Voice communication device and voice communication program

Abstract

In voice call systems, this improves the flexibility of voice response scenarios and the versatility of their applications. [Solution] The voice call device according to the present invention is a voice call device that assists the caller in achieving their objective through a voice call with a user, and comprises: a first prompt input means for inputting a prompt to a language model to guide the caller toward achieving their objective; an answer acquisition means for obtaining an answer including a question to the user from the language model into which the prompt has been input; a voice conversion means for converting the first text of the question into a first call voice; a call voice transmission means for transmitting the first call voice to the user's telephone; a call voice receiving means for receiving a second call voice from the telephone in response to the first call voice; a text conversion means for converting the second call voice into a second text; and a second prompt input means for inputting a prompt including the second text to a language model.

Description

Technical Field

[0001] The present invention relates to a voice call device and a voice call program.

Background Art

[0002] For example, in the case of a manned call center system, there is a limit to the number of operators who can handle calls simultaneously, and it may not be possible to handle all incoming calls. To address this, automatic voice response systems, such as those called IVR and voice bots, have been known. By playing voice guidance to customers and having them press phone buttons, they can select requirements or answer questions to reach the target information or the responsible department (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the above conventional automatic voice response system, depending on the reason for the call and the inquiry requirements of the customer, there are cases where the question branching of the voice response (voice guidance / senario) becomes long and the customer has to press buttons many times. In addition, since there are only leads for the scenarios of voice responses assumed in advance and it is limited to fixed and stereotyped requirement responses, it may lack flexibility in scenarios and versatility in usage applications.

[0005] The present invention has been proposed in view of the above points, and in one aspect, aims to improve the flexibility of voice response scenarios or the versatility of usage applications in a voice call system.

Means for Solving the Problems

[0006] To solve the above problems, the present invention provides a voice call device that assists a caller in achieving their objective through voice communication with a user, comprising: a first prompt input means for inputting a prompt to a language model to guide the caller toward achieving the objective; an answer acquisition means for obtaining an answer including a question to the user from the language model into which the prompt has been input; a voice conversion means for converting the first text of the question into a first call voice; a call voice transmission means for transmitting the first call voice to the user's telephone; a call voice receiving means for receiving a second call voice from the telephone in response to the first call voice; a text conversion means for converting the second call voice into a second text; and a second prompt input means for inputting a prompt including the second text to the language model. [Effects of the Invention]

[0007] According to embodiments of the present invention, the flexibility of voice response scenarios and the versatility of their applications can be improved in a voice call system. [Brief explanation of the drawing]

[0008] [Figure 1] This figure shows an example of the network configuration of the voice call system according to this embodiment. [Figure 2] This figure shows an example of the hardware configuration of the voice call server according to this embodiment. [Figure 3] This figure shows an example of the software configuration of the voice call system according to this embodiment. [Figure 4] This is a sequence diagram related to a voice call system. [Figure 5] This figure shows an example (part 1) of a text dialogue between the voice call server and LLM according to this embodiment. [Figure 6] This figure shows an example (part 1) of a voice call between a user and a voice call server according to this embodiment. [Figure 7] This figure shows an example (part 2) of a text dialogue between the voice call server and LLM according to this embodiment. [Figure 8] This figure shows an example (part 2) of a voice call between a user and a voice call server according to this embodiment. [Figure 9] This is a sequence diagram relating to the voice call system according to this embodiment. [Figure 10] This figure shows an example (part 3) of a text dialogue between the voice call server and LLM according to this embodiment. [Figure 11] This figure shows an example (part 4) of a text dialogue between the voice call server and LLM according to this embodiment. [Figure 12] This figure shows an example (part 3) of a voice call between a user and a voice call server according to this embodiment. [Figure 13] This figure shows an example (part 4) of a voice call between a user and a voice call server according to this embodiment. [Figure 14] This is a sequence diagram relating to the voice call system shown in this modified example. [Figure 15] This figure shows an example (part 5) of the text dialogue between the voice call server and LLM related to this modified example. [Modes for carrying out the invention]

[0009] Embodiments of the present invention will be described in detail with reference to the drawings. [Embodiment 1] <System Configuration> (Network configuration) Figure 1 shows an example of the network configuration of a voice call system according to this embodiment. The voice call system 100 in Figure 1 includes a voice call server 10, an LLM 20, and a telephone 30, which are connected via a network 60.

[0010] The voice call server 10 is a server device that is used, for example, as a reception system in a call center to enable voice calls with users. In particular, the voice call server 10 according to this embodiment is an AI telephone server that responds to user inquiries by obtaining answers to those inquiries via the LLM 20 when it receives a voice inquiry from a user.

[0011] The LLM (Large Language Model) 20 is a language model that further learns (fine-tuning or RAG) a knowledge base that aggregates knowledge documents (such as product manuals, service manuals, brochures, QA sets, etc.) for answering inquiries from users for a general-purpose natural language processing model (LLM: Large Language Models) that has pre-learned grammar, word meanings, etc. using a large amount of text data. Note that for the general-purpose natural language processing model, for example, language models (generative AI) based on BERT (Bidirectional Encoder Representations from Transformers), GPT (Generative Pre-trained Transformer)-3.5, GPT-4, etc. can be used.

[0012] The telephone 30 is a telephone capable of voice calls, such as a general telephone, mobile phone, smartphone, IP phone, etc., and is a calling device of a user (caller) who uses the voice call system 100. The user can make a voice call inquiry by making a call to a predetermined telephone number using the telephone 30 to connect to the voice call server 10. Note that when the telephone 30 is particularly a smartphone, a predetermined telephone application program is installed in advance. [[ID=VIII]] [[ID=IX]]

[0013] [[ID=⑩]] [[ID=⑪]] [[ID=⑫]]

[0014] [[ID=⑬]] [[ID=⑭]](Hardware Configuration)[[ID=⑮]] ​Figure 2 shows an example of the hardware configuration of a voice call server according to this embodiment. The voice call server 10 includes a CPU (Central Processing Unit) 11, ROM (Read Only Memory) 12, RAM (Random Access Memory) 13, HDD (Hard Disk Drive) 14, and a communication device 15.

[0015] The CPU 11 executes various programs and performs calculations. The ROM 12 stores programs necessary for startup. The RAM 13 is a work area that temporarily stores processing by the CPU 11 and stores data. The HDD 14 stores various data and programs. The communication device 15 communicates with other devices via the network 60.

[0016] (Software configuration) Figure 3 shows an example of the software configuration of the voice call system according to this embodiment.

[0017] The voice call server 10 has, as its main functional units, a call receiving unit 101, a prompt input unit 102, a response acquisition unit 103, a text-to-speech conversion unit 104, and a call audio transmission / reception unit 105.

[0018] The call receiving unit 101 has the function of receiving calls from the user's telephone and making calls to the user's telephone.

[0019] The prompt input unit 102 has the function of inputting a goal seek prompt to the language model. The prompt input unit 102 also has the function of inputting a prompt containing text (user response) to the language model.

[0020] The response acquisition unit 103 has the function of acquiring a response, including a question sentence (AI question sentence) for the user, from a language model into which a goal seek prompt has been input.

[0021] The text-to-speech conversion unit 104 has the function of converting text to voice calls or voice calls to text.

[0022] The call audio transmission / reception unit 105 has the function of transmitting call audio (AI question audio) to the telephone and receiving call audio (user response audio) from the telephone.

[0023] Each functional unit is implemented by a program executed on the hardware resources such as the CPU, ROM, and RAM of the computer constituting the voice call server 10. These functional units may be read as "means," "modules," "units," or "circuits." Furthermore, each functional unit can be placed on a separate device on the network 60. The computer program of the voice call server 10 may be stored on a storage medium that is readable by the computer.

[0024] <System Flow> Figure 4 is a sequence diagram relating to the voice call system according to this embodiment. Figure 5 is a diagram showing an example (part 1) of text dialogue between the voice call server and LLM according to this embodiment. Figure 6 is a diagram showing an example (part 1) of voice call between the user and the voice call server according to this embodiment. Below, we will explain in detail, following Figure 4 and referring to Figures 5 and 6 as appropriate, using an example where a user makes a telephone inquiry to a call center regarding a printer malfunction. Regarding the voice call server 10, each step (hereinafter referred to as "S") can be realized by the CPU 11 loading and executing a program capable of realizing this sequence.

[0025] S1: The user making the call uses the telephone 30 to dial a designated telephone number (for example, the telephone number of a call center).

[0026] S2: The voice call server 10 receives a call from a user, establishes a voice call, and inputs (sends) a predetermined goal seek prompt 70a to the LLM20 to prompt the LLM20 to respond to the user's inquiry / requirement.

[0027] For example, as shown in Figure 5, the goal-seek prompt 70a is a type of prompt that instructs the LLM20. It can guide the user towards achieving their goal (sometimes with the LLM20 creating prompts itself) by repeatedly asking questions, such as resolving a request the user wants to inquire about. In other words, the goal-seek prompt 70a is a prompt used to set a goal to be achieved and instruct the LLM20 to guide the user towards that goal. In the goal-seek prompt, the goal to be achieved is set, but the intermediate scenarios (paths) to reach that goal are not set.

[0028] When the goal seek prompt 70a is input, LLM20 responds to the input goal seek prompt with a set of questions (sentences) designed to guide the user towards the goal. The user answers these questions, and this process is repeated until the user's objective is achieved. Normally, reaching a goal requires considerable effort on the user's part, such as creating prompts each time. However, with goal seek prompts, the user does not need to create new prompts each time. By answering the questions output by LLM20 in the goal seek prompt, it is possible to efficiently move towards the goal.

[0029] S3: When a goal seek prompt 70a is input, LLM20 immediately returns a response including a question (so to speak, an AI question) 71a as a response to the input goal seek prompt. In this embodiment, the response to the input goal seek prompt 70a includes a question 71a to the user, such as "This is the XX Call Center. Please tell us the details of your inquiry." Furthermore, if the goal seek prompt instructs LLM20 to create a prompt, this response may also include a prompt generated by LLM20 (a prompt including the user response text 72a in S8 described later, or an example / template of such a prompt).

[0030] Since the question text 71a received from S4:LLM20 is text data, the voice call server 10 converts the text data of the question text 71a into voice call data (so to speak, AI question voice).

[0031] Furthermore, since LLM20 may sometimes provide unexpected or inappropriate answers, the voice call server 10 may modify (process) the text data of the question 71a before converting it into voice call data. For example, even if the prompt instructs LLM20 to answer within 100 characters, if it responds with 200 characters, the voice call server 10 may modify it to within 100 characters. In this case, the voice call server 10 may again have LLM20 modify or summarize the answer to within 100 characters before converting the text data of the question 71a into voice call data.

[0032] S5: The voice call server 10 responds to the telephone 30 with the converted call voice data (AI question voice) 71b. The call voice data 71b is an AI question voice, such as "This is the XX Call Center. Please tell us the nature of your inquiry," which asks the user for specific details about the inquiry and prompts them to make the call (speak).

[0033] S6: Upon receiving the AI ​​question voice 71b, the user freely communicates their inquiry as a user response voice 72b, for example, "I have a question about the product."

[0034] S7: The voice call server 10 receives the user's response voice 72b to the AI ​​question voice 71b from S5 as call audio data, and converts the received call audio data into text data (user's response text).

[0035] Furthermore, since users may sometimes give unexpected or inappropriate answers, the voice call server 10 may modify (process) the user's response audio 72b and convert it into text data. Alternatively, the voice call server 10 may prompt the user to speak again to make a correction, and then convert the user's response audio 72b into text data.

[0036] S8: The voice call server 10 generates a prompt (a prompt based on the user response text 72a) that includes the user response text 72a as the user's response to the AI ​​question voice 71b from S5, and inputs it to the LLM 20.

[0037] S9:LLM20 receives a prompt containing user response text 72a from S8 and immediately replies with a response containing question text 73a. For example, the response may include a question text 73a to the user, such as "This is the XX call center. Please tell us the nature of your inquiry."

[0038] S10: The voice call server 10 converts the text data of the answer question (AI question) 73a received from LLM20 into call audio data (AI question audio).

[0039] S11: The voice call server 10 responds to the telephone 30 with the converted call voice data (AI question voice) 73b. The call voice data 73b is an AI question voice such as, "Thank you for speaking with us. So, which specific product are you referring to?" and prompts the user to make a further response (speak).

[0040] Then, the exchange between the AI ​​question voice (AI question text) and the user answer voice (user answer text) is repeated until the user's objective is achieved (S6-S11). Figure 7 shows an example (part 2) of the text dialogue between the voice call server and LLM according to this embodiment. Figure 8 shows an example (part 2) of the voice call between the user and the voice call server according to this embodiment. Because the LLM20 according to this embodiment has learned, for example, the printer service manual, it is able to identify the user's printer malfunction and answer with a solution according to the service manual. Also, although user answer voices vary from person to person and sometimes include unexpected answers, the LLM20 can flexibly respond with a question voice that corresponds to the user answer voice each time.

[0041] Furthermore, the voice call server 10 can be linked with various systems and databases (Figure 1). For example, it can query the customer management database for the phone number of the calling user, identify the user information, and utilize that user information in the voice call. Also, if it is necessary to identify past call history, for example, it can be linked with a call history system, etc., and past call history can be utilized in the voice call.

[0042] [Embodiment 2] Embodiment 2 will now be described. Embodiment 1 showed an example of handling inquiries based on calls from users (inbound support) (S1 in Figure 4). Embodiment 2 shows an example of providing guidance and sales support based on calls from the voice call server 10 to the user (outbound support).

[0043] Figure 9 is a sequence diagram relating to the voice call system according to this embodiment. Figure 10 is a diagram showing an example (part 3) of text dialogue between the voice call server and LLM according to this embodiment. Figure 11 is a diagram showing an example (part 4) of text dialogue between the voice call server and LLM according to this embodiment. Figure 12 is a diagram showing an example (part 3) of voice call between a user and the voice call server according to this embodiment. Figure 13 is a diagram showing an example (part 4) of voice call between a user and the voice call server according to this embodiment. An example of providing sales information on insurance products to a user will be given, and the details will be explained in accordance with Figure 9 and, as appropriate, with reference to Figures 10-13.

[0044] S1-2: The voice call server 10, which corresponds to the caller, makes a call to the user, i.e., a predetermined telephone number (for example, a telephone number listed in the customer list). Compared to S1 in Figure 4, in S1... S2: When a user answers the phone, the voice call server 10 establishes a voice call and inputs a predetermined goal seek prompt 80a to the LLM20 to prompt the LLM20 to respond to the user's inquiries and requirements.

[0045] When S3:LLM20 receives a goal seek prompt 80a, it immediately responds to the entered goal seek prompt with a question (an AI question, so to speak) 81a. The response includes a question 71a directed at the user, such as, "This is the XX call center. Please tell us the nature of your inquiry."

[0046] S4: The voice call server 10 converts the text data of the answer question 81a received from LLM20 into call audio data (AI question voice).

[0047] S5: The voice call server 10 responds to the telephone 30 with the converted call voice data (AI question voice) 81b.

[0048] S6: The user receives the AI ​​question voice 81b and freely engages in conversation using the user's response voice 82b.

[0049] S7: The voice call server 10 receives the user's response voice 82b to the AI ​​question voice 81b from S5 as call audio data, and converts the received call audio data into text data (user response text).

[0050] S8: The voice call server 10 generates a prompt (a prompt based on the user response sentence 82a) that includes the user response sentence 82a as the user's response to the AI ​​question voice 81b from S5, and inputs it to the LLM 20.

[0051] S9:LLM20 receives a prompt containing user response 82a from S8 and immediately replies with question 83a.

[0052] S10: The voice call server 10 converts the text data of the answer question (AI question) 83a received from LLM20 into call audio data (AI question audio).

[0053] S11: The voice call server 10 responds to the telephone 30 with the converted call voice data (AI question voice) 83b.

[0054] Then, the exchange between the AI ​​question voice (AI question text) and the user answer voice (user answer text) is repeated until the user's objective is achieved (S6-S11). In this embodiment, the LLM20 has learned from, for example, the insurance product brochure and sales manual, and is therefore capable of outputting information related to the insurance product.

[0055] [Differentiation] Figure 14 is a sequence diagram relating to the voice call system according to this modified example. Figure 15 is a diagram showing an example (5) of text dialogue between the voice call server and LLM according to this modified example. As a goal that the user wants to achieve, specific user inquiries and requirements may be explicitly set as goals in the goal seek prompt.

[0056] S1: The user uses telephone 30 to make a call to a designated telephone number (for example, a call center's telephone number).

[0057] S1-3: The voice call server 10 receives a call from a user, establishes a voice call, and responds with guidance to solicit the user's inquiry or request, such as, "This is the XX Call Center. Please tell us the nature of your inquiry."

[0058] S1-4: After receiving guidance, the user can freely make a call to state their inquiry, for example, "I have a question about the product."

[0059] S1-5: The voice call server 10 performs voice analysis processing and extracts keywords to identify the requirements the user wants to inquire about (a means of identifying the user's desired purpose).

[0060] S2: The voice call server 10 inputs a goal seek prompt 90a to the LLM20, which sets the user's desired inquiry as the goal. Compared to the goal seek prompt 70a in Figure 5, the goal seek prompt 90a in Figure 15 shows a more specific goal, such as, "First, I have some questions about the product. Please create prompts using the goal seek prompt, ask me questions carefully, and guide me until I achieve my goal..." This improves the likelihood of the user achieving their desired goal (for example, resolving their inquiry).

[0061] <Summary> According to the voice call system 100 of this embodiment, since a human operator is not required, the number of users that can be handled simultaneously can be dramatically increased. Furthermore, because of the free voice call (response) provided by LLM (AI), there is no need for pre-determined response voices (voice guidance) as in the past, and the response voice is not limited to fixed and standardized requirement responses, so it is possible to provide more flexible voice guidance until the objective is achieved according to the user's response. In addition, the range of application fields and uses and versatility of the voice response system can be improved.

[0062] While the present invention has been described with specific examples based on preferred embodiments, it is clear that various modifications and changes can be made to these examples without departing from the broad spirit and scope of the invention as defined in the claims. In other words, the details of the examples and the accompanying drawings should not be construed as limiting the present invention.

[0063] The voice call system according to this embodiment can be used for inbound calls based on user initiation and outbound calls based on user initiation, and can be used in various situations such as telephone consultation and inquiry lines, procedures, product reservations, product purchase applications, product orders, complaint handling lines, guidance (sales), and telephone research / surveys. Multilingual support is also possible. [Explanation of Symbols]

[0064] 10 Voice call server 20 LLM 30 telephone 60 Networks 100 Voice Call System 101 Power receiving section 102 Prompt Input Section 103 Answer acquisition part 104 Text-to-Speech Conversion Unit 105 Voice transmission and reception unit

Claims

1. A voice communication device that assists the caller in achieving their objectives through voice communication with the user, A language model is provided with a first prompt input means for inputting prompts to guide it toward achieving the aforementioned objective, An answer acquisition means that obtains an answer including a question sentence for the user from the language model into which the prompt has been input, A voice conversion means that converts the first text of the aforementioned question into a first call audio, A call audio transmission means for transmitting the first call audio to the user's telephone, A call audio receiving means that receives a second call audio in response to the first call audio from the aforementioned telephone, A text conversion means that converts the second call audio into a second text, The language model includes a second prompt input means for inputting a prompt containing the second text, A voice communication device characterized by having the following features.

2. The prompt input by the first prompt input means is a goal seek prompt. The voice communication device according to claim 1, characterized by the following:

3. A means for identifying the objective desired by the user through voice analysis processing via a voice call with the user, The voice communication device according to claim 1 or 2, characterized by having the following features.

4. A computer that assists callers in achieving their objectives through voice calls with users, A language model is provided with a first prompt input means for inputting prompts to guide it toward achieving the aforementioned objective, An answer acquisition means that obtains an answer including a question sentence for the user from the language model into which the prompt has been input, A voice conversion means that converts the first text of the aforementioned question into a first call audio, A call audio transmission means for transmitting the first call audio to the user's telephone, A call audio receiving means that receives a second call audio in response to the first call audio from the aforementioned telephone, A text conversion means that converts the second call audio into a second text, The language model includes a second prompt input means for inputting a prompt containing the second text, A voice call program to enable this function.

Images