TWI767339B - Voice service system and method - Google Patents

Abstract

A service server of a voice service system establishes voice communication between itself and a first electronic device and outputs a query message through the first electronic device, when the service server receives a use message from the first electronic device When the user's voice data is used, a natural language processing model is used to determine whether the user's voice data matches the query message. When the judgment result is yes, the service server edits a form for online service programs according to the user's voice data. After the judgment result of the service server is no and it is also judged that the user's voice data indicates a demand, the service server uses the natural language processing model to generate a reply message corresponding to the user's voice data and pass The first electronic device outputs.

Description

Voice service system and method

The present invention relates to a voice service system and method, in particular to a voice service system and method using natural language processing technology.

In recent years, although voice services on the phone and the Internet have gradually become popular, the existing voice services can only output voice messages and provide options according to the preset process. That is to say, when users use the existing voice services, Voice messages are often only passively listened to and selected from the limited options provided by voice services. Therefore, the existing voice service lacks flexibility and cannot meet the needs of users outside the preset process, and there is still inconvenience in use.

One of the objects of the present invention is to provide a voice service system which can improve the inconvenience of the prior art.

The voice service system of the present invention includes a first electronic device and a service server for electrically connecting the first electronic device and storing a natural language processing model. The service server establishes a voice communication between itself and the first electronic device, and outputs a query message through the first electronic device. When the service server receives a user's voice data from the first electronic device, the natural language processing model is used to determine whether the user's voice data matches the query message. When the judgment result of the service server is yes, the service server edits a form data for the online service program according to the user's voice data. After the judgment result of the service server is no, and it is also judged that the user's voice data indicates a demand, the service server uses the natural language processing model to generate a reply message corresponding to the user's voice data, and sends a reply message through the user's voice data. The first electronic device outputs the reply message.

In some implementation aspects of the voice service system of the present invention, the user's voice data is generated by the first electronic device according to the user's voice input when it receives a user's voice input.

In some implementation aspects of the voice service system of the present invention, when the service server receives the user's voice data during the process of outputting the query message through the first electronic device, the service server stops outputting the query message .

In some implementation aspects of the voice service system of the present invention, the voice service system further includes a second electronic device for electrically connecting to the service server. Wherein, when the service server determines that the user's voice data does not match the query message, and also determines that a transfer condition related to the user's voice data is satisfied, the service server sends a transfer request to The second electronic device establishes a voice communication between the first electronic device and the second electronic device when receiving an approval transfer instruction corresponding to the transfer request from the second electronic device.

In some implementation aspects of the voice service system of the present invention, the transfer condition includes that the service server determines that a reply message corresponding to the user's voice data cannot be generated, and the transfer request includes a request for the second The insurance application progress data output by the electronic device in a display manner, and the insurance application progress data indicates one of the current progress corresponding to the inquiry message among a plurality of progresses of a form data collection program.

Another object of the present invention is to provide a voice service method implemented by the service server of the voice service system.

The voice service method of the present invention is implemented by a service server to a first electronic device, the service server stores a natural language processing model, and the voice service method comprises: the service server establishes a relationship between itself and the first electronic device The service server outputs a query message through the first electronic device; when the service server receives a user's voice data from the first electronic device, the natural language processing model is used to determine the user Whether the voice data matches the query message, when the judgment result of the service server is yes, the service server edits a form data for the online service program according to the user voice data, and the service server judges The result is no, and it is also determined that the user's voice data indicates a demand, the service server uses the natural language processing model to generate a reply message corresponding to the user's voice data, and outputs the response message through the first electronic device Reply to a message.

In some implementation aspects of the voice service method of the present invention, the user voice data is generated by the first electronic device according to the user voice input when receiving a user voice input.

In some implementation aspects of the voice service method of the present invention, when the service server receives the user's voice data during the process of outputting the query message through the first electronic device, the service server stops outputting the query message .

In some implementation aspects of the voice service method of the present invention, when the service server determines that the user's voice data does not match the query message, and also determines that a transfer condition related to the user's voice data is satisfied , the service server transmits a transfer request to a second electronic device, and upon receiving an approval transfer command from the second electronic device corresponding to the transfer request, establishes the relationship between the first electronic device and the second electronic device Voice communication between two electronic devices.

In some implementation aspects of the voice service method of the present invention, the transfer condition includes that the service server determines that a reply message corresponding to the user's voice data cannot be generated, and the transfer request includes a request for the second The insurance application progress data output by the electronic device in a display manner, and the insurance application progress data indicates one of the current progress corresponding to the inquiry message among a plurality of progresses of a form data collection program.

The effect of the present invention is that the service server can use the natural language processing module to generate the reply message and output it through the first electronic device when it determines that the user's voice data does not match the inquiry message and indicates the demand, thereby Therefore, the service server can allow the user to freely ask questions in the voice service process, so as to satisfy the user's initiative and needs outside the preset voice service process, so it can indeed improve the inconvenience of the prior art.

Before the present invention is described in detail, it should be noted that the "electrical connection" mentioned in this patent specification generally refers to a wired electrical connection between a plurality of electronic devices/devices/elements connected through conductive materials, as well as through wireless communication. Technology A radio connection for wireless signal transmission. In addition, the "electrical connection" mentioned in this patent specification also generally refers to the "direct electrical connection" formed by the direct connection between two electronic devices/devices/components, and the two electronic devices/devices/components are also connected through An "indirect electrical connection" formed by connecting other electronic equipment/devices/components.

Referring to FIG. 1 , an embodiment of a voice service system 1 of the present invention includes, for example, a service server 10 , a first electronic device 11 for electrically connecting to the service server 10 via a network, and a first electronic device 11 for electrically connecting to the service server 10 via a network The second electronic device 12 of the service server 10 is connected.

The service server 10 stores a natural language processing (Natural Language Processing, NLP for short) model M. More specifically, in this embodiment, the natural language processing model M is, for example, implemented as a natural language processing neural network. network, and the natural language processing neural network is trained by, for example, machine learning, but not limited thereto. Further, in this embodiment, the natural language processing neural network may include, for example, a speech recognition module, a semantic analysis module, and a dialogue generation module. Wherein, the speech recognition module is implemented by, for example, speech recognition technology (Automated Speech Recognition in English, ASR for short), and the semantic analysis module is implemented by, for example, natural language understanding technology (Natural Language Understanding in English, NLU for short) , and the dialogue generation module may be implemented by, for example, natural language generation technology (Natural Language Generation in English, NLG for short), but not limited thereto.

In this embodiment, the first electronic device 11 is implemented as a smart phone, for example, and the second electronic device 12 is implemented as a desktop computer, for example, however, in similar implementations , each of the first electronic device 11 and the second electronic device 12 can be implemented as one of a smart phone, a tablet computer, a notebook computer and a desktop computer, but is not limited to this embodiment.

Further, in order to facilitate the description of the application of this embodiment, the first electronic device 11 is used as a client electronic device 11 in this embodiment, and the second electronic device 12 is used as a server in this embodiment. Electronic device 12 . In this embodiment, the client electronic device 11 belongs to a user, for example, and the service server 10 and the server electronic device 12 are managed by, for example, a business organization that provides online services, and the service The terminal electronic device 12 is, for example, used for operation by a service specialist working in the business organization, but not limited thereto.

More specifically, in the application of this embodiment, the business organization is, for example, a financial gold purchase, and the online service provided by the financial gold purchase is, in this embodiment, for example, a travel insurance for the user The online insurance application service of Ping An Insurance, and the online insurance application service is realized by, for example, the service server 10 implementing a voice service method on the client electronic device 11 , but not limited thereto.

Referring to FIG. 2 , how the service server 10 of this embodiment implements the voice service method on the client electronic device 11 is exemplarily described in detail below.

First, in step S1, when the service server 10 receives a voice service call request from the client electronic device 11, the service server 10 accepts the voice service call request and establishes itself with the client electronic device A two-way instant voice communication between devices 11. More specifically, the voice service call request may be generated by the client electronic device 11 according to the operation of the user and sent to the service server 10, for example, and the two-way real-time voice communication may be, for example, the service server 10 It is established based on the Public Switched Telephone Network (PSTN in English) or, for example, the Voice over Internet Protocol (Voice over Internet Protocol in English, VoIP for short), but not based on this limited. Moreover, in similar applications of this embodiment, the two-way instant voice communication may also be established by the service server 10 actively sending a voice call request to the client electronic device 11, instead of the present embodiment limit.

Next, in step S2 following step S1, by means of the two-way real-time voice communication, the service server 10 outputs a voice-presented query message through the client electronic device 11 for the user to listen to. Taking this embodiment as an example, the inquiry message may be, for example, "Which country are you going to travel?", but it is not limited thereto.

Next, in step S3, when the service server 10 receives a user's voice data from the client electronic device 11, the service server 10 uses the natural language processing model M to determine whether the user's voice data is the same as the user's voice data. The query message matches. In this embodiment, the user voice data can be implemented as a piece of audio data, for example, however, in other embodiments, the user voice data can also be implemented as a piece of text data converted from voice, and It is not limited to this embodiment.

More specifically, in this embodiment, the user voice data is, for example, generated by the client electronic device 11 when the client electronic device 11 receives a user voice input through a radio function according to the user voice input, and is generated by the client The electronic device 11 provides the service server 10 through the two-way instant voice communication. It is worth mentioning that, in this embodiment, if the service server 10 receives the user's voice data during the process of outputting the inquiry message through the client electronic device 11, the service server 10 will stop Output the query message, in other words, if the service server 10 receives the user's voice data before the query message is completely output, it means that the user starts to speak before the query message is output ( It is equivalent to generating the user's voice input to the client electronic device 11 ), at this time, the service server 10 of this embodiment will stop outputting the query message, but will continue to receive the use from the client electronic device 11 . Therefore, the present embodiment can achieve the function of allowing the user to actively interrupt the computer voice.

Further, in this embodiment, the service server 10 uses, for example, the speech recognition module and the semantic analysis module of the natural language processing model M to analyze the semantics expressed by the user's speech data, and determine the Whether the semantics of the user's speech data indicates an answer that semantically logically matches the query message. For example, in the case where the query message is "Which country are you going to travel?", if the user's voice data is "I am going to Japan", "Going to the United States for business trip" or "Korea", the query is made If the message responds with semantic logic, the service server 10 will determine that the user's voice data matches the query message. "?" or "Can inconvenience insurance be added by the way?" and the query message is not answered with semantic logic, the service server 10 will determine that the user's voice data does not match the query message.

If the service server 10 determines that the user's voice data matches the query message, the process proceeds to step S4. On the other hand, if the service server 10 determines that the user's voice data does not match the query message, the process proceeds to step S4. Then go to step S5.

In step S4 following step S3, once the service server 10 determines that the user's voice data matches the query message, the service server 10 edits an online service program based on the user's voice data. form data. For the application of this embodiment, the form data may be, for example, a policy data for applying for travel safety insurance, and the online service program may be, for example, a process to be performed by the service server 10 according to the policy data. Online insurance application procedures, but not limited to this. In addition, taking the user's voice data as "I want to go to Japan", the service server 10 edits the form data (in this embodiment, the policy data), for example, to represent "Japan" The information of is added to a travel location field of the form data, but not limited thereto.

In step S5 subsequent to step S3, once the service server 10 determines that the user's voice data does not match the query message, the service server 10 uses the natural language processing model M to further determine a data related to the usage Whether the transfer conditions for the voice data of the recipient are met.

In this embodiment, the service server 10, for example, firstly uses the natural language processing model M to determine whether the semantics of the user's voice data indicates a demand, and if the judgment result is yes, the service server 10 It is further judged whether the natural language processing model M can be used to generate a reply message for responding and satisfying the requirement. In this embodiment, the transfer condition, for example, means that the service server 10 determines that it cannot generate a reply message corresponding to the user's voice data by using the natural language processing model M. More specifically, when the service server 10 The device 10 determines that the semantics of the user's voice data does not indicate any demand, or determines that although the user's voice data indicates a demand, the natural language processing model M cannot be used to generate a reply for responding to the demand. message, the service server 10 determines that the transfer condition is met, but not limited thereto.

To give a more specific example, suppose that the user's voice data is "Can multiple people apply for insurance?" or "Can I add inconvenience insurance by the way?", and suppose that the online insurance service supports online multiple people to apply for insurance and add insurance Inconvenient function, the service server 10 will determine that the user's voice data indicates a demand and can generate a reply message corresponding to the user's voice data, and determine that the transfer condition is not met. For another example, if the user's voice data is "Which country is more fun?" or "Please help me transfer to a customer service person", the service server 10 will determine that although the user's voice data indicates a demand, It is not possible to use the natural language processing model M to generate a reply message that can meet the demand indicated by the user's voice data, and it is determined that the transfer condition is met.

If the service server 10 determines that the transfer condition is met, the flow proceeds to step S6, on the other hand, if the service server 10 determines that the transfer condition is not met, the flow proceeds to step S8.

In step S6 following step S5, once the service server 10 determines that the transfer condition is met, the service server 10 sends a transfer request to the server electronic device 12 for the server electronic device 12 outputs the transfer request. It is worth noting that, in this embodiment, the transfer request includes, for example, a call transfer notification for the server electronic device 12 to output by playing a sound, and a call transfer notification for the server electronic device 12 to output The insurance application progress data output in the display mode, and the insurance application progress data indicates, for example, one of the current progress corresponding to the inquiry message among the multiple progresses of a form data collection program, whereby the insurance application progress data can allow the service specialist Real-time knowing in which stage of the form data collection process the service server 10 determines that the transfer condition is met, that is, letting the service specialist know the voice communication between the service server 10 and the user in real time Which progress is facing obstacles, but not limited to this.

Taking the application of “Applying for Travel Ping An Insurance” in this embodiment as an example, the form data collection program is, for example, a policy data collection program used to collect insurance application data of Travel Ping An Insurance, wherein the progress of the policy data collection program For example, it has a predetermined sequence, and the progress includes, for example, a member information confirmation progress for confirming the user's membership, and a travel destination for confirming the country the user wants to travel. Confirmation progress, one is used to confirm the number of insured persons to be insured by the user, one is used to confirm the progress of the travel start and end time when the user starts and ends the travel, one is used for the user to confirm The progress of confirming the correctness of the data currently collected, one is the progress of recommending the insurance plan for recommending the insurance plan to the user, one is the progress of confirming the demand for medical planes for confirming whether the user needs medical plane services, one is using For the user to confirm the progress of the trial calculation and confirmation of the premium in the trial calculation of the total premium The correct insurance content announces the confirmation progress, but it is not limited to this. In addition, when the inquiry message is "Which country are you going to travel to?" as an example, the service server 10, for example, takes the confirmation progress of the travel location that should be inquired as the current progress, that is, in In the case where the inquiry message is "Which country are you going to travel?" as an example, the insurance application progress data, for example, indicates the confirmation progress of the travel location, so that the service specialist can instantly know that the service server 10 is in The user encountered a communication barrier in confirming the travel location, but not limited to this.

Next, in step S7 following step S6, when the service server 10 receives an approval transfer instruction from the server electronic device 12 corresponding to the transfer request, the service server 10 creates the client electronic device The two-way real-time voice communication between the device 11 and the server-side electronic device 12 enables the service specialist and the user to conduct real-time two-way voice chat through the client-side electronic device 11 and the server-side electronic device 12 . Wherein, the approval transfer instruction may be generated by the server electronic device 12 according to the operation of the service specialist and sent to the service server 10, for example, but not limited thereto. It is worth mentioning that, since the service specialist can know the current progress of the form data collection procedure through the insurance application progress data before the consent transfer instruction is sent, this embodiment can effectively assist the service specialist in Before making a call with the user, enter the status quickly, and then take over the call with the user from the service server 10 .

In step S8 following step S5, once the service server 10 determines that the transfer condition is not met, the service server 10 uses the natural language processing model M to generate a reply message corresponding to the user's voice data , and output the reply message through the client electronic device 11 for the user to listen to. For example, if the user's voice data is "Can multiple people apply for insurance?", the reply message generated by the service server 10 can be, for example, "No problem, I will confirm the number of people applying for insurance later", For another example, if the user's voice data is "Can you add inconvenience insurance by the way?", the reply message generated by the service server 10 can be, for example, "No problem, inconvenience insurance will be explained to you later. "Insurance Information", but not limited to this.

The above is an exemplary description of the voice service method of this embodiment.

It should be added that, after step S4 in this embodiment, the service server 10 will, for example, output the next inquiry message (for example, "May I ask you to ask me") through the client electronic device 11 according to the sequence of the progress of the policy data collection procedure Do you want to insure one person or multiple people?", but not limited to). On the other hand, after step S8 in this embodiment, the process starts again from step S2, for example, to output the query message that the user has not answered yet. On the other hand, the process of the online insurance application service ends, for example, after the service server 10 executes step S7, or after the form data is edited by the service server 10, and the process of the online insurance application service ends Afterwards, the service server 10 will, for example, automatically perform subsequent operations such as underwriting, requesting money, transmitting and recording policy information according to the edited form data. However, the aforementioned subsequent operations are not the technical focus of this patent specification, so the This will not be described in detail.

In addition, although this embodiment takes the application of "online insurance" as an example to describe the voice service system 1 and the voice service method, it should be understood that the voice service system 1 and the voice service method can also be implemented applications in other areas. More specifically, the service server 10 and the server-side electronic device 12 are not limited to being managed by the financial gold purchase, and the voice service method is not limited to realize online insurance service, but can also be applied to positioning, for example , purchasing tickets, ordering commodities, consulting and other services that can be implemented by using computer voice, therefore, the actual implementation of the voice service method is not limited to this embodiment.

To sum up, by implementing the voice service method, the service server 10 of this embodiment can further determine whether the transfer condition is met when it is determined that the user's voice data does not match the query message, and in the When the transfer condition is not met, the reply message is generated and output through the client electronic device 11, and when the transfer condition is met, the transfer request is sent to the server electronic device 12 to establish the client electronic device 11 and the client electronic device 11. In the voice communication between the server electronic devices 12, by using the natural language processing model M to generate the reply message and output it through the client electronic device 11, the service server 10 can provide the user with the process of voice service Freely ask questions in the middle, so as to meet the needs of users that are outside the preset voice service process, so it can indeed improve the inconvenience of the prior art and achieve the purpose of the present invention. Sending the transfer request indicating the current progress to the server electronic device 12, the service server 10 can effectively assist the service specialist to quickly understand the situation, and can more smoothly take over the service proposed by the user but cannot be served by the service Server 10 addresses the needs.

However, the above are only examples of the present invention, and should not limit the scope of the present invention. Any simple equivalent changes and modifications made according to the scope of the application for patent of the present invention and the content of the patent specification are still within the scope of the present invention. within the scope of the invention patent.

1•••••• Voice service system 10・・・・ Service Server 11··········The first electronic device (client electronic device) 12••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• Second electronic device M········Natural language processing model S1~S8•••••• Steps

Other features and effects of the present invention will be clearly presented in the embodiments with reference to the drawings, wherein: 1 is a block diagram of an embodiment of a voice service system of the present invention; and FIG. 2 is a flow chart illustrating how a service server of this embodiment implements a voice service method.

1•••••• Voice service system 10・・・・ Service Server 11··········The first electronic device (client electronic device) 12••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• Second electronic device M········Natural language processing model

Claims (8)

A voice service system, comprising: a first electronic device; and a service server for electrically connecting the first electronic device and storing a natural language processing model, the service server establishes itself with the first electronic device voice communication between devices, and output a query message through the first electronic device, when the service server receives a user's voice data from the first electronic device, use the natural language processing model to determine the user's voice Whether the data matches the query message, when the judgment result of the service server is yes, the service server edits a form data for the online service program according to the user's voice data, and the judgment result of the service server is If no, and after judging that the user's voice data indicates a demand, the service server uses the natural language processing model to generate a reply message corresponding to the user's voice data, and outputs the reply through the first electronic device The message, wherein when the service server receives the user voice data during the process of outputting the query message through the first electronic device, the service server stops outputting the query message and continues to receive the user voice data. The voice service system of claim 1, wherein the user's voice data is generated by the first electronic device according to the user's voice input when it receives a user's voice input. The voice service system according to claim 1, further comprising a second electronic device for electrically connecting to the service server, wherein when the service server determines that the user's voice data does not match the query message, and also judge When a transfer condition related to the user's voice data is satisfied, the service server sends a transfer request to the second electronic device, and receives a transfer request from the second electronic device corresponding to the transfer request. When the transfer instruction is agreed, the voice communication between the first electronic device and the second electronic device is established. The voice service system of claim 3, wherein the transfer condition includes that the service server determines that a reply message corresponding to the user's voice data cannot be generated, and the transfer request includes a request for the second The insurance application progress data output by the electronic device in a display manner, and the insurance application progress data indicates one of the current progress corresponding to the inquiry message among a plurality of progresses of a form data collection program. A voice service method is implemented on a first electronic device by a service server, the service server stores a natural language processing model, and the voice service method comprises: the service server establishes a relationship between itself and the first electronic device The service server outputs a query message through the first electronic device; when the service server receives a user's voice data from the first electronic device, the natural language processing model is used to determine the user Whether the voice data matches the query message, when the judgment result of the service server is yes, the service server edits a form data for the online service program according to the user voice data, and the service server judges The result is no, and after it is also determined that the user's voice data indicates a demand, the service server uses the natural language processing model to generate a pair that should be used the reply message of the user's voice data, and output the reply message through the first electronic device, wherein, when the service server receives the user's voice data in the process of outputting the query message through the first electronic device, The service server stops outputting the inquiry message and continues to receive the user's voice data. The voice service method of claim 5, wherein the user's voice data is generated by the first electronic device according to the user's voice input when it receives a user's voice input. The voice service method according to claim 5, wherein when the service server determines that the user's voice data does not match the query message, and also determines that a transfer condition related to the user's voice data is satisfied , the service server transmits a transfer request to a second electronic device, and upon receiving an approval transfer command from the second electronic device corresponding to the transfer request, establishes the relationship between the first electronic device and the second electronic device Voice communication between two electronic devices. The voice service method of claim 7, wherein the transfer condition includes that the service server determines that a reply message corresponding to the user's voice data cannot be generated, and the transfer request includes a The insurance application progress data output by the electronic device in a display manner, and the insurance application progress data indicates one of the current progress corresponding to the inquiry message among a plurality of progresses of a form data collection program.

Images