Intelligent question and answer voice processing method and device, equipment and storage medium

Abstract

The application relates to the field of artificial intelligence, and particularly discloses an intelligent question and answer voice processing method and device, equipment and a storage medium. The method comprises the following steps: acquiring audio data of voice interaction between an intelligent robot and each customer; performing fragmentation processing on each audio data based on a VAD detection algorithm to obtain a voice sequence corresponding to each audio data; converting each voice sequence into text information, extracting a question and a corresponding first answer from the same text information; acquiring customer information, binding the question, the first answer and the customer information of the same customer to form a plurality of customer data, and constructing a database based on the customer data; acquiring a second answer preset by a user for each question, calling the database, matching the second answer and the first answer of the same question, determining a target customer according to a matching result, extracting customer information of the target customer and sending the customer information to the user. Through the above method, the efficiency and accuracy of customer intention judgment can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of natural language processing of artificial intelligence, and in particular to an intelligent question and answer voice processing method and device, equipment and a storage medium. BACKGROUND

[0002] Telephone sales personnel call customers through telephone outbound, and then ask customers questions to obtain customer questionnaire answers. With the popularization of intelligent robot outbound technology, after the intelligent robot and the customer complete the question and answer communication, the recording data is saved, how to quickly infer the customer's preferences from the recording data, and make corresponding analysis on the purchase desire and purchase ability becomes particularly important, and then the customer can be further recommended products. Now many cases are that customer service personnel listen to the recording, and then analyze whether the user has the intention to purchase our products while listening, which exists problems of time-consuming, low efficiency and easy misjudgment of manual operation. SUMMARY

[0003] The present application provides an intelligent question and answer voice processing method, device, equipment and storage medium, which can improve the efficiency and accuracy of customer intention judgment, and solves the problem that the user judges the customer intention by listening to the question and answer recording, resulting in high time cost, low efficiency and inaccurate intention judgment.

[0004] To solve the above technical problems, one technical solution adopted by the present application is to provide an intelligent question and answer voice processing method, comprising:

[0005] Obtain audio data of question and answer voice interaction between an intelligent robot and each customer;

[0006] Perform slicing processing on each audio data based on a VAD detection algorithm to obtain a voice sequence corresponding to each audio data;

[0007] Convert each voice sequence into text information, extract a question and a first answer corresponding to the question from the same text information;

[0008] Obtain customer information, bind the question, the first answer and the customer information of the same customer to form a plurality of customer data, and construct a database based on the customer data;

[0009] Obtain a second answer preset by a user for each question, call the database and match the second answer and the first answer of the same question, determine a target customer according to the matching result, extract customer information of the target customer and send it to the user.

[0010] According to one embodiment of the present application, the VAD-based detection algorithm performs slicing on each of the audio data to obtain a speech sequence corresponding to each of the audio data, which includes:

[0011] detecting a start point and an end point of a speech signal from the audio data based on the VAD-based detection algorithm;

[0012] extracting the speech signal between the start point and the end point from the audio data in a detection order;

[0013] sequentially arranging each of the speech signals in an extraction order to generate the speech sequence.

[0014] According to one embodiment of the present application, the VAD-based detection algorithm detects a start point and an end point of a speech signal from the audio data, which includes:

[0015] detecting the start point of the speech signal from the audio data starting from a starting frame of the audio data based on a VAD detection algorithm;

[0016] detecting the end point of the speech signal starting from a next frame of the speech signal of the start point based on the VAD detection algorithm.

[0017] According to one embodiment of the present application, the VAD-based detection algorithm detects a start point and an end point of a speech signal from the audio data, which includes:

[0018] calculating a short-time energy and a zero-crossing rate of each frame of the speech signal starting from a starting frame of the audio data;

[0019] comparing the short-time energy of each frame of the speech signal with a preset first threshold and comparing the zero-crossing rate of each frame of the speech signal with a preset second threshold;

[0020] taking a first frame of the speech signal with a short-time energy greater than the first threshold as a current frame, and determining the current frame as the start point of the speech signal if the short-time energy of the current frame and a plurality of continuous frames after the current frame of the speech signal are all greater than the first threshold and the zero-crossing rate of the current frame and the plurality of continuous frames after the current frame of the speech signal are all greater than the second threshold.

[0021] According to one embodiment of the present application, the VAD-based detection algorithm detects a start point and an end point of a speech signal from the audio data, which includes:

[0022] comparing the short-time energy of each frame of the speech signal with the first threshold starting from the next frame of the speech signal of the start point;

[0023] The first frame with short-time energy of the voice signal less than the first threshold is regarded as a current frame, if the short-time energy of the voice signal of the current frame and continuous multiple frames after the current frame are all less than the first threshold, the current frame is determined as the end point of the voice signal.

[0024] According to one embodiment of the present application, the converting each voice sequence into text information, extracting a question and a first answer corresponding to the question from the same text information comprises:

[0025] Calling a third-party interface to convert each voice sequence into text information;

[0026] Extracting a keyword from the same text information, determining the position of the question in the text information according to the keyword, and determining the position of the first answer according to the position of the question;

[0027] Extracting the question according to the position of the question and extracting the first answer according to the position of the first answer.

[0028] According to one embodiment of the present application, the obtaining a second answer preset by the user for each question, calling the database and matching the second answer of the same question with the first answer, determining a target customer according to the matching result, extracting customer information of the target customer and sending to the user comprises:

[0029] Key word matching the second answer of the same question with the first answer;

[0030] Scoring according to the matching result of each question of the same customer and calculating the total score of the corresponding customer according to each score result;

[0031] Comparing the total score of each customer with a preset value, if the total score exceeds the preset value, the corresponding customer is determined as a target customer, customer information of the target customer is extracted and sent to the user.

[0032] To solve the above technical problems, another technical solution adopted by the present application is to provide an intelligent question and answer voice processing device, comprising:

[0033] An acquisition module is configured to acquire audio data of voice interaction between an intelligent robot and each customer;

[0034] A fragmentation processing module is configured to perform fragmentation processing on each audio data based on a VAD detection algorithm to obtain a voice sequence corresponding to each audio data;

[0035] The extraction module is configured to convert each of the speech sequences into text information, and extract a question and a first answer corresponding to the question from the same text information.

[0036] The construction module is configured to obtain customer information, bind the question, the first answer and the customer information of the same customer to form a plurality of customer data, and construct a database based on the customer data.

[0037] The determination module is configured to obtain a second answer preset by a user for each of the questions, call the database and match the second answer and the first answer of the same question, determine a target customer according to a matching result, extract customer information of the target customer and send the customer information to the user.

[0038] To solve the above technical problems, the present application employs another technical scheme, which provides a computer device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the intelligent question and answer speech processing method when executing the computer program.

[0039] To solve the above technical problems, the present application employs another technical scheme, which provides a computer storage medium having a computer program stored thereon, and the computer program is executable on a processor to implement the intelligent question and answer speech processing method.

[0040] The present application has the beneficial effects that the audio data of the interaction between the intelligent robot and each customer is processed by the VAD detection algorithm to obtain speech sequences, the speech sequences are converted into text information, a database is constructed based on the text information, the customer intention is determined by matching the user preset value with the data in the database, the target customer and the corresponding customer information are obtained, the efficiency and accuracy of the customer intention judgment are improved by the above-mentioned method, the user experience is improved, the manual operation is reduced, and the problems of high time cost, low efficiency and inaccurate intention judgment caused by the user listening to the question and answer recording to determine the customer intention are solved. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 is a flowchart of the intelligent question and answer speech processing method of the first embodiment of the present application;

[0042] Figure 2 is a flowchart of the intelligent question and answer speech processing method of the second embodiment of the present application;

[0043] Figure 3 is a flowchart of the intelligent question and answer speech processing method of the third embodiment of the present application;

[0044] Figure 4is a flowchart of an intelligent question and answer voice processing method of a fourth embodiment of the present application;

[0045] Figure 5 is a flowchart of an intelligent question and answer voice processing method of a fifth embodiment of the present application;

[0046] Figure 6 is a flowchart of an intelligent question and answer voice processing method of a sixth embodiment of the present application;

[0047] Figure 7 is a flowchart of an intelligent question and answer voice processing method of a seventh embodiment of the present application;

[0048] Figure 8 is a structural diagram of an intelligent question and answer voice processing device of an embodiment of the present application;

[0049] Figure 9 is a structural diagram of a computer device of an embodiment of the present application;

[0050] Figure 10 is a structural diagram of a computer storage medium of an embodiment of the present application. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0052] The terms "first", "second", "third" in the present application are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features limited by "first", "second", "third" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present application are only used to explain the relative positional relationship, movement condition, etc. between components in a certain posture (as shown in the drawings), and if the certain posture changes, the directional indications also change accordingly. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units is not limited to the listed steps or units, but can optionally include steps or units not listed, or can optionally include other steps or units inherent to the process, method, product or device.

[0053] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0054] Figure 1 This is a flowchart illustrating the intelligent question-and-answer speech processing method according to an embodiment of the present invention. It should be noted that if substantially the same result is obtained, the method of the present invention is not necessarily identical. Figure 1 The illustrated process sequence is limited. For example... Figure 1 As shown, the method includes the following steps:

[0055] Step S101: Obtain audio data of the question-and-answer voice interactions between the intelligent robot and each customer.

[0056] In step S101, the intelligent robot obtains multiple customer mobile phone numbers and interacts with the corresponding customers through these numbers. Further, after the intelligent robot successfully connects with the customer via outbound calls, it asks the customer questions according to stored preset questions. The customer answers each question, and the intelligent robot collects audio data from the question-and-answer voice interactions with each customer.

[0057] In one feasible implementation, the intelligent robot asks a question, the customer answers the question, thus generating an answer to that question. The intelligent robot then continues to ask the next question, the customer answers the next question, and so on, generating an answer to that question. For example, if the preset questions include five questions, and the customer answers each question, then five answers are generated. In this case, the audio data includes the first question, the answer to the first question, the second question, the answer to the second question, the third question, the answer to the third question, the fourth question, the answer to the fourth question, the fifth question, and the answer to the fifth question. As another example, if the preset questions include five questions, and the customer answers the first, second, fourth, and fifth questions, then four answers are generated. In this case, the audio data includes the first question, the answer to the first question, the second question, the answer to the second question, the third question, the fourth question, the answer to the fourth question, the fifth question, and the answer to the fifth question. In the above embodiments, interruption signals may occur during the interaction between the intelligent robot and the customer; therefore, the audio data also includes interruption signals between questions and answers.

[0058] Step S102: Based on the VAD detection algorithm, the audio data is segmented to obtain the speech sequence corresponding to each audio data.

[0059] In step S102, voice activity detection (VAD), also known as voice endpoint detection, can detect whether there is a voice signal in the audio data, and specifically detect the start point of the valid voice and the end point of the valid voice. In an implementable embodiment, the VAD algorithm detects the voice signal by short-time energy and zero-crossing rate, wherein the short-time energy is the energy of a frame of voice signal, and the zero-crossing rate is the number of times that a frame of voice time-domain signal crosses 0 (time axis). Specifically, when detecting the start point of the voice signal, the VAD algorithm includes two levels of judgment, the first level of judgment is to compare the short-time energy with a first threshold, if the short-time energy is greater than the first threshold, then compare the zero-crossing rate with a second threshold, if the zero-crossing rate is greater than the second threshold, then determine the start point of the voice signal. When detecting the end point of the voice signal, compare the short-time energy with the first threshold, if the short-time energy is less than the first threshold, then determine the end point of the voice signal.

[0060] The embodiment according to the start point and the end point of each voice signal performs slicing processing, can obtain multiple voice signals, and eliminates noise signals other than the intelligent machine and the customer. In the embodiment, the sliced voice signals are arranged in the detection order of the voice signals, to form a voice sequence.

[0061] Further, referring to Figure 2 , step S102 further includes the following steps:

[0062] Step S201: detecting the start point and the end point of the voice signal from the audio data based on the VAD detection algorithm.

[0063] In an implementable embodiment, referring to Figure 3 , step S201 further includes the following steps:

[0064] Step S301: starting from the starting frame of the audio data, detecting the start point of the voice signal from the audio data by using the VAD detection algorithm.

[0065] In step S301, the short-time energy and the zero-crossing rate are used to detect the start point of the voice signal.

[0066] In an implementable embodiment, referring to Figure 4 , step S301 further includes the following steps:

[0067] Step S401: starting from the starting frame of the audio data, calculating the short-time energy and the zero-crossing rate of each frame of voice signal.

[0068] The calculation method of the short-time energy of the embodiment is the square sum of the in-frame signal, and the calculation method of the zero-crossing rate is to shift all the samples in the frame by 1, and then multiply the corresponding points. If the sign is negative, it means that the zero-crossing occurs at this point. Only the number of negative products in all the samples in the frame is obtained to obtain the zero-crossing rate of the frame.

[0069] Step S402: compare the short-time energy of each frame of the speech signal with a preset first threshold, and compare the zero-crossing rate of each frame of the speech signal with a preset second threshold.

[0070] The first threshold of the embodiment is the maximum value of the short-time energy, and the second threshold is the maximum value of the zero-crossing rate.

[0071] Step S403: take the first frame of the speech signal whose short-time energy is greater than the first threshold as the current frame. If the short-time energy of the current frame and the continuous multiple frames of the speech signal after the current frame are all greater than the first threshold, and the zero-crossing rate of the current frame and the continuous multiple frames of the speech signal after the current frame are all greater than the second threshold, it is determined that the current frame is the starting point of the speech signal.

[0072] In the embodiment, if the short-time energy of the current frame is greater than the first threshold and the zero-crossing rate of the current frame is greater than the second threshold, it is judged that the current frame is possibly the starting point of the speech signal. It is necessary to judge whether the short-time energy of the continuous frames of the speech signal after the current frame is greater than the first threshold, and whether the zero-crossing rate is greater than the second threshold. When the short-time energy of the continuous frames of the speech signal is greater than the first threshold and the zero-crossing rate is greater than the second threshold, it is determined that the current frame is the starting point of the speech signal.

[0073] Step S302: starting from the next frame of the speech signal of the starting point, a VAD detection algorithm is used to detect the end point of the speech signal.

[0074] In step S302, the short-time energy is used to detect the end point of the speech signal.

[0075] In an implementable embodiment, please refer to Figure 5 , step S302 further includes the following steps:

[0076] Step S501: starting from the next frame of the speech signal of the starting point, compare the short-time energy of each frame of the speech signal with the first threshold.

[0077] Step S502: take the first frame of the speech signal whose short-time energy is less than the first threshold as the current frame. If the short-time energy of the current frame and the continuous multiple frames of the speech signal after the current frame are all less than the first threshold, it is determined that the current frame is the end point of the speech signal.

[0078] The embodiment compares the short-time energy of each frame of the speech signal with the first threshold after the starting point. If the short-time energy of a frame is less than the first threshold, it is determined whether the short-time energy of the continuous frames after the frame is less than the first threshold. If yes, it is determined that the frame is the ending point of the speech signal.

[0079] Step S202: extracting the speech signal between the starting point and the ending point from the audio data according to the detection sequence.

[0080] In step S202, the starting point and the ending point of all the speech signals in the audio data can be detected according to the above method. Starting from the first starting point, the speech between the adjacent starting point and ending point is the speech signal to be extracted. In an embodiment, if three starting points and three ending points are detected from the audio data, the detection sequence is the first starting point, the first ending point, the second starting point, the second ending point, the third starting point, and the third ending point. The extracted speech signals include the first speech signal between the first starting point and the first ending point, the second speech signal between the second starting point and the second ending point, and the third speech signal between the third starting point and the third ending point.

[0081] Step S203: arranging the speech signals in sequence according to the extraction sequence to generate a speech sequence.

[0082] In step S203, the speech sequence includes the first speech signal, the second speech signal, and the third speech signal according to the extraction sequence. In an implementable embodiment, if the customer answers the first question, the second speech signal can be the answer to the first question. In another implementable embodiment, if the customer does not answer the first question, the second speech signal can be the second question initiated by the intelligent robot. In still another implementable embodiment, if the customer answers the first question but there is an interrupting signal in the answering process, the second speech signal and the third speech signal can both be the answer to the first question. In the embodiment, each question initiated by the intelligent robot includes a question number and a question content, that is, the speech signal corresponding to the intelligent robot includes the question number and the question content.

[0083] Step S103: converting the speech sequence into text information, and extracting the question and the first answer corresponding to the question from the same text information.

[0084] In step S103, the voice sequence is converted into corresponding text information. For example, in a scenario where the smart robot answers each question initiated by the client, if the voice sequence includes a first voice signal, a second voice signal, a third voice signal and a fourth voice signal, the corresponding text information includes a first question (including a question number one and a question content), an answer to the first question, a second question (including a question number two and a question content), and an answer to the second question.

[0085] In an implementable embodiment, referring to Figure 6 , step S103 further includes the following steps:

[0086] Step S601: calling a third-party interface to convert each voice sequence into text information.

[0087] In a first implementable embodiment, a scenario where the smart robot answers each question initiated by the client is taken as an example for illustration. The text information can be a first question (including a question number one and a question content), an answer to the first question, a second question (including a question number two and a question content), an answer to the second question, a third question (including a question number three and a question content), and an answer to the third question. In a second implementable embodiment, a scenario where the smart robot answers part of the questions initiated by the client is taken as an example for illustration. The text information can be a first question (including a question number one and a question content), an answer to the first question, a second question (including a question number two and a question content), a third question (including a question number three and a question content), and an answer to the third question. In a third implementable embodiment, a scenario where the smart robot answers each question initiated by the client but the answering process can be interrupted is taken as an example for illustration. The text information can be a first question (including a question number one and a question content), a first partial answer to the first question, a second partial answer to the first question, a second question (including a question number two and a question content), an answer to the second question, a third question (including a question number three and a question content), and an answer to the third question.

[0088] Step S602: extracting keywords from the same text information, determining a position of a question in the text information according to the keywords, and determining a position of a first answer according to the position of the question.

[0089] In this embodiment, the keywords are question numbers, the positions of the question numbers are the positions of the questions, and the positions between the question numbers are the positions of the answers.

[0090] Step S603: extracting a question according to the position of the question and extracting a first answer according to the position of the first answer.

[0091] Step S104: Obtain customer information, bind the same customer's questions and first answers with the customer information, form multiple customer data, and build a database based on the customer data.

[0092] In step S104, the customer information includes the customer's mobile phone number, the customer's name, and the customer's number. In an implementable embodiment, all questions of the same customer are bound with the corresponding answers one by one and stored in a preset table, and all questions and answers of the customer are bound with the customer information. For each customer, a piece of customer data is formed, and a database is built based on multiple customer data. For example, one piece of customer data includes the customer number, the customer's name, and the customer's mobile phone number; question one: have you ever bought personal insurance? Answer one: yes; question two: which insurance company's product did you buy? Answer two: xx insurance company; question three: if there is a 0 yuan purchase insurance activity, which of the following insurance categories do you prefer: car insurance, personal insurance, life insurance, accident insurance, etc.; answer three: personal insurance, and so on.

[0093] Step S105: Obtain the second answers preset by the user for each question, call the database and match the second answers of the same question with the first answers, determine the target customer according to the matching result, extract the customer information of the target customer and send it to the user.

[0094] In step S105, the more similar the second answers of the same question are to the first answers, the higher the matching degree. Customers with a matching degree exceeding a preset value are determined as target customers, the customer information of the target customers is extracted and sent to the user, the user calls the corresponding target customers for telemarketing through the customer's mobile phone number, improves the customer intention judgment efficiency, and thus improves the user experience, reduces manual operation, and improves the accuracy of intention judgment and communication efficiency.

[0095] In an implementable embodiment, please refer to Figure 7 Step S105 further includes the following steps:

[0096] Step S701: Perform keyword matching on the second answers and the first answers of the same question;

[0097] Step S702: Score according to the matching results of each question of the same customer, and calculate the total score of the corresponding customer according to each score result;

[0098] Step S703: Compare the total score of each customer with a preset value, if the total score exceeds the preset value, the corresponding customer is determined as a target customer, the customer information of the target customer is extracted and sent to the user.

[0099] The intelligent question and answer voice processing method of the embodiment of the present application can improve the efficiency and accuracy of customer intention judgment, and further improve the user experience and reduce manual operation by the following manner: obtaining the target customer and corresponding customer information by obtaining the voice sequence through the VAD detection algorithm for the audio data of the interaction between the intelligent robot and each customer, converting the voice sequence into text information, constructing a database based on the text information, and determining the customer intention by matching the user preset value with the data in the database.

[0100] Figure 8 is a structural schematic diagram of the intelligent question and answer voice processing device of the embodiment of the present application. As shown in the figure, Figure 8 the device 80 comprises an acquisition module 81, a fragmentation processing module 82, an extraction module 83, a construction module 84 and a determination module 85.

[0101] The acquisition module 81 is configured to acquire audio data of the question and answer voice interaction between the intelligent robot and each customer.

[0102] The fragmentation processing module 82 is configured to perform fragmentation processing on each audio data based on a VAD detection algorithm to obtain a voice sequence corresponding to each audio data.

[0103] The extraction module 83 is configured to convert each voice sequence into text information, and extract a question and a first answer corresponding to the question from the same text information.

[0104] The construction module 84 is configured to acquire customer information, bind the question, the first answer and the customer information of the same customer to form a plurality of customer data, and construct a database based on the customer data.

[0105] The determination module 85 is configured to acquire a second answer preset by the user for each question, call the database and match the second answer with the first answer of the same question, determine a target customer according to the matching result, extract the customer information of the target customer and send it to the user.

[0106] Please refer to Figure 9 , Figure 9 is a structural schematic diagram of the computer device of the embodiment of the present application. As shown in the figure, Figure 9 the computer device 90 comprises a processor 91 and a memory 92 coupled with the processor 91.

[0107] The memory 92 stores program instructions for implementing the intelligent question and answer voice processing method of any of the above embodiments.

[0108] The processor 91 is configured to execute the program instructions stored in the memory 92 to process the intelligent question and answer voice.

[0109] The processor 91 can also be called a CPU (Central Processing Unit). The processor 91 can be an integrated circuit chip having a processing capability of signals. The processor 91 can also be a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application-Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0110] Referring to Figure 10 , Figure 10 is a structural schematic diagram of the computer storage medium of the embodiment of the present application. The computer storage medium of the embodiment of the present application stores a program file 101 capable of realizing all the methods described above, wherein the program file 101 can be stored in the computer storage medium in the form of a software product, including a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the method described in various embodiments of the present application. The aforementioned computer storage medium includes: a U disk, a mobile hard disk, a ROM (Read-Only Memory), a RAM (Random Access Memory), a magnetic disk or an optical disk, and various media capable of storing program codes, or a terminal device such as a computer, a server, a mobile phone, a tablet, etc.

[0111] In several embodiments provided by the present application, it should be understood that the disclosed system, device and method can be implemented by other ways. For example, the device embodiments described above are only schematic, and for example, the division of the units is only a logical function division, and there can be another division manner in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0112] In addition, each functional unit in the various embodiments of the present application can be integrated into one processing unit, or each unit can exist physically, or two or more units can be integrated into one unit. The integrated unit can be realized in the form of hardware, or in the form of a software functional unit.

[0113] The above merely illustrates the embodiments of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation, or direct or indirect application in other related technical fields, which are made by using the content of the present application specification and drawings, are also included in the patent protection scope of the present application.

Claims

1. An intelligent question-answering speech processing method, characterized by, The method comprises: obtaining audio data of voice interaction between an intelligent robot and each customer; performing slicing processing on each audio data based on a VAD detection algorithm to obtain a voice sequence corresponding to each audio data; converting each voice sequence into text information, extracting a question and a first answer corresponding to the question from the same text information; obtaining customer information, binding the question, the first answer and the customer information of the same customer to form a plurality of customer data, and constructing a database based on the customer data; obtaining a second answer preset by a user for each question, calling the database, performing keyword matching between the second answer and the first answer of the same question, determining a target customer according to a matching result, extracting customer information of the target customer and sending the customer information to the user; the slicing processing on each audio data based on the VAD detection algorithm to obtain a voice sequence corresponding to each audio data comprises: detecting a starting point and an ending point of a voice signal from the audio data based on the VAD detection algorithm; extracting the voice signal between the starting point and the ending point from the audio data in a detection order; sequentially arranging each voice signal in the extraction order to generate the voice sequence; the detecting a starting point and an ending point of a voice signal from the audio data based on the VAD detection algorithm comprises: starting from a starting frame of the audio data, detecting a starting point of a voice signal from the audio data by using a VAD detection algorithm; starting from a next frame of the voice signal of the starting point, detecting an ending point of the voice signal by using the VAD detection algorithm; the starting from a starting frame of the audio data, detecting a starting point of a voice signal from the audio data by using a VAD detection algorithm comprises: starting from a starting frame of the audio data, calculating a short-time energy and a zero-crossing rate of each frame of the voice signal; comparing the short-time energy of each frame of the voice signal with a preset first threshold, and comparing the zero-crossing rate of each frame of the voice signal with a preset second threshold; taking a first frame with a short-time energy greater than the first threshold as a current frame, and if the short-time energy of the current frame and a plurality of continuous frames after the current frame are all greater than the first threshold, and the zero-crossing rate of the current frame and the plurality of continuous frames after the current frame are all greater than the second threshold, determining the current frame as the starting point of the voice signal; the starting from a next frame of the voice signal of the starting point, detecting an ending point of the voice signal by using the VAD detection algorithm comprises: starting from a next frame of the voice signal of the starting point, comparing the short-time energy of each frame of the voice signal with the first threshold; taking a first frame with a short-time energy less than the first threshold as a current frame, and if the short-time energy of the current frame and a plurality of continuous frames after the current frame are all less than the first threshold, determining the current frame as the ending point of the voice signal. 2.The intelligent question-answering voice processing method of claim 1, wherein, The converting each of the voice sequences into text information, extracting a question and a first answer corresponding to the question from the same text information comprises: calling a third-party interface to convert each of the voice sequences into text information; extracting a keyword from the same text information, determining a position of the question in the text information according to the keyword, and determining a position of the first answer according to the position of the question; extracting the question according to the position of the question and extracting the first answer according to the position of the first answer. 3.The intelligent question-answering voice processing method of claim 1, wherein, The obtaining a second answer preset by a user for each of the questions, calling the database, and performing keyword matching between the second answer and the first answer of the same question, scoring according to a matching result to determine a target customer, and extracting customer information of the target customer and sending the customer information to the user comprises: performing keyword matching between the second answer and the first answer of the same question; scoring according to a matching result of each of the questions of the same customer and calculating a total score of the corresponding customer according to each of the scoring results; comparing the total score of each customer with a preset value, and if the total score exceeds the preset value, determining the corresponding customer as the target customer, extracting customer information of the target customer, and sending the customer information to the user.

4. An intelligent question answering voice processing apparatus, characterized by, The intelligent question and answer voice processing method is applied to any one of claims 1-3, comprising: an acquisition module configured to acquire audio data of voice interaction between an intelligent robot and each customer; a fragmentation processing module configured to perform fragmentation processing on each of the audio data based on a VAD detection algorithm to obtain a voice sequence corresponding to each of the audio data; an extraction module configured to convert each of the voice sequences into text information, and extract a question and a first answer corresponding to the question from the same text information; a construction module configured to acquire customer information, bind the question, the first answer, and the customer information of the same customer, form a plurality of customer data, and construct a database based on the customer data; a determination module configured to obtain a second answer preset by a user for each of the questions, call the database, perform keyword matching between the second answer and the first answer of the same question, score according to a matching result to determine a target customer, and extract customer information of the target customer and send the customer information to the user.

5. A computer device comprising: A memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the intelligent question and answer voice processing method of any one of claims 1-3 when executing the computer program.

6. A computer storage medium having stored thereon a computer program, characterized in that The computer program is executed by the processor to implement the intelligent question and answer voice processing method of any one of claims 1-3.

Images