Intelligent call forwarding method, device, equipment and readable storage medium

Abstract

The present application provides an intelligent call forwarding method, apparatus, device and readable storage medium, relating to the technical field of call forwarding, including: when the call is connected, sending a first response voice to the callee; receiving a first response voice from the callee for the first response voice; analyzing the first response voice to obtain a voice analysis result; if the voice analysis result meets the configured forwarding rules, converting the first response voice and the first response voice into a first response text and a first response text respectively; forwarding the call with the callee to a manual customer service representative, and sending identity information, historical call data, data on matters to be mediated, the first response text and the first response text to the manual customer service representative. The present application improves the intelligence and automation level of intelligent call forwarding, while also improving mediation efficiency.

Description

Technical Field

[0001] The present application relates to the technical field of call forwarding, and in particular to an intelligent call forwarding method, apparatus, device and readable storage medium. Background Art

[0002] Existing AI customer service call technology can only respond with configured fixed phrases, and cannot generate targeted responses based on the user's actual or historical circumstances, making it insufficiently flexible. Furthermore, its voice differs significantly from a human voice, leading to a high probability that customers will hang up the phone after hearing the voice of the intelligent customer service.

[0003] Although there are platforms in the mediation industry where robots can complete calls, the calls are usually answered by human customer service staff as soon as they are connected. The human customer service staff will then convey general matters or fixed lines of dialogue to the customer before conducting mediation communication. However, some customers will hang up the phone while conveying general matters or fixed lines of dialogue. This not only wastes the time of the human customer service staff, but also causes them to keep conveying repetitive matters, resulting in low mediation efficiency. In addition, the call forwarding is not very intelligent and cannot automatically reply to the customer's voice, which leads to a poor user experience. Summary of the Invention

[0004] The purpose of the embodiments of the present application is to provide an intelligent call forwarding method, device, equipment and readable storage medium to solve the above-mentioned problems existing in the prior art and improve mediation efficiency.

[0005] In a first aspect, a smart call forwarding method is provided, which may include:

[0006] When the call is connected, a first response voice is sent to the called party; wherein the first response voice is generated based on the identity information of the called party, the data of the matters to be mediated, and the historical call data;

[0007] receiving a first reply voice from the calling party in response to the first answer voice;

[0008] Analyzing the first reply speech to obtain a speech analysis result;

[0009] If the voice analysis result satisfies the configured transfer rule, converting the first answer voice and the first reply voice into a first answer text and a first reply text respectively;

[0010] The call with the call object is transferred to the manual customer service, and the identity information, the historical call data, the data of the matters to be mediated, the first response text and the first reply text are sent to the manual customer service.

[0011] In an optional implementation, the method for generating the first response voice includes:

[0012] Obtaining the identity information and data of the matters to be mediated of the callee, as well as the historical call data of the callee;

[0013] Generate a first response text based on the identity information, the data of the matter to be mediated, the historical call data, and the configured response template;

[0014] The first response text is input into a pre-trained AI speech generation model to obtain a first response speech.

[0015] In an optional implementation, the data on matters to be mediated include: the bank to which the debt is owed and the status of the debt;

[0016] The historical call data includes: the number of historical calls, historical call time and call summary.

[0017] In an optional implementation, inputting the first response text into a pre-trained AI speech generation model to obtain a first response speech includes:

[0018] Inputting the first response text into a pre-trained AI speech generation model to obtain an initial response speech; wherein the AI speech generation model is trained using historical mediation data; the historical mediation data is historical call data between different human customer service representatives and different call recipients;

[0019] The initial response voice is subjected to voiceprint processing to obtain a first response voice.

[0020] In an optional implementation, analyzing the first reply speech to obtain a speech analysis result includes:

[0021] Converting the first reply voice into text to obtain a first response text;

[0022] Preprocessing the first response text to obtain a preprocessed first response text;

[0023] Performing semantic analysis on the preprocessed first response text to obtain a semantic analysis result;

[0024] Inputting the preprocessed first response text and the semantic analysis result into a pre-trained text classification model to obtain a semantic classification result of the first response text;

[0025] The semantic analysis result and the semantic classification result are determined as the speech analysis result of the first reply speech.

[0026] In an optional implementation, after obtaining the speech analysis result, the method further includes:

[0027] Matching the second response text corresponding to the semantic analysis result and the semantic classification result from the configured comparison table of different semantic analysis results, different semantic classification results and corresponding second response texts;

[0028] The second response text is input into the AI voice generation model to obtain a second response voice, and the second response voice is sent to the calling party.

[0029] In an optional implementation, the method further includes:

[0030] If the semantic analysis result, the semantic classification result or the second response text is empty, the call with the calling party will be transferred to the manual customer service, and the identity information, the historical call data, the data on matters to be mediated, the response text and the reply text during this call will be sent to the manual customer service.

[0031] In a second aspect, an intelligent call forwarding device is provided, which may include:

[0032] A sending unit, configured to send a first response voice to the called party after the call is connected; wherein the first response voice is generated based on the identity information of the called party, data on matters to be mediated, and historical call data;

[0033] A receiving unit, configured to receive a first reply voice from the calling party in response to the first answer voice;

[0034] an analyzing unit, configured to analyze the first reply speech to obtain a speech analysis result;

[0035] a conversion unit, configured to convert the first answer voice and the first reply voice into a first answer text and a first reply text, respectively, if the voice analysis result satisfies a configured transfer rule;

[0036] The transfer unit is used to transfer the call with the call object to the manual customer service, and send the identity information, the historical call data, the data of the matters to be mediated, the first response text and the first reply text to the manual customer service.

[0037] In a third aspect, an electronic device is provided, the electronic device including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus;

[0038] Memory for storing computer programs;

[0039] The processor is configured to implement any of the method steps described in the first aspect when executing a program stored in the memory.

[0040] In a fourth aspect, a computer-readable storage medium is provided, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, any of the method steps described in the first aspect is implemented.

[0041] This application not only completes batch outbound calls, but also does not transfer the call to manual customer service immediately after the call is connected. Instead, a pre-trained model conveys general matters or fixed lines of speech to the callee; and intelligently makes a second reply based on the customer's response, realizing a dialogue between AI and the callee; and during the conversation, the pre-trained model is used to process the reply text to simulate the human voice, thereby improving the user experience.

[0042] This application not only completes batch outbound calls, but also can replace manual repetitive work on general matters, thereby liberating mediators and allowing them to focus their energy on the specific mediation process that requires the most manpower. It not only reduces useless operations, but also improves mediation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0044] Figure 1 A flowchart of an intelligent call forwarding method provided in an embodiment of the present application;

[0045] Figure 2 A schematic diagram of an intelligent call forwarding method provided in an embodiment of the present application;

[0046] Figure 3 A schematic diagram of the structure of an intelligent call forwarding device provided in an embodiment of the present application;

[0047] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0049] The intelligent call forwarding method provided in the embodiments of the present application can be applied in a server or a terminal with strong computing capabilities. The server can be a physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. The terminal can be a user equipment (UE) such as a mobile phone, smartphone, laptop, digital broadcast receiver, personal digital assistant (PDA), tablet computer (PAD), handheld device, vehicle-mounted device, wearable device, computing device, other processing device connected to a wireless modem, mobile station (MS), mobile terminal, etc. The terminal and server can be connected directly or indirectly via wired or wireless communication methods, which is not limited in this application.

[0050] The preferred embodiments of the present application are described below in conjunction with the drawings in the specification. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present application and are not used to limit the present application. In addition, the embodiments and features in the embodiments of the present application can be combined with each other if there is no conflict.

[0051] Figure 1 A flow chart of an intelligent call forwarding method provided in an embodiment of the present application. Figure 1 As shown, the method may include:

[0052] Step S110: When the call is connected, a first response voice is sent to the called party; a first reply voice from the called party to the first response voice is received; and the first reply voice is analyzed to obtain a voice analysis result.

[0053] In an embodiment of the present application, a method for generating a first response voice includes:

[0054] Obtain the identity information of the callee, data on matters to be mediated, and historical call data of the callee; generate a first response text based on the identity information, data on matters to be mediated, historical call data, and the configured response template; input the first response text into a pre-trained AI voice generation model to obtain a first response voice.

[0055] In an embodiment of the present application, the data on matters to be mediated include: the type of matters to be mediated; different types of matters to be mediated and different historical call data correspond to different response templates, and the response templates corresponding to the matters to be mediated and the historical call status corresponding to each call object are matched from the configured different types of matters to be mediated, different historical call statuses and corresponding response templates; for example, if the type of matter to be mediated is overdue mortgage, and the call object in the historical call data of the call object did not answer the phone, then the corresponding response template is for overdue mortgage and previous failure to communicate.

[0056] In an embodiment of the present application, the historical call status is obtained by analyzing the historical call data; it includes: not answered, answered but not replied, and answered and replied; among them, answered and replied can be further divided into answered and replied repayment date, answered but not replied repayment date, and answered and replied the current status of the person.

[0057] In the embodiment of the present application, the AI speech generation model includes:

[0058] A multimodal text preprocessing module is used to perform semantic analysis on the input first response text to obtain a text feature vector;

[0059] The adaptive hybrid coding acoustic processing module is used to map the obtained text feature vector into the acoustic space. It uses a pre-trained Transformer-CNN-based conversion model to perform acoustic conversion on the text feature vector to obtain the initial acoustic features. It also uses a sparse attention mechanism to generate an acoustic feature vector based on the initial acoustic features.

[0060] An adjustment module is used to obtain the configured sound parameters; generate an adjusted acoustic feature vector based on the sound parameters and the acoustic feature vector;

[0061] The sound conversion module is used to convert the adjusted acoustic feature vector into a first response voice.

[0062] In the embodiment of the present application, the multimodal text preprocessing module is specifically used to:

[0063] The first answer text is standardized using regular expressions and a natural language toolkit (such as NLTK); the standardization process is used to remove irrelevant characters (punctuation marks, special symbols, etc.) in the first answer text and to convert the first answer text into uppercase and lowercase characters;

[0064] Use BERT or other advanced NLP models to perform deep semantic analysis on the standardized first-response text to generate a text feature vector containing semantic information.

[0065] In the embodiment of the present application, the voice parameters are obtained by analyzing the voice characteristics of the mediator in the target historical mediation call data in the mediation library; specifically, the target historical mediation call data is the historical call data of successful mediation.

[0066] In an embodiment of the present application, a response text is generated in advance based on the corresponding data of the call object, and the response text is sent to the AI voice generation model.

[0067] In an embodiment of the present application, the AI speech generation model is trained using historical call data of different mediators.

[0068] In an embodiment of the present application, the data on matters to be mediated include: the bank where the debt is owed, the status of the debt, the amount of the debt, the mediation name of the current mediation project, the source of the mediation data, and data such as banned words and banned phrases; specifically, the status of the debt is the duration of overdue debt and other data.

[0069] In the embodiment of the present application, the historical call data includes: the number of historical calls, the historical call time and the call summary.

[0070] In an embodiment of the present application, the system predicts the concurrency, determines the number of call recipients for batch outbound calls based on the concurrency, and initiates batch outbound calls to the call recipients; and sends corresponding response voice to the call recipients after the call is connected.

[0071] In an embodiment of the present application, the first response text is input into a pre-trained AI speech generation model to obtain the first response speech, including:

[0072] The first response text is input into a pre-trained AI voice generation model to obtain an initial response voice; the initial response voice is voiceprint processed to obtain a first response voice.

[0073] In an embodiment of the present application, the AI speech generation model is trained using historical mediation data; the historical mediation data is historical call data between different manual customer service representatives and different call recipients.

[0074] In the embodiment of the present application, the first reply speech is analyzed to obtain a speech analysis result, including:

[0075] Convert the first reply speech into text to obtain a first reply text; preprocess the first reply text to obtain a preprocessed first reply text; perform semantic analysis on the preprocessed first reply text to obtain a semantic analysis result; input the preprocessed first reply text and the semantic analysis result into a pre-trained text classification model to obtain a semantic classification result of the first reply text; determine the semantic analysis result and the semantic classification result as the speech analysis result of the first reply speech.

[0076] In the embodiment of the present application, after obtaining the speech analysis result, the method further includes:

[0077] Match the second response text corresponding to the semantic analysis result and the semantic classification result from the configured comparison table of different semantic analysis results, different semantic classification results and corresponding second response text; input the second response text into the AI voice generation model to obtain the second response voice, and send the second response voice to the call object.

[0078] In another embodiment of the present application, the second response text may be generated based on a pre-trained AI dialogue model.

[0079] For example, suppose the caller is Mr. B, the data on the matter to be mediated is Bank A, the debt is 100,000 yuan, and the mediation is in the name of a third-party mediation agency entrusted by Bank A; the mediation data comes from Bank A; based on the identity information of the caller and the specific data on the matter to be mediated, a first response voice is generated, including: Hello, Mr. B, I am Xiao C, a staff member of the third-party mediation agency entrusted by Bank A. Now I am communicating with you on behalf of Bank A regarding the 100,000 yuan debt you owe Bank A, and would like to inform you of the following: Bank A has communicated with you on the issue of overdue debt on a certain day of a certain month of a certain year, but you have not taken any action to repay the debt. Bank A has now entrusted the court with the matter; we are now conducting pre-trial mediation with you on the matter. Do you have any difficulties?

[0080] In the above content, the contents such as "Bank A has entrusted the court with the matter", "We are now conducting pre-litigation mediation with you on the matter, do you have any difficulties?" and "I am Xiao C, a staff member of a third-party mediation agency entrusted by Bank A" are derived from the configured response template. The response template will vary depending on the data of the matters to be mediated corresponding to the call object.

[0081] In an embodiment of the present application, if the first reply voice of the callee is analyzed and the voice analysis result is that the callee is currently stating the current situation and reasons for being unable to repay the loan, the second response text is matched, and the second response text and the configured response template are combined to generate a second response voice.

[0082] In another embodiment of the present application, when analyzing the first reply voice, keyword extraction technology will be combined to extract pre-configured keywords such as "unemployment" and "sickness"; when there are corresponding keywords in the first reply voice, the response template and preparation materials corresponding to the keyword will be matched from the configured different keywords and different response templates and different preparation material comparison tables according to the keywords; a second response voice will be generated based on the response template, the second response text and the preparation materials; for example, OK, Mr. B, your situation has been received, please submit the following preparation materials to a certain email address, we will verify and contact you as soon as possible.

[0083] Step S120: If the voice analysis result satisfies the configured transfer rule, the first answer voice and the first reply voice are converted into the first answer text and the first reply text respectively.

[0084] In an embodiment of the present application, the transfer rule, i.e., the semantic analysis result, the semantic classification result, or the second response text is empty, or the semantic classification result is a transfer semantic classification result.

[0085] In an embodiment of the present application, if the response text can be matched, it means that the AI voice generation model can continue to communicate with the caller. When it is impossible to communicate with the caller, for example, the content of the caller's speech cannot be recognized or the semantic analysis results cannot be obtained, a transfer is required.

[0086] For example, in the first reply voice, the callee does not speak, or there are no pre-set keywords, or the voice analysis result cannot extract valid information (such as dialect, etc.), then it is determined that the transfer rules are met.

[0087] Step S130: Transfer the call with the callee to manual customer service, and send the identity information, historical call data, data on matters to be mediated, the first response text, and the first reply text to the manual customer service.

[0088] In an embodiment of the present application, when the AI voice generation model is unable to continue the conversation with the callee, manual customer service intervention is required. However, when the call is directly transferred, the manual customer service is not clear about the identity and other data of the callee, so the corresponding data needs to be sent to the manual customer service.

[0089] In another embodiment of the present application, the system will make outbound calls in batches, and the number of outbound calls is predicted based on the historical call connection situation: obtain historical call data and the corresponding historical call connection number and the number of historical call conversations within the configured time (that is, connected and not hung up within a certain period of time), and obtain the number of mediators who are online at the same time in the historical time period corresponding to the historical call data; input the above data into a pre-built time series prediction model to obtain a trained time series prediction model; input the current number of mediators and the time data into the trained time series prediction model to obtain the target number of outbound calls; the system calls the target number of outbound calls.

[0090] In an embodiment of the present application, the time series prediction model may be ARIMA or SARIMA.

[0091] like Figure 2 As shown, the intelligent call forwarding method of the embodiment of the present application includes:

[0092] 1. The mediator initiates AI intelligent outbound calls in batches, and the system automatically starts making batch calls.

[0093] 2. Once the call is connected, the AI's multi-round dialogue capability is called upon to start outputting audio, and the system begins a real-time conversation with the parties involved. During the conversation, the AI selects and outputs appropriate conversation content based on the contextual constraints.

[0094] 3. The system records the conversation between AI and the parties in real-time voice-to-text format.

[0095] 4. If the system determines that the AI cannot continue the conversation within 1 second, it will automatically switch to the mediator's line and synchronize the previous multiple rounds of conversation records between the AI and the parties to the mediator.

[0096] 5. The mediator will understand the progress of the AI conversation and what was said before based on the previous conversation records, so that there will be no obvious inconsistencies in language logic when the real person talks to the parties, until the mediator completes the conversation alone and ends the call.

[0097] Corresponding to the above method, the embodiment of the present application also provides an intelligent call forwarding device, such as Figure 3 As shown, the intelligent call forwarding device includes:

[0098] The sending unit 310 is configured to send a first response voice to the called party after the call is connected; wherein the first response voice is generated based on the identity information of the called party, the data of the items to be mediated, and the historical call data;

[0099] The receiving unit 320 is configured to receive a first reply voice from the callee in response to the first answer voice;

[0100] An analysis unit 330 is configured to analyze the first reply speech and obtain a speech analysis result;

[0101] a conversion unit 340 for converting the first answer voice and the first reply voice into a first answer text and a first reply text, respectively, if the voice analysis result satisfies the configured transfer rule;

[0102] The transfer unit 350 is used to transfer the call with the call object to the manual customer service, and send the identity information, historical call data, data on matters to be mediated, the first response text and the first reply text to the manual customer service.

[0103] The functions of each functional unit of the intelligent call forwarding device provided in the above embodiments of the present application can be realized through the above method steps. Therefore, the specific working process and beneficial effects of each unit in the intelligent call forwarding device provided in the embodiments of the present application will not be repeated here.

[0104] The present application also provides an electronic device, such as Figure 4 As shown, it includes a processor 410 , a communication interface 420 , a memory 430 and a communication bus 440 , wherein the processor 410 , the communication interface 420 , and the memory 430 communicate with each other via the communication bus 440 .

[0105] Memory 430, for storing computer programs;

[0106] The processor 410 is configured to execute the program stored in the memory 430 by performing the following steps:

[0107] When the call is connected, a first response voice is sent to the callee; wherein the first response voice is generated based on the identity information of the callee, data on matters to be mediated, and historical call data;

[0108] receiving a first reply voice from the call recipient in response to the first answer voice;

[0109] Analyze the first reply speech to obtain a speech analysis result;

[0110] If the voice analysis result meets the configured transfer rule, the first answer voice and the first reply voice are converted into the first answer text and the first reply text respectively;

[0111] The call with the callee will be transferred to the manual customer service, and the identity information, historical call data, data on matters to be mediated, first response text and first reply text will be sent to the manual customer service.

[0112] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. This communication bus can be divided into address buses, data buses, and control buses. For ease of illustration, the figure uses only one thick line, but this does not mean that there is only one bus or only one type of bus.

[0113] The communication interface is used for communication between the above electronic device and other devices.

[0114] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage. Alternatively, the memory may be at least one storage device located away from the processor.

[0115] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components.

[0116] The implementation methods and beneficial effects of the various components of the electronic device in the above embodiments to solve the problems can be found in Figure 1 The various steps in the embodiment shown are implemented, therefore, the specific working process and beneficial effects of the electronic device provided by the embodiment of the present application are not repeated here.

[0117] In another embodiment provided by the present application, a computer-readable storage medium is further provided. The computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute any of the intelligent call forwarding methods in the above embodiments.

[0118] In another embodiment provided by the present application, a computer program product including instructions is also provided. When the computer program product is run on a computer, the computer is enabled to execute any one of the intelligent call forwarding methods in the above embodiments.

[0119] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the embodiments of the present application may take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware. Furthermore, the embodiments of the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0120] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0121] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0122] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0123] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic creative concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0124] Obviously, those skilled in the art can make various changes and modifications to the embodiments of the present application without departing from the spirit and scope of the embodiments of the present application. Thus, if these modifications and variations of the embodiments of the present application fall within the scope of the claims and their equivalents, the embodiments of the present application are also intended to include these modifications and variations.

Claims

1. An intelligent call forwarding method, characterized in that: The method comprises: When the call is connected, a first response voice is sent to the callee; wherein, the first response voice is generated based on the identity information of the callee, the data of items to be mediated and the historical call data; the data of items to be mediated include the type of items to be mediated; the method for generating the first response voice comprises: obtaining the identity information and data of items to be mediated, as well as the historical call data of the callee; matching the items to be mediated and the response templates corresponding to the historical call data of the callee from the configured different types of items to be mediated, different historical call data and corresponding response templates; generating a first response text based on the identity information, the data of items to be mediated, the historical call data and the response template; inputting the first response text into a pre-trained The first response speech is obtained from the AI speech generation model; the AI speech generation model includes: a multimodal text preprocessing module, which is used to perform semantic analysis on the input first response text to obtain a text feature vector; an adaptive hybrid coding acoustic processing module, which is used to map the obtained text feature vector into the acoustic space: using a pre-trained Transformer-CNN-based conversion model to perform acoustic conversion on the text feature vector to obtain initial acoustic features; using a sparse attention mechanism to generate an acoustic feature vector based on the initial acoustic features; an adjustment module, which is used to obtain configured sound parameters; generating an adjusted acoustic feature vector based on the sound parameters and the acoustic feature vector; and a sound conversion module, which is used to convert the adjusted acoustic feature vector into the first response speech; receiving a first reply voice from the calling party in response to the first answer voice; Analyzing the first reply speech to obtain a speech analysis result; If the voice analysis result satisfies the configured transfer rule, converting the first answer voice and the first reply voice into a first answer text and a first reply text respectively; The call with the call object is transferred to the manual customer service, and the identity information, the historical call data, the data of the matters to be mediated, the first response text and the first reply text are sent to the manual customer service.

2. The method according to claim 1, wherein The data on matters to be mediated include: the bank where the debt is owed and the status of the debt; The historical call data includes: the number of historical calls, historical call time and call summary.

3. The method according to claim 2, wherein Inputting the first response text into a pre-trained AI speech generation model to obtain a first response speech includes: Inputting the first response text into a pre-trained AI speech generation model to obtain an initial response speech; wherein the AI speech generation model is trained using historical mediation data; the historical mediation data is historical call data between different human customer service representatives and different call recipients; The initial response voice is subjected to voiceprint processing to obtain a first response voice.

4. The method according to claim 1, wherein Analyze the first reply speech to obtain a speech analysis result, including: Converting the first reply voice into text to obtain a first response text; Preprocessing the first response text to obtain a preprocessed first response text; Performing semantic analysis on the preprocessed first response text to obtain a semantic analysis result; Inputting the preprocessed first response text and the semantic analysis result into a pre-trained text classification model to obtain a semantic classification result of the first response text; The semantic analysis result and the semantic classification result are determined as the speech analysis result of the first reply speech.

5. The method according to claim 3, wherein After obtaining the speech analysis result, the method further includes: Matching the second response text corresponding to the semantic analysis result and the semantic classification result from the configured comparison table of different semantic analysis results, different semantic classification results and corresponding second response texts; The second response text is input into the AI voice generation model to obtain a second response voice, and the second response voice is sent to the calling party.

6. The method according to claim 5, wherein The method further comprises: If the semantic analysis result, the semantic classification result or the second response text is empty, the call with the calling party will be transferred to the manual customer service, and the identity information, the historical call data, the data on matters to be mediated, the response text and the reply text during this call will be sent to the manual customer service.

7. An intelligent call forwarding device, characterized in that: The device comprises: The sending unit is used to send a first response voice to the call object when the call is connected; wherein, the first response voice is generated based on the identity information of the call object, the data of the items to be mediated and the historical call data; the data of the items to be mediated include the type of items to be mediated; the method for generating the first response voice comprises: obtaining the identity information and the data of the items to be mediated, as well as the historical call data of the call object; matching the items to be mediated and the response templates corresponding to the historical call data of the call object from the configured different types of items to be mediated, different historical call data and corresponding response templates; generating a first response text based on the identity information, the data of the items to be mediated, the historical call data and the response template; inputting the first response text into a pre-set A first response speech is obtained from the trained AI speech generation model; the AI speech generation model includes: a multimodal text preprocessing module, which is used to perform semantic analysis on the input first response text to obtain a text feature vector; an adaptive hybrid coding acoustic processing module, which is used to map the obtained text feature vector into the acoustic space: using a pre-trained Transformer-CNN-based conversion model to perform acoustic conversion on the text feature vector to obtain initial acoustic features; using a sparse attention mechanism to generate an acoustic feature vector based on the initial acoustic features; an adjustment module, which is used to obtain configured sound parameters; generating an adjusted acoustic feature vector based on the sound parameters and the acoustic feature vector; and a sound conversion module, which is used to convert the adjusted acoustic feature vector into the first response speech; A receiving unit, configured to receive a first reply voice from the calling party in response to the first answer voice; an analyzing unit, configured to analyze the first reply speech to obtain a speech analysis result; a conversion unit, configured to convert the first answer voice and the first reply voice into a first answer text and a first reply text, respectively, if the voice analysis result satisfies a configured transfer rule; The transfer unit is used to transfer the call with the call object to the manual customer service, and send the identity information, the historical call data, the data of the matters to be mediated, the first response text and the first reply text to the manual customer service.

8. An electronic device, characterized in that: The electronic device includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; Memory for storing computer programs; A processor, configured to implement the method according to any one of claims 1 to 6 when executing a program stored in a memory.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images