedback Loops

[1316] User (field staff): Answers the customer based on the displayed information. If more specific information is needed, the system asks additional questions. The system receives feedback and searches for information again.

[1317] Specific examples

[1318] Scenario 1: Consultation regarding opening a bank account

[1319] 1. Audio Acquisition

[1320] Customer: "I'd like to open a new account and would like to confirm the procedure."

[1321] Device: Record this conversation and store the audio data in a buffer.

[1322] 2. Voice Recognition

[1323] On the device: Sends the audio data in the buffer to the Google Speech-to-Text API in real time.

[1324] Server: Receives the returned text data "I would like to open a new account and would like to confirm the procedure." and saves it in the database with a timestamp.

[1325] 3. Keyword detection

[1326] Server: Analyzes the text data and detects the keyword "confirm." At this point, the part "I would like to open a new account. What are the procedures?" is retained as the analysis target.

[1327] 4. Analysis of conversation content

[1328] Server: Uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[1329] 5. Generating Search Queries

[1330] Server: Generate a specific search query: "new account opening procedure."

[1331] 6. Contacting Search Engines

[1332] Server: Sends the generated query as an HTTP request to the search engine's API.

[1333] 7. Retrieving and Displaying Search Results

[1334] Server: Sends search results received from the search engine to the terminals of field staff.

[1335] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[1336] 8. Feedback Loops

[1337] User (on-site staff): Answers the customer based on the displayed information. For example, "To open a new account, you need the following documents." If the customer has a more detailed question, the system detects the keywords again and searches for the necessary information.

[1338] This system allows field staff to efficiently obtain information and quickly answer customer questions, which is expected to reduce customer service time and improve customer satisfaction.

[1339] The processing flow will be explained below.

[1340] Step 1:

[1341] Terminal: A microphone installed at the customer service counter continuously records conversations and collects voice data in real time. The voice data is temporarily stored in a buffer within the system.

[1342] Step 2:

[1343] Terminal: The acquired voice data is sent from the buffer to the voice recognition engine, which converts the voice data into text data using, for example, the Google Speech-to-Text API.

[1344] Step 3:

[1345] Server: Receives converted text data from the speech recognition engine and stores it in a database with a timestamp.

[1346] Step 4:

[1347] Server: Analyzes the character data and detects the set keyword (e.g., "confirm") and stores the partial string up to the appearance of the keyword.

[1348] Step 5:

[1349] Server: When a keyword is detected, the conversation content before and after it (usually within one minute) is extracted.

[1350] Step 6:

[1351] Server: The extracted text data is sent to a natural language processing engine to extract key words and phrases.

[1352] Step 7:

[1353] Server: Generates a search query based on the extracted words and phrases. For example, it generates a specific query such as "open a new account."

[1354] Step 8:

[1355] Server: Sends the generated query as an HTTP request to the search engine's API.

[1356] Step 9:

[1357] Server: Receives search results from the search engine, formats them, and sends them to the terminals of field staff.

[1358] Step 10:

[1359] Terminal: The received search results are displayed on the on-site staff's display.

[1360] Step 11:

[1361] User (field staff): Responds to the customer based on the displayed information. If more specific information is required, the user asks additional questions. The system receives feedback and searches for information again based on this.

[1362] Example 1

[1363] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1364] In customer service, staff need a lot of information to respond to customer questions quickly and accurately. However, manually searching for information is time-consuming and can lead to a decrease in customer satisfaction. Furthermore, a noisy and multitasking environment on-site makes it difficult to respond quickly. Therefore, there is a need for a system that can recognize voice in real time and automatically present relevant information.

[1365] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1366] In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, means for analyzing the text data and detecting set keywords, means for retroactively analyzing the conversation content before and after a keyword is detected and extracting key words and phrases, means for generating a search query based on the extracted words and phrases and making an inquiry to a search engine, means for receiving and displaying search results from the search engine, and means for searching for additional information based on the feedback. This enables on-site staff to respond to customer questions quickly and accurately and improve customer satisfaction.

[1367] "Conversational audio" refers to the verbal exchanges that take place between customers and on-site staff during customer service.

[1368] "Audio data" means a digital representation of speech collected using a device such as a microphone.

[1369] A "voice recognition engine" is software or a service that receives voice data as input and converts it into text-based character data.

[1370] "Character data" refers to text-format data converted by a voice recognition engine, and primarily refers to information expressed using characters (such as alphabets or kanji).

[1371] "Keywords" are words or phrases that are considered particularly important in the conversation and serve as triggers for the system to perform specific operations.

[1372] The "conversation content" is a series of words exchanged between the customer and the on-site staff, and is a group of text data containing specific information.

[1373] A "natural language processing engine" is an algorithm or software that analyzes text data and extracts key words and phrases.

[1374] A "search query" is a specific phrase or combination of words generated to ask a search engine.

[1375] A "search engine" is a system or software that searches the Internet and databases for relevant information based on an input search query.

[1376] "Search results" refers to the list or content of information that a search engine provides based on a search query.

[1377] "Feedback" refers to the process and results of field staff conducting additional information searches based on the information and results they receive from the system.

[1378] This invention is a speech recognition system that enables on-site staff to respond quickly and accurately to customer questions during customer service duties. This system captures conversational voices, performs real-time speech recognition, and automatically searches for and displays necessary information, significantly improving on-site efficiency.

[1379] 1. Hardware and Software Used

[1380] The system hardware configuration is as follows:

[1381] Microphone (to capture speech)

[1382] Terminal (to acquire audio data and perform initial processing)

[1383] Server (for data analysis, generating search queries, and communicating with search engines)

[1384] Display (to show results to field staff)

[1385] The system software configuration is as follows:

[1386] Google Speech-to-Text API (voice recognition engine)

[1387] NLP engines (for natural language processing, e.g. SpaCy)

[1388] Search engine APIs (for information searches, e.g., Google and Bing APIs)

[1389] 2. System processing overview

[1390] The process of this speech recognition system involves the following steps:

[1391] Acquiring audio data:

[1392] The device collects conversational audio in real time through a microphone and temporarily stores it in a buffer. This audio data is stored in PCM format.

[1393] Real-time speech recognition:

[1394] The device splits the audio data in the buffer into regular batches (e.g., every 5 seconds) and sends them to the Google Speech-to-Text API, which converts the audio into text data.

[1395] The server stores the received character data in a database with a timestamp.

[1396] Keyword detection and conversation analysis:

[1397] The server periodically scans new text data for set keywords, and if a keyword is found, sends the surrounding conversational content to an NLP engine to extract key words and phrases.

[1398] Search query generation and query:

[1399] The server then combines the extracted words and phrases to generate a search query, which it then sends as an HTTP request to the search engine's API, which then returns the search results.

[1400] Displaying search results:

[1401] The server displays the search results received from the search engine on the terminal's display, and the field staff responds to the customer based on the displayed information.

[1402] Feedback Loop:

[1403] The user (field staff) responds to the customer based on the displayed information. If more details are needed, the user provides feedback to the system and searches for additional information.

[1404] 3. Explanation of specific examples

[1405] Scenario 1: Consultation regarding opening a bank account

[1406] 1. Audio capture:

[1407] Customer: "I'd like to open a new account and would like to confirm the procedure."

[1408] The device records this conversation and stores the audio data in a buffer.

[1409] 2. Speech Recognition:

[1410] The device sends the buffered audio data to the Google Speech-to-Text API in real time.

[1411] The server receives the returned text data, "I would like to open a new account and would like to confirm the procedure," and stores it in the database with a timestamp.

[1412] 3. Keyword detection:

[1413] The server analyzes the text data and detects the keyword "confirmation." At this point, it retains the part "I would like to open a new account. What are the procedures?" as the analysis target.

[1414] 4. Conversation Analysis:

[1415] The server uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[1416] 5. Generating search queries:

[1417] The server generates a specific search query, "new account opening procedure," and sends it as an HTTP request to the search engine's API.

[1418] 6. Retrieving and displaying search results:

[1419] The server sends the search results received from the search engine to the terminal of the on-site staff and displays them on the screen.

[1420] The terminal then displays the search results on the on-site staff's screen, such as information about the documents and conditions required to open a new account.

[1421] 7. Feedback Loop:

[1422] The user (field staff) responds to the customer based on the displayed information. If the customer has a more detailed question, the system will again detect the keywords and search for the necessary information.

[1423] This system allows field staff to efficiently obtain information and quickly answer customer questions, which is expected to reduce customer service time and improve customer satisfaction.

[1424] Example prompt sentence:

[1425] I would like to open a new account and would like to confirm the procedure.

[1426] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1427] Step 1: Acquire audio data

[1428] The terminal captures conversational audio in real time from a microphone installed at the customer service counter. The input is the conversational audio between the customer and the front-line staff, which is converted into digital format and stored in a buffer in the terminal's memory. Specifically, the audio signal is converted from analog to digital format and continuously written to the buffer in PCM format.

[1429] Step 2: Real-time speech recognition

[1430] The device buffers the audio data in regular batches (for example, every 5 seconds). The input is audio data, which is sent to the Google Speech-to-Text API to be converted into text. At this stage, the audio data is analyzed by a speech recognition engine, and text data is generated as output. This text data is then sent from the device to the server.

[1431] Step 3: Storing Character Data

[1432] The server stores the text data received from the device in a database with a timestamp. The input is text data, and a timestamp is added during the process of saving it in the database. Specifically, the text data is stored in an SQLite database, and synchronization between voice and text is maintained.

[1433] Step 4: Keyword detection

[1434] The server periodically scans new text data to detect configured keywords. The input is character data, and it searches for specific keywords using regular expressions, etc. This detection process outputs the part containing the configured keyword. Specifically, when a keyword is detected, the conversation content before and after it is temporarily stored in memory.

[1435] Step 5: Analyzing the conversation

[1436] The server sends the conversational content, including one minute before and after the text where the keyword was detected, to a natural language processing (NLP) engine. The input is text data, which is analyzed by an NLP engine (e.g., SpaCy) to extract key words and phrases. The output is a list of extracted words and phrases. Specifically, the NLP engine tokenizes the text and identifies and extracts key words such as nouns and verbs.

[1437] Step 6: Generating a search query

[1438] The server generates a query based on the extracted words and phrases in a format that is easy for users to search. The input is a list of key words and phrases, and a specific search query is generated by string concatenation. The output is the search query. For example, a phrase such as "new account opening procedure" is generated.

[1439] Step 7: Contact the search engines

[1440] The server then sends the generated search query to the search engine's API. The input is the search query, which is queried to the search engine (e.g., Google or Bing) using an HTTP GET request. The output is the search results. Specifically, the query is appended to a URL and sent to the search engine, which returns the results in JSON format.

[1441] Step 8: Retrieving and Displaying Search Results

[1442] The server sends the search results received from the search engine to the terminal and displays them on the on-site staff's display. The input is the search results from the search engine, which are formatted for display on the terminal's display. The output is the information displayed on the on-site staff's display. Specifically, information such as "Documents and conditions required for opening a new account" is displayed.

[1443] Step 9: Feedback Loop

[1444] The user (on-site staff) responds to the customer based on the displayed information. If more specific information is required, they ask an additional question. The input is the customer's additional question, and the content of this question is picked up again from the microphone, and processing returns to step 1 for reprocessing. The output is updated search results and a new answer. Specifically, more detailed information is searched again based on the customer's additional question and displayed.

[1445] (Application example 1)

[1446] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1447] Conventional voice recognition systems were not sufficient for quickly obtaining necessary information during conversations with customers. Furthermore, search functions based on specific keywords alone sometimes made it difficult for on-site staff to accurately answer customer questions. As a result, customer satisfaction and response efficiency declined.

[1448] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1449] In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, means for analyzing the text data and detecting set keywords, means for retroactively analyzing the conversation content before and after a keyword is detected and extracting key words and phrases, means for generating a search query based on the extracted words and phrases and making an inquiry to a search engine, means for receiving search results from the search engine and displaying them on a user terminal, and means for performing another information search based on the feedback. This enables on-site staff to quickly and accurately answer customer questions.

[1450] "Conversational voice" refers to the voice exchanged between the user and the person answering the call.

[1451] "Means for acquiring" refers to a device or system for collecting audio data.

[1452] "Send to the speech recognition engine in real time" means that collected speech data is immediately transferred to the speech recognition engine.

[1453] "Means for converting to text data" refers to a process or device that converts audio data into text data.

[1454] "Analyzing character data" refers to the process of interpreting the converted text data and extracting necessary information.

[1455] "Means for detecting set keywords" refers to a system that identifies specific predefined words or phrases within text data.

[1456] "When a keyword is detected, the conversation content before and after it is analyzed" refers to analyzing the text before and after the detected keyword and extracting related content.

[1457] "Means for extracting key words and phrases" refers to a system that finds important words and phrases from the analyzed text data.

[1458] "Means for generating a search query and sending it to a search engine" refers to creating a search query based on the extracted words and phrases and sending that query to a search engine.

[1459] "Means for receiving search results from a search engine and displaying them on a user's device" refers to a system that receives responses from a search engine and displays that information on a user's device.

[1460] "Means of searching for information again based on feedback" refers to the process of generating a search query again based on additional questions or requests from the user to obtain new information.

[1461] "User devices" refer to devices used by customers and field staff, such as smartphones and tablets.

[1462] A "natural language processing engine" is a system that analyzes text data and extracts meaning and related information.

[1463] "General-purpose device" refers to a device that has a wide range of functions and is not limited to a specific use, such as a general-purpose smartphone or tablet.

[1464] The system for implementing this invention uses the following hardware and software. First, the hardware used includes a smartphone with a microphone for recording voice and a server for data processing. The software includes a Python program, a voice recognition library (e.g., the SpeechRecognition library), the Google Speech-to-Text API, and a natural language processing engine (e.g., an NLP engine for natural language processing).

[1465] The server uses a speech recognition engine to convert voice data into text data in real time. The converted text data is analyzed to detect set keywords. When a keyword is detected, the conversation content before and after it is analyzed to extract key words and phrases. A search query is generated based on these extracted words and phrases, and this search query is used as data to query a search engine.

[1466] The search results returned by the search engine are displayed on the user's smartphone. Here, "user device" refers to the device used by the customer or field staff, such as a smartphone or tablet. Furthermore, the system also incorporates a mechanism for re-searching information based on user feedback, enabling more accurate responses to questions.

[1467] For example, when a store clerk asks, "Is this item in stock?", the system instantly recognizes the keyword "in stock" and searches for and displays inventory information. This information is detailed information that customers want, and helps field staff respond quickly.

[1468] Example prompt sentence:

[1469] Input Sentence: Customer: "Is this item in stock?"

[1470] Example output: System: In response to the question "Is this item in stock?", the system recognizes the keyword "in stock" and searches for and displays stock information.

[1471] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1472] Step 1:

[1473] The user uses the microphone on their smartphone to record the conversation. This voice data is temporarily stored in a buffer within the system. The input is the conversation between the user and the customer, and the output is voice data.

[1474] Step 2:

[1475] The device sends the captured voice data in real time to a voice recognition engine (for example, Google Speech-to-Text API). This engine converts the voice data into text data. The input is voice data, and the output is text data.

[1476] Step 3:

[1477] The server receives the converted character data and stores it in a database with a timestamp. The input is character data, and the output is character data with a timestamp. This timestamp is used for later analysis.

[1478] Step 4:

[1479] The server analyzes the text data and detects the set keywords. For example, keywords such as "stock" and "price" are detected. The input is the text data, and the output is the detected keywords.

[1480] Step 5:

[1481] When a keyword is detected, the server analyzes the conversation content before and after the keyword and extracts key words and phrases. For example, key phrases such as "this product" and "is it in stock" are extracted. The input is the keyword and text data, and the output is the extracted key words and phrases.

[1482] Step 6:

[1483] The server generates a search query based on the extracted words and phrases. This query is intended to retrieve specific information. For example, a search query might be generated for "this product in stock." The input is the main word or phrase, and the output is the search query.

[1484] Step 7:

[1485] The server sends the generated search query to the search engine as an HTTP request and waits for a response from the search engine (e.g., Google). The input is the search query, and the output is the search results from the search engine.

[1486] Step 8:

[1487] The server receives search results from the search engine and displays them on the user's terminal. These results are information that the user provides to their customers. The input is the search results from the search engine, and the output is the information displayed on the user's terminal.

[1488] Step 9:

[1489] The user responds to the customer based on the displayed information. If the customer requires more detailed information, the user asks an additional question, and the system searches for information again based on the feedback. The input is the additional question from the customer, and the output is the information obtained again.

[1490] This process allows the system to provide quick and accurate answers to customer questions.

[1491] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1492] Overview of speech and emotion recognition systems

[1493] This invention is a speech and emotion recognition system that aims to provide more accurate service by not only enabling on-site staff to respond quickly to customer questions while serving customers, but also by recognizing customer emotions and utilizing that information. This system captures conversational voice, performs speech and emotion recognition in real time, and automatically searches for and displays the necessary information.

[1494] Overview of program processing

[1495] 1. Acquiring audio data

[1496] Terminal: A microphone installed at the customer service counter constantly records conversations, collecting voice data in real time. This voice data is temporarily stored in a buffer within the system.

[1497] 2. Real-time speech recognition

[1498] Terminal: The acquired voice data is sent from the buffer to the speech recognition engine, which converts the voice data into text data. For example, the Google Speech-to-Text API is used to convert voice to text.

[1499] Server: The character data received from the speech recognition engine is saved in a database with a timestamp.

[1500] 3. Keyword detection

[1501] Server: Analyzes the character data and detects the set keyword (e.g., "confirm") and stores the partial string up to the appearance of the keyword.

[1502] 4. Analysis of conversation content

[1503] Server: When a keyword is detected, the server extracts the surrounding conversation (usually within one minute) and sends the extracted text data to a natural language processing (NLP) engine to extract key words and phrases.

[1504] 5. Emotional Recognition

[1505] Device: Voice data is input into an emotion recognition engine to analyze the user's emotions. For example, it determines whether the user is angry or happy based on the tone and speed of the voice.

[1506] Server: Stores the user's emotional data based on the analyzed emotional information.

[1507] 6. Search Query Generation

[1508] Server: Generates a search query based on the extracted words and phrases and sentiment information. For example, "Open a new account" and if the user is angry, a quick response is required.

[1509] 7. Contacting Search Engines

[1510] Server: Sends the generated query as an HTTP request to the search engine's API.

[1511] 8. Retrieving and Displaying Search Results

[1512] Server: Receives search results from the search engine and sends them to the terminals of field staff. Search results also reflect emotional information, for example, prioritizing content that requires a quick response.

[1513] Terminal: The received search results are displayed on the on-site staff's display.

[1514] 9. Feedback Loops

[1515] User (on-site staff): Responds to the customer based on the displayed information. If more specific information is required, the system asks additional questions. The system receives feedback and searches for information again. During this process, the user's emotions are continuously monitored.

[1516] Specific examples

[1517] Scenario 1: Consultation regarding opening a bank account

[1518] 1. Audio Acquisition

[1519] Customer: "I'd like to open a new account and would like to confirm the procedure."

[1520] Device: Record this conversation and store the audio data in a buffer.

[1521] 2. Voice Recognition

[1522] On the device: Sends the audio data in the buffer to the Google Speech-to-Text API in real time.

[1523] Server: Receives the returned text data "I would like to open a new account and would like to confirm the procedure." and saves it in the database with a timestamp.

[1524] 3. Keyword detection

[1525] Server: Analyzes the text data and detects the keyword "confirm." At this point, the part "I would like to open a new account. What are the procedures?" is retained as the analysis target.

[1526] 4. Analysis of conversation content

[1527] Server: Uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[1528] 5. Emotional Recognition

[1529] On the device: An emotion recognition engine is used to determine the user's emotion. In this case, it is determined that the customer has a question.

[1530] 6. Search Query Generation

[1531] Server: Generate a specific search query, such as "open a new account," based on sentiment information. If the customer is in a hurry, a quick response is required.

[1532] 7. Contacting Search Engines

[1533] Server: Sends the generated query as an HTTP request to the search engine's API.

[1534] 8. Retrieving and Displaying Search Results

[1535] Server: Sends search results received from the search engine to the terminals of field staff. The results are displayed with priority given to items that require immediate attention.

[1536] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[1537] 9. Feedback Loops

[1538] User (on-site staff): Responds to the customer based on the displayed information. For example, "To open a new account, you need the following documents." If the customer has a more detailed question, the system again detects keywords, searches for the necessary information, and continuously monitors the customer's emotions.

[1539] This system allows on-site staff to efficiently obtain information and respond to customers with consideration for their feelings, which is expected to reduce customer service time and improve customer satisfaction.

[1540] The processing flow will be explained below.

[1541] Step 1:

[1542] Terminal: A microphone installed at the customer service counter continuously records conversations and collects voice data in real time. The voice data is temporarily stored in a buffer within the system.

[1543] Step 2:

[1544] Terminal: The acquired voice data is sent from the buffer to a voice recognition engine, for example, using the Google Speech-to-Text API to convert the voice data into text data.

[1545] Step 3:

[1546] Server: Receives converted text data from the speech recognition engine and stores it in a database with a timestamp.

[1547] Step 4:

[1548] Server: Analyzes the character data and detects the set keyword (e.g., "confirm"), and stores the substring up to the detected keyword.

[1549] Step 5:

[1550] Server: When a keyword is detected, the conversation content before and after it (usually within one minute) is extracted.

[1551] Step 6:

[1552] Server: The extracted text data is sent to a natural language processing (NLP) engine to extract key words and phrases.

[1553] Step 7:

[1554] Device: Voice data is input into an emotion recognition engine to analyze the user's emotions. For example, the emotional state is determined from the tone, speed, and pitch of the voice, and emotions such as joy, anger, and sadness are identified.

[1555] Step 8:

[1556] Server: Based on the analyzed emotional information, the user's emotional data is stored in a database.

[1557] Step 9:

[1558] Server: Generates search queries based on the extracted words and phrases and emotion information. For example, it generates a specific query such as "new account opening procedure" and sets a high priority if the user is angry.

[1559] Step 10:

[1560] Server: Sends the generated search query as an HTTP request to the search engine's API.

[1561] Step 11:

[1562] Server: Receives search results from the search engine, formats them, and sends them to the terminals of field staff. Search results also reflect emotional information, and priority is given to content that requires a quick response.

[1563] Step 12:

[1564] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[1565] Step 13:

[1566] User (on-site staff): Responds to the customer based on the displayed information. If more specific information is required, the system asks additional questions. The system receives feedback and searches for information again. During this process, the system also continuously monitors the user's emotions.

[1567] Specific examples

[1568] Scenario 1: Consultation regarding opening a bank account

[1569] 1. Audio Acquisition

[1570] Customer: "I'd like to open a new account and would like to confirm the procedure."

[1571] Device: Record this conversation and store the audio data in a buffer.

[1572] 2. Voice Recognition

[1573] On the device: Sends the audio data in the buffer to the Google Speech-to-Text API in real time.

[1574] Server: Receives the returned text data "I would like to open a new account and would like to confirm the procedure." and saves it in the database with a timestamp.

[1575] 3. Keyword detection

[1576] Server: Analyzes the text data and detects the keyword "confirm." At this point, the part "I would like to open a new account. What are the procedures?" is retained as the analysis target.

[1577] 4. Analysis of conversation content

[1578] Server: Uses a natural language processing engine to analyze the conversation content before and after the keywords and extract keywords such as "new account" and "opening procedure."

[1579] 5. Emotional Recognition

[1580] On the device: An emotion recognition engine is used to determine the user's emotion. In this case, it is determined that the customer has a question.

[1581] 6. Search Query Generation

[1582] Server: Generate a specific search query, such as "open a new account," based on sentiment information. If the customer is in a hurry, a quick response is required.

[1583] 7. Contacting Search Engines

[1584] Server: Sends the generated query as an HTTP request to the search engine's API.

[1585] 8. Retrieving and Displaying Search Results

[1586] Server: Sends search results received from the search engine to the terminals of field staff. The results are displayed with priority given to items that require immediate attention.

[1587] Terminal: The received search results are displayed on the on-site staff's screen. For example, information such as "Documents and conditions required for opening a new account" is displayed.

[1588] 9. Feedback Loops

[1589] User (on-site staff): Responds to the customer based on the displayed information. For example, "To open a new account, you need the following documents." If the customer has a more detailed question, the system again detects keywords, searches for the necessary information, and continuously monitors the customer's emotions.

[1590] This system allows on-site staff to efficiently obtain information and respond to customers with consideration for their feelings, which is expected to reduce customer service time and improve customer satisfaction.

[1591] Example 2

[1592] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1593] While serving customers, on-site staff need to be able to respond quickly to customer questions and accurately recognize their emotions to provide more appropriate service. However, conventional systems only use voice recognition and do not analyze emotions, making it difficult to improve customer satisfaction. In addition, the need to process large amounts of data in real time requires specialized hardware and advanced software, making on-site implementation and operation complicated.

[1594] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1595] In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data, means for analyzing the text data and detecting set keywords, means for, when a keyword is detected, retroactively analyzing the conversation content before and after the keyword and extracting key words and phrases, means for analyzing the tone, pitch, and speed of the voice data and recognizing emotions, means for generating search queries based on the emotional information and the extracted words and phrases and making inquiries to a search engine, and means for receiving and displaying search results from the search engine. This enables quick and accurate responses to customer questions and the provision of services that take emotions into consideration.

[1596] "Conversational audio" refers to the audio information exchanged between customers and on-site staff during customer service.

[1597] "Audio data" refers to information that is a digital recording of conversational voice.

[1598] A "voice recognition engine" is software that analyzes voice data and converts it into text data.

[1599] "Character data" refers to text information generated from voice data by a voice recognition engine.

[1600] A "keyword" is a particularly important word or phrase in the character data to be analyzed.

[1601] A "natural language processing engine" is software that analyzes text data and extracts key words and phrases.

[1602] A "search query" is a string of characters or phrases generated to search for specific information.

[1603] A "search engine" is a system or software that searches for information based on a search query.

[1604] "Tone" is an attribute that represents the pitch and quality of sound in audio data.

[1605] "Pitch" is an attribute that represents the frequency of sound in audio data.

[1606] "Speed" is an attribute that indicates the speaking speed of the voice in the voice data.

[1607] "Emotion recognition" is the process of identifying a speaker's emotions by analyzing the tone, pitch, and rate of speech data.

[1608] "Field staff" refers to employees who handle customer service.

[1609] "Customer" means the person receiving the service.

[1610] "Display" means a device that visually displays search results or other information.

[1611] A "buffer" is a memory space for temporarily storing audio data.

[1612] This invention relates to a system that enables on-site staff to provide more appropriate service by quickly and accurately answering customer questions while serving customers and recognizing their emotions. The system collects conversational voices, performs real-time speech and emotion recognition, and automatically searches for and displays the necessary information.

[1613] Specifically, the device uses a microphone installed at the customer service counter to continuously record the conversation between the customer and the on-site staff, temporarily storing the audio data in a buffer, and then sends this audio data in real time to a speech recognition engine (for example, Google Speech-to-Text API) and converts it into text data.

[1614] The server then stores the acquired text data in a database with a timestamp. The stored text data is analyzed to detect set keywords (e.g., "confirmation"). If a keyword is detected, the server extracts the surrounding conversation content and uses a natural language processing (NLP) engine (e.g., spaCy or NLTK) to extract key words and phrases.

[1615] Furthermore, the device analyzes the acoustic characteristics of the voice data, such as tone, pitch, and speed, and identifies the user's emotions using an emotion recognition engine (e.g., IBM Watson Tone Analyzer). This emotion information is analyzed together with the text data, and the server performs processing according to the emotion.

[1616] The server generates a search query based on the extracted key words and phrases and the identified sentiment information. The generated search query is sent as an HTTP request to a search engine (e.g., a general-purpose search engine API). The search results from the search engine are received by the server, appropriately filtered, and then sent to the device.

[1617] The terminal displays the received search results on the on-site staff's screen, allowing them to obtain the necessary information in a timely manner and respond to customers quickly and efficiently.If additional questions arise, a feedback loop is created in which the keywords are detected again and the necessary information is searched for.

[1618] As a concrete example, consider a consultation about opening a bank account. When a customer says, "I'd like to open a new account, but I'd like to confirm the procedure," the system records this conversation and saves the voice data in a buffer. A speech recognition engine converts this voice data into text data, generating the text data, "I'd like to open a new account, but I'd like to confirm the procedure."

[1619] Next, the server detects the keyword "confirmation" and analyzes and extracts the conversation content before and after it, obtaining key words and phrases such as "new account" and "opening procedure." At the same time, the device performs emotion recognition to identify the customer's question. The server generates a search query for "new account opening procedure" and displays appropriate search results if a quick response is needed.

[1620] An example of a prompt sentence is, "I would like to know the procedure for opening a new account at the bank. What documents do I need?" By using this, the system can quickly search for the relevant information and provide it to the field staff, allowing the customer to be served appropriately.

[1621] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1622] Step 1: Acquire audio data

[1623] Input: Voice conversation between customer and on-site staff

[1624] Specific operation: The terminal uses a microphone installed at the customer service counter to continuously record the conversation between the customer and the on-site staff. The recorded audio data is temporarily stored in a buffer within the terminal.

[1625] Output: Buffered audio data

[1626] Step 2: Real-time speech recognition

[1627] Input: Buffered audio data

[1628] Specific operation: The device sends the acquired voice data to a voice recognition engine (e.g., Google Speech-to-Text API) in real time, and converts the voice data into text data.

[1629] Output: Text data returned from the speech recognition engine

[1630] Step 3: Storing Character Data

[1631] Input: Character data returned from the speech recognition engine

[1632] Specific operation: The server adds a timestamp to the received character data and saves it in the database.

[1633] Output: Character data stored in the database with a timestamp

[1634] Step 4: Keyword detection

[1635] Input: Character data stored in a database

[1636] Specific operation: The server analyzes the text data and detects the set keyword (e.g., "confirm"), triggering the detection of the keyword and saving the conversation content before and after the keyword.

[1637] Output: The part where the keyword was detected and the conversation content before and after it

[1638] Step 5: Analyzing the conversation

[1639] Input: Conversation content before and after the keyword

[1640] What it does: The server uses a natural language processing engine (e.g., spaCy or NLTK) to extract key words and phrases, including tokenization, part-of-speech tagging, and named entity recognition.

[1641] Output: Extracted key words and phrases

[1642] Step 6: Recognize emotions

[1643] Input: Buffered audio data

[1644] Specific operation: The device analyzes acoustic features such as tone, pitch, and speed of the voice data and uses an emotion recognition engine (e.g., IBM Watson Tone Analyzer) to identify the user's emotion.

[1645] Output: Identified emotion information

[1646] Step 7: Generating a search query

[1647] Input: extracted key words and phrases, sentiment information

[1648] Specific operation: The server fuses the extracted words and phrases with sentiment information to generate search queries.

[1649] Output: Generated search query

[1650] Step 8: Contact the search engines

[1651] Input: Generated search query

[1652] Specific operation: The server creates and sends an HTTP request to send the generated query to the search engine's API.

[1653] Output: HTTP request to the search engine's API

[1654] Step 9: Retrieving and Displaying Search Results

[1655] Input: Search results from a search engine

[1656] Specific operation: The server analyzes search results from the search engine, filters and prioritizes them appropriately, and then sends them to the on-site staff's terminal. The terminal then displays the received search results on the on-site staff's display.

[1657] Output: Search results displayed on the device display

[1658] Step 10: Feedback Loop

[1659] Input: Search results displayed on the device display, additional questions from the customer

[1660] Specific operation: The user (on-site staff) answers the customer based on the displayed information. If the customer has a more detailed question, the system will detect keywords again, search for the necessary information, and continuously monitor the customer's emotions.

[1661] Output: Updated search results and continuously monitored sentiment information

[1662] (Application example 2)

[1663] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1664] When on-site staff deal with customers, they are required to respond to their questions quickly and accurately. Furthermore, understanding customer emotions and providing appropriate responses leads to improved customer satisfaction. However, with conventional systems, it has been difficult to accurately recognize customer emotions and provide information in real time. Therefore, there is a need for a system that integrates emotion recognition and voice recognition to provide appropriate services in real time that take customer emotions into consideration.

[1665] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for acquiring conversational voice, means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into character data, means for analyzing the character data and detecting set keywords, means for recognizing emotions based on the voice data, means for generating a search query based on the extracted words, phrases, and emotion information and making an inquiry to a search engine, and means for receiving and displaying search results from the search engine. This enables field staff to understand customer emotions and quickly provide information that is appropriate for those emotions.

[1666] "Means for acquiring conversational audio" refers to devices or technologies for recording conversations between customers and wait staff in real time and collecting audio data.

[1667] "Means for transmitting voice data to a voice recognition engine in real time and converting it into text data" refers to a device or technology that immediately sends acquired voice data to a voice recognition engine and converts it from voice to text data.

[1668] "Means for analyzing character data and detecting set keywords" refers to a device or technology that analyzes the converted character data and finds specific keywords that have been set in advance.

[1669] "Means for analyzing the conversational content before and after a keyword is detected and extracting key words and phrases" refers to devices or technology that, when a keyword is found, analyzes the conversational content before and after the keyword and identifies key words and phrases.

[1670] "Means for recognizing emotions based on voice data" refers to devices or technologies that analyze characteristics of voice, such as tone and speed, to determine the speaker's emotions.

[1671] "Means for generating a search query based on the extracted words, phrases, and emotional information, and sending a query to a search engine" refers to a device or technology that creates a search query based on the words and phrases obtained through the analysis and the recognized emotional information, and performs an information search.

[1672] "Means for receiving and displaying search results from a search engine" refers to a device or technology that receives search results from a search engine and displays them in an easy-to-view format for the user.

[1673] This invention builds a system that integrates voice recognition and emotion recognition, centered around smart glasses worn by customer service staff. The system aims to respond quickly and accurately to customer requests and provide appropriate service according to the customer's emotions.

[1674] The server receives real-time voice data from the smart glasses' microphone and converts it into text using the Google Speech-to-Text API. This text data is then sent to the server and analyzed for keyword detection. If a keyword is detected, the server analyzes the surrounding conversation using a natural language processing engine to extract key words and phrases.

[1675] At the same time, the server uses the voice data to perform emotion recognition and obtain the customer's emotional information. Based on this emotional information and the extracted words and phrases, a search query is generated and sent to a search engine. The information received from the search engine is then displayed on the smart glasses' display, allowing the customer service staff to provide the customer with a quick and appropriate answer.

[1676] As a concrete example, consider a scenario in a consumer electronics retail store. If a customer asks, "Is this TV on sale?", the server detects this conversation, finds the keyword "sale" and analyzes the conversation before and after. At the same time, it recognizes the customer's emotions from the tone of their voice to determine whether they are anxious or angry about the question. Based on this, the server generates a search query, "TV sale now," and uses it to query the inventory management system, immediately displaying the results to the staff. The staff can quickly respond through the smart glasses, "Yes, it is currently on sale."

[1677] An example of a prompt sentence is, "By using the Google Speech-to-Text API to convert voice into text, and when specific keywords are detected, the system searches for information related to those keywords and displays it in real time on the staff's smart glasses," the system can generate appropriate search queries and provide accurate information in real time.

[1678] As a result, the on-site staff can understand the customer's emotions and quickly provide information that is in line with those emotions, thereby improving customer satisfaction.

[1679] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1680] Step 1:

[1681] The server acquires conversational voice from the microphone of the smart glasses. The acquired voice data is temporarily stored in a buffer in the smart glasses. The input is conversational voice and the output is voice data. This voice data is kept in the buffer for subsequent processing.

[1682] Step 2:

[1683] The device sends the audio data in the buffer to the speech recognition engine in real time, which uses the Google Speech-to-Text API to convert the audio data into text data. The input is audio data, and the output is text data converted from the audio.

[1684] Step 3:

[1685] The server analyzes the text data received from the speech recognition engine. Here, it detects set keywords (e.g., "special price" or "returns"). The input is the text data, and the output is the analysis results including the detected keywords.

[1686] Step 4:

[1687] When a keyword is detected, the server analyzes the conversation content before and after it, and uses a natural language processing engine to extract key words and phrases. The input is the text data and the detected keyword, and the output is the extracted words and phrases.

[1688] Step 5:

[1689] The server performs emotion recognition based on the voice data. It analyzes the tone and speed of the voice data to determine the customer's emotion (e.g., anger, joy, confusion). The input is the voice data, and the output is the analyzed emotional information.

[1690] Step 6:

[1691] The server generates a search query based on the extracted words and phrases and emotional information. Here, the search query is adjusted taking into account emotional information, such as when the customer is in a hurry. The input is the extracted words and phrases and emotional information, and the output is the generated search query.

[1692] Step 7:

[1693] The server queries the search engine using the generated search query. It sends the query as an HTTP request to the search engine's API. The input is the search query, and the output is the search results from the search engine.

[1694] Step 8:

[1695] The server receives search results from the search engine and sends them to the on-site staff's terminal, which then displays the search results on the display of the smart glasses. The input is the search results from the search engine, and the output is the search results converted into a displayable information format.

[1696] Step 9:

[1697] The user (on-site staff) responds to the customer based on the displayed information. If more specific information is required, an additional question is asked, and the system receives feedback again to search for the required information. The input is the customer's voice asking the additional question and the search results from the system, and the output is an appropriate answer to the customer.

[1698] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[1699] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1700] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[1701] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1702] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[1703] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[1704] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[1705] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[1706] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[1707] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[1708] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[1709] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[1710] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[1711] 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.

[1712] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[1713] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[1714] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[1715] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[1716] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[1717] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[1718] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[1719] The following is further disclosed regarding the above embodiment.

[1720] (Claim 1)

[1721] A means for acquiring speech;

[1722] A means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data;

[1723] A means for analyzing character data and detecting set keywords;

[1724] When a keyword is detected, the conversation content before and after it is analyzed to extract key words and phrases.

[1725] A means for generating search queries based on the extracted words and phrases and sending them to a search engine;

[1726] means for receiving and displaying search results from a search engine;

[1727] A system including:

[1728] (Claim 2)

[1729] 10. The system of claim 1, further comprising means for using a natural language processing engine in analyzing the conversation content surrounding the specified keyword.

[1730] (Claim 3)

[1731] 10. The system of claim 1, further comprising means for the search engine to provide information based on a search query.

[1732] "Example 1"

[1733] (Claim 1)

[1734] A means for acquiring speech;

[1735] A means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data;

[1736] A means for analyzing character data and detecting set keywords;

[1737] When a keyword is detected, the conversation content before and after it is analyzed to extract key words and phrases.

[1738] A means for generating search queries based on the extracted words and phrases and sending them to a search engine;

[1739] means for receiving and displaying search results from a search engine;

[1740] a means of conducting additional information searches based on the feedback;

[1741] A system including:

[1742] (Claim 2)

[1743] 10. The system of claim 1, further comprising means for using a natural language processing engine in analyzing the conversation content surrounding the specified keyword.

[1744] (Claim 3)

[1745] 10. The system of claim 1, further comprising: means for sending the generated search query as an HTTP request to an API of a search engine.

[1746] "Application Example 1"

[1747] (Claim 1)

[1748] A means for acquiring speech;

[1749] A means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data;

[1750] A means for analyzing character data and detecting set keywords;

[1751] When a keyword is detected, the conversation content before and after it is analyzed to extract key words and phrases.

[1752] A means for generating search queries based on the extracted words and phrases and sending them to a search engine;

[1753] means for receiving and displaying search results from the search engine on the user's device;

[1754] A means to conduct information search again based on the feedback, and

[1755] A system including:

[1756] (Claim 2)

[1757] 10. The system of claim 1, further comprising means for using a natural language processing engine in analyzing the conversation content surrounding the specified keyword.

[1758] (Claim 3)

[1759] 10. The system of claim 1, further comprising: means for a search engine to provide information based on a search query; and means for using a general-purpose device as a user terminal.

[1760] "Example 2: Combining Emotion Engines"

[1761] (Claim 1)

[1762] A means for acquiring speech;

[1763] A means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data;

[1764] A means for analyzing character data and detecting set keywords;

[1765] When a keyword is detected, the conversation content before and after it is analyzed to extract key words and phrases.

[1766] means for analyzing tone, pitch, and rate of the audio data and recognizing emotions;

[1767] A means for generating search queries based on the emotion information and extracted words and phrases and sending them to a search engine;

[1768] means for receiving and displaying search results from a search engine;

[1769] A system including:

[1770] (Claim 2)

[1771] 10. The system of claim 1, further comprising means for using a language processing engine in analyzing the conversation content surrounding the specified keyword.

[1772] (Claim 3)

[1773] 10. The system of claim 1, further comprising means for the search engine to provide information based on the generated search query.

[1774] "Application example 2 when combining emotion engines"

[1775] (Claim 1)

[1776] A means for acquiring speech;

[1777] A means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data;

[1778] A means for analyzing character data and detecting set keywords;

[1779] When a keyword is detected, the conversation content before and after it is analyzed to extract key words and phrases.

[1780] A means for recognizing emotions based on voice data;

[1781] A means for generating a search query based on the extracted words, phrases and sentiment information and sending the query to a search engine;

[1782] means for receiving and displaying search results from a search engine;

[1783] A system including:

[1784] (Claim 2)

[1785] 10. The system of claim 1, further comprising means for using a natural language processing engine in analyzing the conversation content surrounding the specified keyword.

[1786] (Claim 3)

[1787] 10. The system of claim 1, further comprising means for adjusting the prioritization of search results based on emotion information. [Explanation of symbols]

[1788] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. A means for acquiring speech; A means for transmitting the acquired voice data to a voice recognition engine in real time and converting it into text data; A means for analyzing character data and detecting set keywords; When a keyword is detected, the conversation content before and after it is analyzed to extract key words and phrases. A means for generating search queries based on the extracted words and phrases and sending them to a search engine; means for receiving and displaying search results from a search engine; A system including:

2. 10. The system of claim 1, further comprising means for using a natural language processing engine in analyzing the conversation content surrounding the specified keyword.

3. 10. The system of claim 1, further comprising means for the search engine to provide information based on a search query.

Images