KR102193307B1 - Artificial intelligence automated Response system with customer verification function using speech verification and method thereof - Google Patents

Abstract

The present invention relates to an artificial intelligence automated response system and a method thereof. More specifically, the present invention relates to the artificial intelligence automated response system having a customer authentication function using speaker authentication and the method thereof, wherein a call, control data, and voice of the automated response system can be integrated and relayed, and the voice recognition for the customer′s spoken voice and the speaker authentication are performed at the same time by using the integrated relay so as to provide a service responding to a verified speaker.

Description

Artificial intelligence automated response system with customer verification function using speech verification and method thereof

The present invention relates to an artificial intelligence automatic response system and method, and more particularly, it is possible to integrate and relay the call, control data and voice of the automatic response system, and use the integrated relay to recognize the voice of the customer's spoken voice and It relates to an artificial intelligence automatic response system and method having a customer authentication function using speaker authentication that simultaneously performs speaker authentication and provides a service corresponding to a verified speaker.

In general, one of the important factors that determines the customer's purchase of a certain product or satisfaction of a service is the customer service of a company, government office, etc. to resolve customer complaints and provide quick and quick after service. . Recently, the automatic response system (ARS, or Interactive Voice Response: IVR) is performing the window area that is in charge of customer service at the forefront.

The automatic answering system recognizes the user's voice through the second generation that relied on the function of the switchboard and the automatic answering system, and the third generation that combines telephone and computer, and a computer telephony integration (CTI) automatic answering system, and recognizes the user's voice and responds optimally to the recognized voice. It is evolving into a 4th generation artificial intelligence (AI) automatic response system that automatically selects and responds with voice or text.

The automatic response system can reduce customer service processing costs from the standpoint of a company, not only play the role of a branch office, but also quickly and efficiently obtain customer information, and provides the advantage of providing rapid service. Is developing.

Currently, automatic answering systems in which phone and computer are integrated and artificial intelligence are installed are the mainstream, and such artificial intelligence automatic answering systems include call infrastructure such as switchboard (PBX), CTI, STT (Speech to Text), SV (Speech Verification). ), TTS (Text to Speech), statistics, AI infrastructure such as AI analysis, call infrastructure and AI infrastructure are linked, and IVR, real-time voice processing, and third-party linkage are developed and applied for integration of service scenarios and unit statistics. Are doing.

As described above, the conventional artificial intelligence automatic response system requires a lot of time to implement the service because IVR, real-time voice processing, and third-party interworking must be developed each time to respond to scenarios and statistics according to the applied customer's field. There was a problem in that the investment cost increased.

In addition, it is important to identify and authenticate who the customer is in order to provide the customer with accurate services and information corresponding to the customer, and to provide information corresponding to the customer only to the legitimate customer.

In the conventional artificial intelligence automatic response system, identification information such as a customer's resident registration number and card number is input to identify and authenticate the customer, or when connecting to a consultant, the consultant listens to the customer through voice identification information that can directly identify the customer. Identifying and authenticating customers.

As described above, the conventional AI automatic response system identifies and authenticates customers only by the identification information spoken or input by the customer, so a third party who illegally obtained the customer's identification information can pretend to be a customer with the customer's identification information. There is a problem in that, due to this, services and information can be provided to third parties.

In order to prevent such problems recently, Korean Patent Registration No. 10-1209033 [Customer authentication system and method using speaker recognition of an automatic response system] (hereinafter referred to as "priority technology") is a technology that authenticates customers through speaker recognition. Start.

However, in the prior art, since the voice information on the customer and the customer's voice characteristics is stored in the system, if the customer's voice and voice information is leaked, the voice of the customer can be imitated, and thus, there is a problem that a third party can impersonate a customer.

Korean Patent Registration No. 10-1209033 (2012.12.06.announcement)

Accordingly, an object of the present invention is to be able to relay the call, control data and voice of the automatic response system by integrating, and to respond to the verified speaker by performing speaker authentication and voice recognition for the spoken voice of the customer using the integrated relay. It is to provide an artificial intelligence automatic response system and method having a customer authentication function using speaker authentication that provides a service.

An artificial intelligence automatic response system having a customer authentication function using speaker authentication according to the present invention for achieving the above object includes: voice recognition for received voice, registration of voice prints for each customer, and speaker authentication by registered voice prints. A voice recognition/speaker authentication unit (STT/SV) for performing and transmitting the authentication result information; A response unit for transmitting a response voice to the voice-recognized text data; And, when a call is received from the switchboard (PBX), a call path is formed, and an automatic response service is provided according to the set scenario, but when a virtual counseling service is selected from a registered customer, a voice input in real time to the voice recognition/speaker authentication unit And, when it is determined that the authentication result information received from the voice recognition/speaker authentication unit is successful by comparing the voice print for the real-time voice of the customer and the registered voice print, the scenario through the response unit And an artificial intelligence voice hub (AI VH) that provides a virtual counseling service according to a customer's selection.

The response unit may include an artificial intelligence response unit that receives the speech-recognized text data through the artificial intelligence speech hub unit, searches for a response text corresponding to the text data, and transmits it to the artificial intelligence speech hub unit; And a voice conversion unit (TTS) for receiving the response text through the artificial intelligence voice hub unit and transmitting a response voice converted into voice to the artificial intelligence voice hub unit.

The response unit receives the speech-recognized text data through the artificial intelligence speech hub unit, searches for a response text corresponding to the text data, converts the searched response text into speech, and transmits the speech to the artificial intelligence speech hub unit. It is characterized by being an artificial intelligence response unit.

The artificial intelligence voice hub unit may include a hub communication unit configured to perform data communication and voice communication with the voice recognition/speaker authentication unit and the response unit; A hub storage unit including a scenario DB for storing at least one scenario according to the automatic response service; And setting the scenario according to the automatic response service and setting the interlocking setting of the voice recognition/speaker authentication unit and the response unit, forming a call path when a call is received from the switchboard (PBX), and providing an automatic response service according to the set scenario, Provides voice input in real time to the voice recognition/speaker authentication unit when a virtual counseling service is selected from a customer whose voice print is registered, and recognizes the voice by comparing the voice print for the real-time voice of the customer with the registered voice print. / If it is determined that the authentication result information received from the speaker authentication unit is authentication success, a hub control unit that provides a virtual consultation service according to the scenario and the customer's selection through the response unit and the voice conversion unit.

The hub control unit may include a call processing unit configured to form a call path when a call is received and perform call transfer processing according to a call transfer request; Scenario setting according to the automatic response service and interlocking setting of the voice recognition/speaker authentication unit and response unit are performed, and guidance voice according to the set scenario through the call formed when the call is formed through the call processing unit is provided to the user terminal of the customer A scenario processing unit; A speaker separating unit for separating the customer's voice received during the call and the guidance voice and the response voice according to the scenario and outputting only the customer's voice; A switch mirroring unit that mirrors the customer's voice output from the speaker separation unit and provides it to the voice recognition/speaker authentication unit and the response unit; And the voice recognition/speaker authentication unit and the response unit are interlocked to provide real-time voice input to the voice recognition/speaker authentication unit when a virtual counseling service is selected from a customer registered with a voice print, and the voice for the real-time voice of the customer If it is determined that the authentication result information received from the voice recognition/speaker authentication unit is successful by comparing the print and the registered voice print, the response unit provides a virtual counseling service according to the scenario and the customer's selection. It characterized in that it comprises a link processing unit.

When it is determined that the virtual counseling service is selected through the scenario processing unit, the interlocking processing unit controls the switch mirroring unit to provide a real-time voice input to the speech recognition/speaker authentication unit, and An authentication linking unit for outputting a virtual counseling service start command when the authentication result information is received by comparing the voice print for real-time voice and the registered voice print and it is determined that the authentication result information is successful; And a response linking unit that provides a virtual counseling service according to the scenario and the customer's selection through the response unit when a command to start the virtual counseling service is input from the authentication linking unit.

The system further includes an information analysis unit that stores a call with a call path, a history of use of a virtual counseling service for a corresponding customer, and counseling content information about the virtual counseling service, and stores statistical information and analysis information on the stored information. , The interworking processing unit further comprises a third-party interworking unit that provides information on the call, virtual counseling service scenario information, and consultation content information according to the call upon termination of the call through the call processing unit to the information analysis unit. .

The artificial intelligence voice hub unit includes: an authentication communication unit configured to perform data communication and voice communication with the response unit and the artificial intelligence voice hub unit; An authentication storage unit including a customer voice storage unit for storing customer voices and a voice print DB for storing customer-specific voice prints; And voice recognition for each customer's voice received from the artificial intelligence voice hub unit through the authentication communication unit, registration of a voice print for each customer, and speaker authentication by a registered voice print, and the authentication result information is transmitted to the authentication communication unit. It characterized in that it comprises an authentication control unit for transmitting to the artificial intelligence voice hub through.

The authentication control unit searches the voice print DB in response to a voice print registration confirmation request for the customer of the artificial intelligence voice hub to check whether the customer has registered the voice print, and the voice print registration mode according to whether the voice print is registered. And a mode identification unit for setting one of voice authentication modes. A voice collection unit for collecting the voice provided from the artificial intelligence voice hub unit through an authentication communication unit according to the mode identified by the mode identification unit; A speech recognition unit configured to perform speech recognition on the speech acquired through the speech collection unit and output text data corresponding to the speech; A registration unit that generates a voice print from the voice collected through the voice collection unit in the voice print registration mode, stores it in the voice print DB, and registers it; And generating a voice print from the voice collected through the voice collecting unit in the voice authentication mode, comparing the generated voice print with the registered voice print of the corresponding customer in the voice print DB, and performing speaker authentication according to whether or not they match, And an authentication unit for transmitting the authentication result information to the artificial intelligence voice hub unit through the authentication communication unit.

An artificial intelligence automatic response method having a customer authentication function using speaker authentication according to the present invention for achieving the above object is: The artificial intelligence voice hub forms a call path when incoming calls from the exchange (PBX), and An automatic response service initiation process of starting the automatic response service according to the method; When the customer selects the virtual consultation service, the artificial intelligence voice hub unit provides real-time voice input to the voice recognition/speaker authentication unit if the customer who sent the call is a customer who has registered a voice print, and a voice print for the real-time voice of the customer A speaker authentication process of receiving authentication result information received from the voice recognition/speaker authentication unit by comparing the registered voice print with the registered voice print; And a virtual consultation service process in which the artificial intelligence voice hub unit provides a virtual consultation service according to the scenario and the customer's selection through a response unit and a speech conversion unit when it is determined that the speaker authentication is successful.

In the process of starting the automatic response service, the artificial intelligence voice hub forms a call path when a call is received from the switchboard (PBX), provides voice guidance according to the scenario through the call path, and transmits the customer's voice according to the scenario. A voice mirroring step of receiving and mirroring to the voice recognition/speaker authentication unit; A voice recognition step of the artificial intelligence voice hub unit interworking with a voice recognition/speaker authentication unit to provide text data, which is a voice recognition result of the mirrored customer's voice, to a response unit; A response obtaining step of acquiring a response text for the text data by the artificial intelligence voice hub unit interworking with a response unit; And a response step of converting the obtained response text into speech by the artificial intelligence voice hub unit and providing it to the user terminal of the customer.

The speaker authentication process may include a step of checking whether or not to register a voice print in which the artificial intelligence voice hub unit interworks with a voice recognition/speaker authentication unit to check whether a voice print is registered for the customer who has sent the call; A mode identification step of identifying and setting a voice print registration mode and a voice authentication mode according to whether or not the voice print is registered by a voice recognition/speaker authentication unit linked by the artificial intelligence voice hub unit; A voice mirroring step in which the artificial intelligence voice hub mirrors the voice of the customer to the voice recognition/speaker authentication unit according to the mode identified by the mode identification unit; A voice print registration step of generating a voice print from the voice by a voice recognition/speaker authentication unit interlocked with the artificial intelligence voice hub unit and storing and registering it in a voice print DB when the set mode is a voice print registration mode; And if the set mode is an authentication mode, a voice print is generated from a voice by a voice recognition/speaker authentication unit linked to the artificial intelligence voice hub unit, and the generated voice print and a registered voice print of the corresponding customer in the voice print DB are It characterized by including the authentication step of performing the speaker authentication according to the match or not by comparing the authentication result information.

The present invention has an artificial intelligence voice hub capable of integrating and relaying calls, control data, and voice, so that when a scenario is changed according to a customer company, the changed scenario can be applied by simply setting.

In the present invention, since a changed scenario can be applied with only a simple setting, it is not necessary to develop IVR, real-time voice processing, and third-party interworking every time, thereby saving time and cost required for development.

In addition, the present invention performs speaker authentication for a customer, but automatically generates and registers a speaker voice print for the customer upon initial use by the customer, thereby providing convenience to customers and managers.

In addition, the present invention deletes the collected customer's voice after creation of a voice print, divides the generated voice print into a receipt and a ticket to hold only tickets, and stores and manages the receipt in a reliable external storage means. There is an effect that can improve sex.

1 is a diagram showing the configuration of a communication system including an artificial intelligence automatic response system having a customer authentication function using speaker authentication according to the present invention.
2 is a diagram showing the configuration of an artificial intelligence voice hub of the artificial intelligence automatic response system according to the present invention.
3 is a diagram showing the configuration of a voice recognition/speaker authentication unit of an artificial intelligence automatic response system according to the present invention.
4 is a flowchart illustrating a method of registering a voice print of a customer among artificial intelligence automatic response methods having a customer authentication function using speaker authentication according to the present invention.
5 is a flowchart showing an automatic response method according to customer authentication among artificial intelligence automatic response methods having a customer authentication function using speaker authentication according to the present invention.

Hereinafter, the configuration and operation of an artificial intelligence automatic response system having a customer authentication function using speaker authentication according to the present invention will be described in detail with reference to the accompanying drawings, and an automatic response method through customer authentication by speaker authentication in the system is described. Explain.

1 is a diagram showing the configuration of a communication system including an artificial intelligence automatic response system having a customer authentication function using speaker authentication according to the present invention.

Referring to FIG. 1, an artificial intelligence automatic response system 200 having a customer authentication function using speaker authentication according to the present invention is connected to a user terminal unit 100 used by a customer through a wired/wireless data communication network 10, Data communication and voice communication are performed.

The wired/wireless data communication network 10 includes an Internet network including a Wi-Fi network, a mobile communication network including a 2nd generation (2G), 3G, 4G, 5G, etc., and a general telephone public network (PSTN). And a packet-based next-generation communication network (NGN) coupled to a mobile communication network.

The user terminal unit 100 includes a mobile terminal 100-1 such as a smart phone or a smart pad that wirelessly connects to the wired/wireless data communication network 10, and a wired telephone 100- connected to the wired/wireless data communication network 10 by wire. It could be either of 2).

The automatic answering system 200 includes a computer telephony integration (CTI) 300, a switchboard (PBX) 400, an artificial intelligence voice hub (AI VH) 500, a response unit. 600 and a speech recognition/speaker authentication unit (STT/SV: Speech to Text/Speech Verification) 700, and further include an information analysis unit 800 and an external storage unit 1000 according to an embodiment. I will be able to.

The CTI 300, the switch 400, the AR VH 500, the response unit 600, the STT/SV 700, the information analysis unit 800 is an internal local area network (LAN) 20 It is preferable to be connected through, and the external storage unit 1000 is preferably connected through a wired/wireless data communication network 10.

The CTI 300 acquires information such as the phone number of the user terminal unit 100 of the customer and customer identification information input from the user terminal unit 100, and provides various information using the obtained information, which is the terminal of the consultant. It is displayed on a counselor terminal (not shown) and provides necessary information to other devices.

The PBX 400 switches a plurality of incoming calls to a counselor terminal (not shown) and an AI VH 500 for connection or performs call switching to connect a connected call to another location.

AI VH (500) forms a call path when a call is received from the switch (PBX) (400), configures call infrastructure such as CTI (300), PBX (400), and responds (600), voice recognition/speaker authentication unit It is configured between artificial intelligence infrastructure components such as 700 and information analysis unit 800, and mirrors the voice received by the call infrastructure configuration to the artificial intelligence infrastructure configuration, and configures the call infrastructure and artificial intelligence by transmitting and receiving control data. In conjunction with infrastructure configurations, it provides automatic response service according to the set scenario.

The AI VH 500 acquires one or more of customer identification information and phone number for the customer who sent the call through CTI, and acquired customer identification information and phone number when the virtual consultation service is selected among the scenarios after the start of the scenario. It is checked whether a voice print for the customer corresponding to the number is registered.

If the voice print for the customer is not registered, the AI VH 500 obtains general customer authentication information through the CTI 300 and provides a virtual consultation service scenario for the virtual consultation service selected by the customer after authentication to the customer, In conjunction with the voice recognition/speaker authentication unit (STT/SV) 700, voices according to the virtual counseling service scenario are collected, and voice prints for the collected voices are generated and registered (hereinafter, the registered voice print is referred to as “registered voice”). It is called "print").

In addition, if the customer is a registered customer with voice print, the AI VH 500 mirrors the customer's voice received through the switch 400 in real time and provides it to the STT/SV 700, and provides the voice for the customer's real-time voice. When it is determined that the authentication result information received from the STT/SV 700 is successful by comparing the print and the registered voice print, the virtual counseling service according to the scenario and the customer's selection through the response unit 600 Provides.

The response unit 600 receives speech-recognized text data from the AI VH 500 and transmits a response voice corresponding to the text data to the AI VH 500.

The response unit 600 may include an artificial intelligence response unit 610 and a text to speech (TTS) 620.

The artificial intelligence response unit 610 is an artificial intelligence system such as a chat artificial intelligence response unit (or "chat bot (robot)") and a voice artificial intelligence response unit (or "voice bot (robot)"). , The response text corresponding to the input text data is extracted, and the extracted response text is provided to the AI VH 500 or the response voice is provided to the AI VH 500.

The voice bot may be a configuration in which the TTS 620 is combined with a chat bot.

When the artificial intelligence response unit 610 is a chatbot, the TTS 620 receives a response text, converts it into a response voice, and transmits it to the VI AH 500.

The STT/SV 700 interworks with the AI VH 500 to receive the mirrored customer's voice, perform voice recognition, output the recognized text data, and create and register a registered voice print in the voice print registration mode, In the authentication mode, authentication is performed by comparing the authentication voice print and the registered voice print registered for the customer, and the authentication result information is transmitted to the AI VH 500.

The STT/SV 700 divides the registered voice print according to an embodiment to generate a ticket and a receipt, stores the ticket in an internal storage means, and stores the receipt in the external storage unit 1000 to store the registered voice print. After registration, the receipt of the customer is obtained from the external storage unit 1000 in the authentication mode, combined with the ticket, and restored the registered voice print, and speaker authentication is performed by comparing the restored registered voice print and the authenticated voice print.

The external storage unit 1000 may be a server of an external supervisory authority having high trust and security, or may be a block chain network.

The information analysis unit 800 stores a call with a call path, a history of using a virtual counseling service for a corresponding customer, and counseling content information for a virtual counseling service, generates and stores statistical information about the stored information, and It creates and stores various analysis information, such as information on services that are mainly used.

2 is a diagram showing the configuration of an artificial intelligence voice hub of the artificial intelligence automatic response system according to the present invention. Hereinafter, a detailed configuration and operation of the AI VH 500 according to the present invention will be described with reference to FIG. 2.

The AI VH 500 includes a hub storage unit 510, a hub communication unit 520 and a hub control unit 511.

The hub storage unit 510 is a program area for storing a control program for controlling the overall operation of the AI VH 500 according to the present invention, a temporary area for temporarily storing data generated during execution of the control program, and executing the control program It includes a data area for semi-permanently storing data necessary for the system and data generated while executing a control program. The scenario DB 511 according to the present invention may be configured in the data area.

The scenario DB 511 stores an automatic response service scenario including a virtual counseling service scenario according to the present invention.

The hub communication unit 520 enables data communication and voice communication with the STT/SV 700 and the response unit 600.

The hub control unit 530 includes a call processing unit 540, a scenario processing unit 550, a switch mirroring unit 560, a speaker separation unit 570, and an interlocking processing unit 580, including the AI VH 500 according to the present invention. Control the overall operation.

Specifically, the call processing unit 540 controls the PBX 400 when incoming calls to the PBX 400 to form a call route with the automatic answering system 200, and controls the PBX 400 according to the call transfer request. To perform call transfer processing.

The scenario processing unit 550 provides a scenario providing unit 551 for providing a guide voice according to a scenario set through a call formed through a call formed through the call processing unit 540 to the user terminal unit 100 of the customer, and an automatic response service. It includes a scenario management unit 552 that performs the scenario change and setting according to the STT / SV (700) and the interlocking setting of the response unit (600).

The speaker separation unit 570 separates the customer's voice received during the call and the guidance voice and response voice according to the scenario, and outputs only the customer's voice.

The switch mirroring unit 560 mirrors the customer's voice output from the speaker separation unit 570 and provides it to the STT/SV 700 and the response unit 600.

The interlocking processing unit 580 includes an authentication linking unit 581, a response linking unit 582, and a third-party linking unit 585, and interlocking with the STT/SV 700 and the response unit 600 to print voice. When selecting a virtual counseling service from a registered customer, a voice input in real time is provided to the STT/SV 700, and the STT/SV is performed by comparing the voice print for the real-time voice of the customer and the registered voice print. When it is determined that the authentication result information received from 700 is authentication success, a virtual counseling service according to the scenario and the customer's selection is provided through the response unit 600.

When it is determined that the virtual counseling service is selected through the scenario processing unit 550, the authentication linking unit 581 controls the switch mirroring unit 560 to provide a voice input in real time to the STT/SV 700, When the authentication result information is received from the STT/SV 700 by comparing the voice print for the real-time voice of the customer and the registered voice print and the authentication result information is determined to be successful, the virtual counseling service is started. Print the command.

The response linking unit 582 provides a virtual counseling service according to the scenario and a customer's selection through the response unit 500 when a command to start a virtual counseling service is input from the authentication linking unit 581.

The response linking unit 582 includes an artificial intelligence linking unit 583 for processing an interlocking operation with the artificial intelligence response unit 610 and a speech synthesis linking unit 584 for processing an interlocking operation with the voice conversion unit 620 do.

The third-party linking unit 585 provides information on the call, virtual consultation service scenario information, and consultation content information according to the call when the call is terminated through the call processing unit 540 to the information analysis unit 800, The information analysis unit 800 stores a call with a call path, a history of using a virtual counseling service for a corresponding customer, and information about counseling contents for a virtual counseling service, and stores statistical information and analysis information about the stored information.

3 is a diagram showing the configuration of a voice recognition/speaker authentication unit of an artificial intelligence automatic response system according to the present invention.

Referring to FIG. 3, the STT/SV 700 includes an authentication storage unit 710, an authentication communication unit 720, and an authentication control unit 730.

The authentication storage unit 710 is a program area for storing a control program for controlling the overall operation of the STT/SV 700 according to the present invention, a temporary area for temporarily storing data generated during execution of the control program, and the control It includes a data area for semi-permanently storing data required for program execution and data generated during execution of the control program. In the data area, a customer voice storage unit 711 and a voice print DB 712 may be configured according to the present invention.

The customer voice storage unit 711 collects and stores the voice mirrored from the AI VH 500 for a first time in the voice print registration mode, and stores the voice mirrored from the AI VH 500 for a second time in the authentication mode. Collect and store. It is preferable that the voice used after each time is deleted from the customer voice storage unit 711. It will be preferable that the first time is a time greater than the second time. For example, the first time may be 40 seconds or more, and the second time may be 5 seconds or the like.

The voice print DB 712 stores a registered voice print for each customer, and stores a ticket for each customer obtained by dividing the registered voice print for each customer according to another embodiment.

The authentication control unit 730 includes a mode identification unit 740, a voice collection unit 750, a voice recognition unit 760, a registration unit 770, and an authentication unit 780, and STT/SV 700 according to the present invention. ) To control the overall operation.

Specifically, the mode identification unit 740 searches the voice print DB 712 in response to a voice print registration confirmation request for the customer of the AI VH 500 to check whether the customer has registered the voice print, and One of a voice print registration mode and a voice authentication mode is set according to whether or not a print is registered.

The voice collection unit 750 collects the voice provided from the AI VH 500 through the authentication communication unit 720 according to the mode identified by the mode identification unit 740 and stores the voice in the customer voice storage unit 711. Save it. The voice collection unit 750 deletes the customer's voice stored in the customer voice storage unit 711 when a voice print of the stored voice is generated, registration is completed, or authentication is completed.

The voice recognition unit 760 performs voice recognition on the voice acquired through the voice collection unit 750 and outputs text data corresponding to the voice.

The registration unit 770 generates a registered voice print from the voice collected through the voice collection unit 750 in the voice print registration mode, stores it in the voice print DB 712 and registers it.

The authentication unit 780 generates an authentication voice print from the voice collected through the voice collection unit 750 in the authentication mode, and the generated authentication voice print and the registered voice print of the corresponding customer in the voice print DB 712 The speaker authentication is performed according to whether or not they match, and the authentication result information is transmitted to the AI VH 500 through the authentication communication unit 720.

According to another embodiment, the registration unit 770 may include a registered voice print generation unit 771, a voice print separation unit 772, a ticket registration unit 773, and a receipt registration unit 774.

The registered voice print generation unit 771 generates and encrypts a registered voice print, which is a voice print to be registered, from the voice collected through the voice collection unit 750 in the voice print registration mode.

The voice print separation unit 772 separates the registered voice print into a ticket and a receipt. A method of separating the voice print from a ticket and a receipt may be applied in various ways, such as a method of separating a simple size (1/2 ratio, 1/3 ratio, etc.).

The ticket registration unit 773 stores and registers the divided tickets in the voice print DB 712.

The receipt registration unit 774 transmits the receipt to the external storage unit 1000 through the authentication communication unit 720 and registers the receipt in an external storage unit such as the supervisory authority server 1000 and a blockchain network.

In the case of the above embodiment, the authentication unit 780 includes an authentication voice print generation unit 781, a ticket acquisition unit 782, a receipt acquisition unit 783, a registration voice print combining unit 784, a decoding unit 785, A comparison unit 786 and an authentication result notification unit 787 may be included.

The authentication voice print generation unit 781 generates an authentication voice print that is a voice print to be authenticated from the voice collected from the voice collection unit 750 in the authentication mode.

The ticket acquisition unit 782 obtains a ticket corresponding to the customer who provided the voice to be authenticated from the voice print DB 712.

The receipt acquisition unit 783 obtains a receipt for the customer from the supervisory authority server 1000, a blockchain network, and the like.

The registered voice print combining unit 784 combines the obtained ticket and receipt to generate an encrypted registered voice print.

The decryption unit 785 decodes and outputs the encrypted registered voice print.

The comparison unit 786 compares the decoded registered voice print and the authenticated voice print, performs speaker authentication according to whether they match, and outputs authentication result information.

The authentication result notification unit 787 transmits the authentication result information to the AI VH 500 through the authentication communication unit 720.

4 is a flowchart showing a method of registering a voice print of a customer among artificial intelligence automatic response methods having a customer authentication function using speaker authentication according to the present invention.

Referring to FIG. 4, first, the AI VH 500 monitors whether a call is received through the switch 400 (S111), and when the call is received, the switch 400 is controlled to form a channel, and then according to the service scenario. Guidance voice is provided to the customer through the formed call path (S113). The guide voice at this time will be a guide voice requesting selection of a service menu.

After providing the guidance voice, the AI VH 500 monitors whether a virtual consultation service application is selected (S115), and when a service other than the virtual consultation service is selected, provides the corresponding service according to the service scenario corresponding to the service ( S117).

On the other hand, when the virtual consultation service application is selected, the AI VH 500 interlocks with the STT/SV 700 to check whether speaker authentication registration, that is, a voice print for the customer who sent the call, is registered (S119). Whether or not it is determined (S121).

If registered, the AI VH 500 executes the speaker authentication virtual counseling service according to FIG. 5, which will be described later (S123). The speaker authentication virtual counseling service will be described in detail with reference to FIG. 5 below.

On the other hand, if it is not registered, the AI VH 500 performs a registered voice print registration process as follows.

The AI VH 500 initiates a virtual consultation service by mirroring and providing the voice received from the customer to the STT/SV 700 and STT 620 (S125, S129, S131).

STT/SV 700 starts to collect the customer's voice when the virtual consultation service is started (S127).

Upon the start of the virtual counseling service for registering the registered voice print, the STT/SV 700 informs the customer that the virtual counseling service for registering the registered voice print is currently being provided according to the scenario and collects voices for more than one hour. It would be desirable to guide.

The AI VH 500 checks whether the service is terminated after the start of the virtual consultation service (S133), and when the service is terminated, it notifies the STT/SV 700 that the virtual consultation service is terminated (S135), and the STT/SV The 700 generates and registers a registered voice print for the customer's voice collected until the service is terminated (S137). At this time, the STT/SV 700 may divide the registered voice print into tickets and receipts according to an embodiment.

When divided into tickets and receipts, the STT/SV 700 transmits the receipt storage request information including the receipt to the external storage unit 1000 (S139).

Then, the external storage unit 1000 will store and manage the receipt for each customer in the internal storage means (S141).

On the other hand, the STT/SV 700 stores the ticket in the voice print DB 712 of the internal authentication storage unit 710, and deletes the customer's voice stored in the customer voice storage unit 711 (S143).

5 is a flowchart showing an automatic response method according to customer authentication among artificial intelligence automatic response methods having a customer authentication function using speaker authentication according to the present invention.

5, the AI VH 500 provides a guide voice to the customer according to the virtual counseling service scenario, and mirrors the voice received from the customer to the STT/SV 700 and the TTS 620, thereby providing a virtual Start counseling service (S211, S213, S215).

When the virtual counseling service is started, the ST/SV 700 interworking with the AI VH 500 acquires the customer's voice in real time every second time, and generates an authentication voice print corresponding to the acquired customer's voice. (S217). In the above description, the authentication voice print is generated by acquiring the voice in the time unit in the authentication mode. However, it may be possible to generate the authenticated voice print by acquiring the voice in a word unit or a sentence unit instead of a time unit.

When the authentication voice print is generated, the STT/SV 700 requests and receives a receipt for the customer who sent the call to the external storage unit 1000 (S219, S221), and combines the ticket and the receipt for the customer. After generating a registered voice print, it is decoded (S223).

When the registered voice print is acquired, the STT/SV 700 performs verification by comparing the authenticated voice print and the registered voice print, and determines whether the verification is successful (S225 and S227).

If the verification fails, the STT/SV 700 notifies the authentication result information according to the verification failure to the AI VH 500 and ends (S229). It would be desirable for the AI VH 500 to receive notification of the verification failure to provide a guide voice in a scenario corresponding to the verification failure to the customer.

On the other hand, if the verification is successful, the STT/SV 700 transmits the authentication result information to the AI VH 500 to notify the verification success (S230).

The AI VH 500, which is notified of the success of the verification, provides a virtual consultation service in conjunction with the artificial intelligence response unit 610 and the TTS 620 until the customer normally terminates the service (S237) (S231, S233, S235). ).

Meanwhile, it is common knowledge in the art that the present invention is not limited to the above-described typical preferred embodiment, but can be implemented in various ways without departing from the gist of the present invention. Anyone who has a will be able to understand it easily. If implementation by such improvement, change, substitution or addition falls within the scope of the following appended claims, the technical idea should also be considered to belong to the present invention.

100: user terminal 200: artificial intelligence automatic response system
300: Computer Telephony Integration System (CTI) 400: Switchboard (PBX)
500: artificial intelligence voice hub unit (AI VH) 510: hub storage unit
511: scenario DB 520: hub communication unit
530: hub control unit 540: call processing unit
550: scenario processing unit 560: switch mirroring unit
570: speaker separation unit 580: interlocking processing unit
600: response unit 610: artificial intelligence response unit
620: voice conversion unit 700: voice recognition/speaker authentication unit (STT/SV)
710: authentication storage unit 711: customer voice storage unit
712: voice print DB 720: authentication communication unit
730: authentication control unit 740: mode identification unit
750: voice collection unit 760: voice recognition unit
770: register 780: certification
800: information analysis unit 1000: supervisory authority server

Claims (13)

A voice recognition/speaker authentication unit (STT/SV) for performing voice recognition for the received voice, registering a voice print for each customer, and performing speaker authentication based on the registered voice print, and transmitting the authentication result information;
A response unit for transmitting a response voice to the voice-recognized text data; And
When a call is received from a switchboard (PBX), a call path is formed, and an automatic response service is provided according to the set scenario, but when a virtual counseling service is selected from a customer registered with a voice print, the voice input in real time to the voice recognition/speaker authentication unit is And, when the authentication result information received from the voice recognition/speaker authentication unit is determined to be successful by comparing the voice print for the real-time voice of the customer and the registered voice print, the scenario and An artificial intelligence automatic response system having a customer authentication function using speaker authentication, characterized in that it includes an artificial intelligence voice hub (AI VH) that provides a virtual consultation service according to the customer's selection.
The method of claim 1,
The response unit,
An artificial intelligence response unit that receives the speech-recognized text data through the artificial intelligence speech hub unit, searches for a response text corresponding to the text data, and transmits it to the artificial intelligence speech hub unit; And
Having a customer authentication function using speaker authentication, characterized in that it comprises a voice conversion unit (TTS) that receives the response text through the artificial intelligence voice hub and transmits the response voice converted into voice to the artificial intelligence voice hub. Artificial intelligence automatic response system.
The method of claim 1,
The response unit,
A voice artificial intelligence response unit that receives the speech-recognized text data through the artificial intelligence voice hub unit, searches for a response text corresponding to the text data, converts the searched response text into speech, and transmits the result to the artificial intelligence speech hub unit. Artificial intelligence automatic response system having a customer authentication function using speaker authentication, characterized in that.
The method of claim 1,
The artificial intelligence voice hub unit,
A hub communication unit for performing data communication and voice communication with the voice recognition/speaker authentication unit and the response unit;
A hub storage unit including a scenario DB for storing at least one scenario according to the automatic response service; And
Scenario setting according to the automatic response service and the interlocking setting of the voice recognition/speaker authentication unit and the response unit are performed, a call path is formed when a call is received from the switchboard (PBX), and an automatic response service according to the set scenario is provided, When selecting a virtual counseling service from a registered customer, a voice inputted in real time is provided to the voice recognition/speaker authentication unit, and the voice recognition / by comparing the voice print of the real-time voice of the customer with the registered voice print. Speaker authentication comprising a hub control unit that provides a virtual consultation service according to the scenario and customer selection through the response unit and the voice conversion unit when the authentication result information received from the speaker authentication unit is determined to be authentication success. Artificial intelligence automatic response system with customer authentication function using.
The method of claim 4,
The hub control unit,
A call processing unit that forms a call path when a call is received and performs call transfer processing according to a call transfer request;
Scenario setting according to the automatic response service and interlocking setting of the voice recognition/speaker authentication unit and response unit are performed, and guidance voice according to the set scenario through the call formed when the call is formed through the call processing unit is provided to the user terminal of the customer A scenario processing unit;
A speaker separating unit for separating a customer's voice received during the call and a guide voice and a response voice according to the scenario and outputting only the customer's voice;
A switch mirroring unit that mirrors the customer's voice output from the speaker separation unit and provides it to the voice recognition/speaker authentication unit and the response unit; And
When a virtual counseling service is selected from a customer for whom a voice print is registered in connection with the voice recognition/speaker authentication unit and the response unit, a voice inputted in real time to the voice recognition/speaker authentication unit is provided, and a voice print for the real-time voice of the customer When it is determined that the authentication result information received from the voice recognition/speaker authentication unit is successful by comparing the registered voice print with the registered voice print, the linkage providing a virtual consultation service according to the scenario and the customer's selection through the response unit Artificial intelligence automatic response system having a customer authentication function using speaker authentication, characterized in that it comprises a processing unit.
The method of claim 5,
The interlocking processing unit,
When it is determined that the virtual counseling service is selected through the scenario processing unit, the switch mirroring unit is controlled to provide a real-time voice input to the voice recognition/speaker authentication unit, and the voice for the real-time voice of the customer from the voice recognition/speaker authentication unit. An authentication linking unit configured to output a virtual counseling service start command when the authentication result information is received by comparing the print and the registered voice print and the authentication result information is determined to be successful;
An artificial having a customer authentication function using speaker authentication, characterized in that it includes a response linking unit that provides a virtual counseling service according to the scenario and the customer's selection through the response unit when a command to start a virtual counseling service is input from the authentication linking unit. Intelligent automatic response system.
The method of claim 6,
Further comprising an information analysis unit for storing the call with a call path, the history of use of the virtual consultation service for the customer, and information on the consultation contents for the virtual consultation service, and storing statistical information and analysis information on the stored information,
The interlocking processing unit,
Customer authentication using speaker authentication, characterized in that it further comprises a third-party linking unit that provides information on the call, virtual counseling service scenario information, and consultation content information according to the call to the information analysis unit when the call is terminated through the call processing unit. Artificial intelligence automatic response system with functions.
The method of claim 1,
The artificial intelligence voice hub unit,
An authentication communication unit for performing data communication and voice communication with the response unit and the artificial intelligence voice hub unit;
An authentication storage unit including a customer voice storage unit for storing customer voices and a voice print DB for storing customer-specific voice prints; And
Voice recognition for each customer's voice received from the artificial intelligence voice hub unit through the authentication communication unit, registration of customer-specific voice prints, and speaker authentication by registered voice prints, and the authentication result information through the authentication communication unit. An artificial intelligence automatic response system having a customer authentication function using speaker authentication, characterized in that it comprises an authentication control unit transmitting to the artificial intelligence voice hub unit.
The method of claim 8,
The authentication control unit,
In response to a voice print registration confirmation request for the customer of the artificial intelligence voice hub, the voice print DB is searched to check whether the customer has registered the voice print, and depending on whether the voice print is registered, the voice print registration mode and the voice authentication mode are selected. A mode identification unit to set one;
A voice collection unit collecting voice provided from the artificial intelligence voice hub unit through an authentication communication unit according to the mode identified by the mode identification unit;
A speech recognition unit configured to perform speech recognition on the speech acquired through the speech collection unit and output text data corresponding to the speech;
A registration unit that generates a voice print from the voice collected through the voice collection unit in the voice print registration mode, stores it in the voice print DB, and registers it; And
In the voice authentication mode, a voice print is generated from the voice collected through the voice collection unit, and speaker authentication is performed according to whether or not they match by comparing the generated voice print with the registered voice print of the corresponding customer in the voice print DB. An artificial intelligence automatic response system having a customer authentication function using speaker authentication, comprising: an authentication unit transmitting authentication result information to the artificial intelligence voice hub unit through the authentication communication unit.
An automatic answering service initiation process in which the artificial intelligence voice hub forms a call path when a call is received from a switchboard (PBX) and initiates an automatic answering service according to a set scenario;
When the customer selects the virtual consultation service, the artificial intelligence voice hub unit provides real-time voice input to the voice recognition/speaker authentication unit if the customer who sent the call is a customer who has registered a voice print, and a voice print for the real-time voice of the customer A speaker authentication process of receiving authentication result information received from the voice recognition/speaker authentication unit by comparing the registered voice print with the registered voice print; And
When it is determined that the speaker authentication is successful, the artificial intelligence voice hub unit provides a virtual consultation service according to the scenario and the customer's selection through a response unit and a voice conversion unit. Artificial intelligence automatic response method with customer authentication function using.
The method of claim 10,
The process of starting the automatic response service,
The artificial intelligence voice hub part forms a call path when a call is received from the switchboard (PBX), provides voice guidance according to the scenario through the call path, and receives the customer's voice according to the scenario to recognize the voice/speaker authentication A voice mirroring step of mirroring to a part;
A voice recognition step of the artificial intelligence voice hub unit interworking with a voice recognition/speaker authentication unit to provide text data, which is a voice recognition result of the mirrored customer's voice, to a response unit;
A response acquiring step of acquiring a response text for the text data by interworking with a response unit by the artificial intelligence voice hub unit; And
And a response step of converting the obtained response text into voice by the artificial intelligence voice hub unit and providing it to the user terminal of the customer. 15. An artificial intelligence automatic response method having a customer authentication function using speaker authentication.
The method of claim 10,
The speaker authentication process,
The artificial intelligence voice hub unit interworking with the voice recognition/speaker authentication unit to check whether or not to register the voice print for the customer who sent the call;
A mode identification step of identifying and setting a voice print registration mode and a voice authentication mode according to whether or not the voice print is registered by a voice recognition/speaker authentication unit linked by the artificial intelligence voice hub unit;
A voice mirroring step in which the artificial intelligence voice hub mirrors the customer's voice to the voice recognition/speaker authentication unit according to the mode identified by the mode identification unit;
A voice print registration step of generating a voice print from the voice by a voice recognition/speaker authentication unit interlocked with the artificial intelligence voice hub unit and storing and registering it in a voice print DB when the set mode is a voice print registration mode; And
If the set mode is an authentication mode, a voice print is generated from the voice by the voice recognition/speaker authentication unit linked to the artificial intelligence voice hub unit, and the generated voice print is compared with the registered voice print of the corresponding customer in the voice print DB. Thus, the artificial intelligence automatic response method having a customer authentication function using speaker authentication, characterized in that it comprises an authentication step of performing speaker authentication according to the match or not, and the authentication result information.
The method of claim 10,
The artificial intelligence voice hub unit interlocks with the information analysis unit to store the call for which the call path is formed, the history of using the virtual counseling service for the customer, and information about the counseling content for the virtual counseling service, and store statistical information and analysis information on the stored information. Artificial intelligence automatic response method having a customer authentication function using speaker authentication, characterized in that it further comprises a storing information analysis process.

Images