Voice communication method based on artificial intelligence and related equipment thereof

Abstract

The embodiment of the invention provides a voice call method based on artificial intelligence, and the method comprises the steps: creating a call task containing target customer information, a first service identifier and a second service identifier, building communication connection through an AI call-out platform, obtaining a customer audio, transmitting a first voice recognition service according to the first service identifier to obtain a customer text, and transmitting the customer text to a second voice recognition service according to the second service identifier; and generating a response text to the intelligent dialogue platform, sending the response text to the second voice generation service according to the second service identifier to obtain a response audio, playing the response audio, judging whether the client speaks during playing, and if yes, stopping playing and restarting the interaction process. According to the method, real-time response to interruption of the user is realized, the defect that the existing technical scheme only requires one question and one answer and does not support interruption is overcome, the natural fluency of dialogue is improved, ASR and TTS suppliers can be dynamically switched through the service identifier, the problem that the manufacturer of the original scheme supports singleness is solved, and the compatibility and expansibility of the system are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of communication, and particularly relates to a voice call method based on artificial intelligence and a related device thereof. BACKGROUND

[0002] In the field of intelligent telephone customer service in the service industry, with the vigorous development of technologies such as artificial intelligence, natural language processing (NLP), automatic speech recognition (ASR), text-to-speech (TTS) and machine learning (ML), the intelligent customer service industry is rapidly evolving. Its core applications include online customer service robots, telephone intelligent customer service, multi-channel integration and intelligent work order systems, and the current technology develops in the direction of supporting semantic understanding and multi-round dialogue with NLP, improving the naturalness of interaction with speech recognition and synthesis, optimizing service efficiency with knowledge graph and automation, and continuously developing towards more intelligent and humanized.

[0003] In the prior art, FreeSWITCH is used as the core application of the call center, interfaces with the ASR services of various manufacturers through the media resource control protocol (MRCP), and realizes the integration with the ASR services by means of the open source component MRCPServer, so as to convert real-time voice into text and lay a foundation for intelligent customer service. This scheme realizes the voice interaction technology support through the MRCP protocol, and to some extent, meets the basic voice recognition and interaction requirements.

[0004] However, when the existing scheme is integrated with the existing call center system, the interactive experience has obvious limitations. The system only supports a simple question and answer mode, the interaction is stiff and unnatural, the user is easy to perceive the mechanicalness of the dialogue, and generates an adverse experience, which reduces the user satisfaction. The existing technology has significant deficiencies in dealing with natural dialogue scenarios, and needs to be optimized to enhance the interrupt processing capability. SUMMARY

[0005] The purpose of the present application is to provide a voice call method based on artificial intelligence, which aims to solve the problem that the existing natural dialogue scenario only supports a simple question and answer mode, the interaction is stiff and unnatural, the user is easy to perceive the mechanicalness of the dialogue, and generates an adverse experience. The present application provides:

[0006] The first aspect of the embodiment of the present application provides a voice call method based on artificial intelligence, comprising:

[0007] creating a call task, the call task comprising target customer information, a first service identifier and a second service identifier;

[0008] perform the outbound call task through an AI outbound call platform based on the target customer information, and establish a communication connection with the target customer;

[0009] obtain customer audio;

[0010] send the customer audio to a first speech recognition service corresponding to the first service identifier to obtain customer text;

[0011] send the customer text to an intelligent dialogue platform to generate a response text;

[0012] send the response text to a second speech generation service corresponding to the second service identifier to obtain response audio;

[0013] play the response audio;

[0014] determine whether the target customer speaks during the playing of the response audio;

[0015] if the target customer speaks, stop the playing of the response audio and re-execute the steps of obtaining customer audio to playing response audio.

[0016] According to the first aspect of the method, the method further comprises: if the target customer speaks, determining whether the response audio is an interruptible statement; if the response audio is the interruptible statement, not executing the step of stopping the playing of the response audio and re-executing the steps of obtaining customer audio to playing response audio.

[0017] According to the first aspect of the method, the method further comprises:

[0018] real-time recognition of the customer text; if the customer text is an alert statement, real-time playing of a preset response statement; the playing of the preset response statement can overlap with the speaking of the target customer.

[0019] According to the first aspect of the method, the step of determining whether the target customer speaks during the playing of the response audio comprises:

[0020] determining whether the target customer speaks for more than a preset time during the playing of the response audio;

[0021] if the target customer speaks for more than the preset time, it is determined that the target customer speaks.

[0022] Based on the method for voice call based on artificial intelligence provided in the first aspect of the embodiments of the present application, optionally, different lengths of silence time can be set in advance for different response audios.

[0023] Based on the method for voice call based on artificial intelligence provided in the first aspect of the embodiments of the present application, optionally,

[0024] The step of sending the customer audio to the first voice recognition service corresponding to the first service identifier is initiated by a call-dp service.

[0025] Based on the method for voice call based on artificial intelligence provided in the first aspect of the embodiments of the present application, optionally, the call task is executed based on a FreeSWITCH application.

[0026] The second aspect of the embodiments of the present application provides a voice call device based on artificial intelligence, comprising:

[0027] A creating unit is configured to create a call task, wherein the call task comprises target customer information, a first service identifier and a second service identifier.

[0028] An establishing unit is configured to execute an outbound call task based on the target customer information through an AI outbound call platform, and establish a communication connection with a target customer.

[0029] An obtaining unit is configured to obtain customer audio.

[0030] A first sending unit is configured to send the customer audio to a first voice recognition service corresponding to the first service identifier, and obtain customer text.

[0031] A generating unit is configured to send the customer text to an intelligent dialogue platform, and generate response text.

[0032] A second sending unit is configured to send the response text to a second voice generation service corresponding to the second service identifier, and obtain response audio.

[0033] A playing unit is configured to play the response audio.

[0034] A judging unit is configured to judge whether the target customer speaks during the playing of the response audio.

[0035] A interrupting unit is configured to, if the target customer speaks, interrupt the playing of the response audio, and re-execute the steps of obtaining customer audio to playing response audio.

[0036] Based on the voice call device based on artificial intelligence provided in the second aspect of the embodiments of the present application, optionally, the interrupting unit is further configured to:

[0037] If the target customer has speech, it is determined whether the response audio is an interruptible sentence, if the response audio is the interruptible sentence, the playing process of the response audio is not interrupted, and the step of obtaining the customer audio to the playing of the response audio is re-executed.

[0038] According to the speech call device based on artificial intelligence provided in the second aspect of the embodiments of the present application, optionally, the device further comprises a warning unit configured to:

[0039] The customer text is identified in real time, if the customer text is a warning sentence, a preset response sentence is played in real time, and the playing process of the preset response sentence can overlap with the speech process of the target customer.

[0040] According to the speech call device based on artificial intelligence provided in the second aspect of the embodiments of the present application, optionally, the judging unit is specifically configured to:

[0041] It is determined whether the target customer has speech for more than a preset time length during the playing process of the response audio.

[0042] If the target customer has speech for more than a preset time length, it is determined that the target customer has speech.

[0043] According to the speech call device based on artificial intelligence provided in the second aspect of the embodiments of the present application, optionally, different lengths of silence time can be set in advance for different response audios.

[0044] According to the speech call device based on artificial intelligence provided in the second aspect of the embodiments of the present application, optionally, the step of sending the customer audio to the first speech recognition service corresponding to the first service identifier is initiated by a call-dp service.

[0045] According to the speech call device based on artificial intelligence provided in the second aspect of the embodiments of the present application, optionally, the call task is executed based on a FreeSWITCH application.

[0046] The third aspect of the embodiments of the present application provides a speech call device based on artificial intelligence, comprising:

[0047] A central processing unit, a memory, an input and output interface, a wired or wireless network interface and a power supply;

[0048] The memory is a transitory memory or a persistent memory;

[0049] The central processing unit is configured to communicate with the memory, execute the instruction operation in the memory on the device to execute the method as described in any one of the first aspects of the embodiments of the present application.

[0050] The fourth aspect of the embodiments of the present application provides a computer readable storage medium comprising instructions which, when executed on a computer, cause the computer to perform the method according to any one of the first aspect of the embodiments of the present application.

[0051] The fifth aspect of the embodiments of the present application provides a computer program product comprising instructions which, when executed on a computer, cause the computer to perform the method according to any one of the first aspect of the embodiments of the present application.

[0052] From the above technical solutions, the embodiments of the present application have the following advantages: the embodiments of the present application provide a voice call method based on artificial intelligence, which comprises: creating a call task, the call task comprising target customer information, a first service identifier and a second service identifier; executing the outbound task based on the target customer information through an AI outbound platform to establish a communication connection with the target customer; obtaining customer audio; sending the customer audio to a first speech recognition service corresponding to the first service identifier to obtain customer text; sending the customer text to an intelligent dialogue platform to generate a response text; sending the response text to a second speech generation service corresponding to the second service identifier to obtain response audio; playing the response audio; determining whether the target customer speaks during the playing of the response audio; if the target customer speaks, stopping the playing of the response audio and re-executing the steps of obtaining customer audio to playing the response audio. Real-time response to user interruption behavior is achieved. Compared with the defect of the MRCP scheme in the background art that "only supports one question and one answer mode and does not support interruption", the scheme enables the system to detect user interruption when the customer speaks and immediately interrupt the current voice playing to receive user input, significantly improving the naturalness and fluency of the dialogue and solving the "interaction mechanicalness" problem. In addition, the scheme also realizes dynamic switching of ASR and TTS suppliers by "including the first service identifier and the second service identifier in the call task to correspondingly call different speech recognition and generation services". The scheme can flexibly connect to multiple manufacturers such as Tencent and iFLYER through the service identifier, solves the problem of single manufacturer support in the original scheme, and improves the compatibility and expansibility of the system. BRIEF DESCRIPTION OF DRAWINGS

[0053] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed in the embodiments or the prior art description will be briefly introduced below. Obviously, the drawings in the following description are only embodiments of the present application, and those skilled in the art can obtain other drawings according to the provided drawings without creative labor. It can be understood that the drawings provided in this part are only for better understanding of the present application and do not constitute a limitation on the present application.

[0054] Figure 1A flowchart of an embodiment of the AI-based voice call method provided in the present application;

[0055] Figure 2 A flowchart of a user speech judgment process in an embodiment of the AI-based voice call method provided in the present application.

[0056] Figure 3 A timing flowchart of an AI voice call process in the prior art;

[0057] Figure 4 A timing flowchart of an embodiment of the AI-based voice call method provided in the present application;

[0058] Figure 5 A structural diagram of an AI-based voice call device provided in the present application;

[0059] Figure 6 Another structural diagram of an AI-based voice call device provided in the present application. DETAILED DESCRIPTION

[0060] In order to make the personnel in the technical field better understand the present application scheme, the technical solutions in the embodiments of the present application are described clearly and completely below. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by the person skilled in the art without creative labor should belong to the scope of protection of the present application. At the same time, in order to describe clearly and concisely, the description of well-known functions and structures is omitted in the following description.

[0061] The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, product or device that includes a series of steps or units does not have to be limited to only those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0062] In the field of intelligent phone customer service in the service industry, with the booming development of technologies such as artificial intelligence, natural language processing (NLP), automatic speech recognition (ASR), text-to-speech (TTS) and machine learning (ML), the intelligent customer service industry is rapidly evolving. Its core applications include online customer service robots, telephone intelligent customer service, multi-channel integration and intelligent work order systems, and the current technology is developing in the direction of supporting semantic understanding and multi-round dialogue with NLP, improving the naturalness of interaction with speech recognition and synthesis, optimizing service efficiency with knowledge graph and automation, and continuously developing towards more intelligent and humanized.

[0063] In the prior art, FreeSWITCH is used as the core application of the call center, interfaces with the ASR services of major manufacturers through the media resource control protocol (MRCP), and integrates with the ASR services by means of the open source component MRCPServer, which can convert real-time speech into text and lay the foundation for intelligent customer service. This scheme realizes voice interaction technology support through the MRCP protocol, and to some extent meets the basic voice recognition and interaction requirements.

[0064] However, the existing scheme has obvious limitations in the interaction experience when integrated with existing call center systems. The system only supports a simple question and answer mode, the interaction is stiff and unnatural, users are easily aware of the mechanical nature of the dialogue, resulting in a poor experience and reducing user satisfaction. The existing technology has significant shortcomings in dealing with natural dialogue scenarios, and needs to be optimized to enhance processing capacity.

[0065] To solve the above problems, please refer to Figure 1 One embodiment of the voice call method based on artificial intelligence provided by the present application includes steps 101-109.

[0066] 101. Create a call task, which includes target customer information, a first service identifier and a second service identifier.

[0067] The task contains target customer information (such as contact information, identity), a first service identifier (specifying an ASR speech recognition service provider such as Tencent or iFLYER), and a second service identifier (specifying a TTS speech synthesis service provider), which clearly specifies the target of the call and the configuration of the service provider.

[0068] 102. Perform the outbound task based on the target customer information through an AI outbound platform to establish a communication connection with the target customer.

[0069] The AI outbound call platform is an automatic call system based on artificial intelligence technology. As an intermediate hub connecting Freeswitch and target customers, its function is to undertake call tasks and perform outbound call operations, which can batch process outbound call requirements and realize call process control, such as automatically screening target customers according to preset strategies, allocating call resources, and connecting subsequent audio stream processing modules (such as call-dp services) after call establishment. The platform sends a call request to the operator or call center through the Freeswitch application based on the target customer information (such as phone number, customer identification, etc.) contained in the call task created in step 101, and calls the underlying communication interface to establish a voice link with the target customer.

[0070] 103, obtaining customer audio;

[0071] Specifically, the customer's voice reply is collected. During the process, the AI customer service can first play a fixed greeting (such as an opening speech), and then receive the audio stream feedback from the customer.

[0072] 104. sending the customer audio to a first speech recognition service corresponding to the first service identification to obtain customer text;

[0073] Specifically, the customer audio is sent to the ASR service provider corresponding to the first service identification, and the customer text is generated through voice-to-text technology. In actual implementation, the ASR service provider can be, for example, Tencent, iFlytek, Byte, and Tuba Rabbit self-research, etc., which is determined according to the actual needs of the outbound call task, and is not limited here.

[0074] 105. sending the customer text to an intelligent dialogue platform to generate a response text;

[0075] The customer text is transmitted to the intelligent dialogue platform, which generates a response text based on enterprise preset dialogues, knowledge graphs, etc. The intelligent dialogue platform can be generated by machine learning, neural network models, etc., and contains various settings, which are not limited here.

[0076] 106. sending the response text to a second speech generation service corresponding to the second service identification to obtain a response audio;

[0077] The response text is sent to the TTS service provider corresponding to the second service identification to generate a playable response audio. Specifically, the TTS service provider can be Tencent, Byte, Tuba Rabbit self-research, etc., which is determined according to the actual needs of the outbound call task, and is not limited here.

[0078] 107. playing the response audio;

[0079] Specifically, the synthesized voice reply is played to the target customer, thereby realizing information feedback in human-computer interaction.

[0080] 108. determining whether the target customer has a speech during the playing of the response audio;

[0081] Specifically, whether the customer has a speech (e.g., the voice energy value exceeds the threshold) during the playing of the response audio is monitored in real time. If there is a speech, the interrupt logic is triggered, otherwise, the playing is continued until the end. In actual implementation, various identification methods can be used to determine whether the customer has a speech. User interruption will cause audio signal overlap (i.e., the "double speaking" scenario), at which time the two voices need to be accurately separated and the user speech needs to be identified. Specifically, in this case, the traditional voice activity detection (VAD) algorithm is difficult to accurately distinguish the voice spectrum features of the AI customer service and the user. Because the voice signals of the two may have similarities in frequency, amplitude, etc., especially when the user's voice is small or the voice frequency range is similar to that of the AI customer service, the algorithm is prone to "missed judgment", i.e., failing to detect the user's speech; or "misjudgment", i.e., mistakenly identifying the AI customer's voice as the user's speech. Therefore, in actual implementation, a deep learning sound source separation model can be used to learn the differences between the two in spectral features, fundamental frequency patterns, and sound timing by training a large amount of "double speaking" scenario data (AI customer service and user voice overlap samples), so as to realize real-time stripping of the user's voice in the mixed audio.

[0082] It should be noted that background noise (e.g., TV sound, keyboard typing), device current sound, or user non-verbal sound (coughing, breathing) may trigger false interruption. For these problems, a model needs to be trained and judged, i.e., the model used can collect background noise features, device noise features, and non-verbal sound features, thereby further improving the judgment accuracy.

[0083] If the judgment result is that there is, it means that the speech overlaps, and step 109 is executed, if there is not, the audio is continuously played until the end.

[0084] 109. If there is a speech, the playing process of the response audio is stopped, and the steps of obtaining the customer audio and playing the response audio are re-executed.

[0085] Specifically, when it is detected that the target customer has a speech during the playing of the response audio, the system immediately stops the playing of the current response audio, and re-executes the whole process of obtaining the customer audio (step 103), sending the customer audio to the first speech recognition service (step 104), generating the customer text and transmitting it to the intelligent dialogue platform (step 105), obtaining the response audio through the second speech generation service (step 106), and playing the response audio (step 107), so as to realize real-time response to user interruption and ensure that the interaction prioritizes processing user input content according to natural dialogue logic.

[0086] From the above technical solution can be seen, the embodiment of the application has the following advantages: the embodiment of the application provides a voice call method based on artificial intelligence, comprising: creating a call task, the call task comprising target customer information, a first service identifier and a second service identifier; executing the outbound task based on the target customer information through an AI outbound platform to establish a communication connection with the target customer; obtaining customer audio; sending the customer audio to a first speech recognition service corresponding to the first service identifier to obtain customer text; sending the customer text to an intelligent dialogue platform to generate a response text; sending the response text to a second speech generation service corresponding to the second service identifier to obtain response audio; playing the response audio; determining whether the target customer speaks during the playing of the response audio; if the target customer speaks, stopping the playing of the response audio and re-executing the steps of obtaining customer audio to playing the response audio. Real-time response to user interruption behavior is realized. Compared with the defect of the MRCP scheme in the background art that "only supports one question and one answer mode and does not support interruption", the scheme enables the system to detect user interruption when the customer speaks and immediately interrupts the current voice playing to receive user input, significantly improving the naturalness and fluency of the dialogue and solving the "interaction mechanicalness" problem. In addition, the present scheme also realizes the dynamic switching of ASR and TTS suppliers by "including the first service identifier and the second service identifier in the call task to correspondingly call different speech recognition and generation services". The present scheme can flexibly connect to multiple manufacturers such as Tencent and iFLYOS through the service identifier, solves the problem of single manufacturer support in the original scheme, and improves the compatibility and expansibility of the system.

[0087] To facilitate the use of the present method in actual real-time process, the present application further describes in detail whether there is an interruption process as follows, please refer to Figure 2 , comprising steps 201-203

[0088] 201. determining whether the target customer speaks during the playing of the response audio,

[0089] Sometimes, the user only replies "um" "yes", and the terminal is not needed at this time. It is determined whether the target customer speaks beyond a preset time length, such as more than 2 seconds.

[0090] If the target customer speaks beyond the preset time length, it is determined that the target customer speaks.

[0091] Alternatively, whether there is speech can also be determined by keyword judgment. For example, the customer says "wait a moment" "I ask" and the like, indicating that the user has the desire to speak. At this time, although the time length is insufficient, it can still be determined that there is speech.

[0092] 202. judging whether the response audio is an un-interruptable sentence

[0093] If there is a speech, judging whether the response audio is an un-interruptable sentence, if the response audio is the un-interruptable sentence, not performing the process of interrupting the playing of the response audio, and re-performing the step of obtaining the customer audio to the playing of the response audio.

[0094] Specifically, the un-interruptable sentence can be, for example, an opening sentence, such as "Hello, I am a decoration engineer of TuBarabbit, and the call is to see that you have browsed the decoration information on the Internet", wherein "Hello, I am a decoration engineer of TuBarabbit" is not allowed to be interrupted, and the following can be interrupted.

[0095] 203. interrupting the playing of the response audio.

[0096] That is, when the user has a speech and the response audio is not an un-interruptable sentence, interrupting the playing of the response audio, and re-answering according to the user audio, and in other cases, the corresponding audio still needs to be played.

[0097] It can be understood that the present scheme can also include the identification of the warning sentence, that is, the customer text is identified in real time, if the customer text is identified as a warning sentence, a preset response sentence is played in real time, and the playing process of the preset response sentence can overlap with the speaking process of the target customer. The warning sentence can be a sentence expressing extreme impatience or abuse of the customer, at which time the customer needs to be responded to as soon as possible, and the preset response sentence can be a response such as "I am sorry to disturb you, now hang up" and the like, and the specific type is not limited here. The identification process of the warning sentence can occur at any time point, and the specific time point is not limited here.

[0098] From the above technical solution can be seen, the embodiment of the application has the following advantages: the embodiment of the application provides a voice call method based on artificial intelligence, comprising: creating a call task, the call task comprising target customer information, a first service identifier and a second service identifier; executing the outbound task based on the target customer information through an AI outbound platform to establish a communication connection with the target customer; obtaining customer audio; sending the customer audio to a first speech recognition service corresponding to the first service identifier to obtain customer text; sending the customer text to an intelligent dialogue platform to generate a response text; sending the response text to a second speech generation service corresponding to the second service identifier to obtain response audio; playing the response audio; determining whether the target customer speaks during the playing of the response audio; if the target customer speaks, stopping the playing of the response audio and re-executing the steps of obtaining customer audio to playing response audio. Real-time response to user interruption behavior is realized. Compared with the defect of the MRCP scheme in the background art that "only supports one question and one answer mode and does not support interruption", the scheme enables the system to detect user interruption when the customer speaks and immediately interrupt the current voice playing, and instead receives user input, significantly improving the naturalness and fluency of the dialogue and solving the "interaction mechanicalness" problem. In addition, the present scheme also realizes the dynamic switching of ASR and TTS suppliers by "including the first service identifier and the second service identifier in the call task and corresponding to calling different speech recognition and generation services". The present scheme can flexibly interface with Tencent, iFlytek and other manufacturers through the service identifier, solving the single manufacturer support problem of the original scheme and improving the compatibility and expansibility of the system.

[0099] The above describes the execution process of the present scheme, and the implementation architecture of the present scheme is described below.

[0100] Specifically, please refer to Figure 3 , Figure 3 is a flowchart of the AI customer service outbound process in the prior art environment, which needs to support the initiation of the speech recognition process through the MRCP-Server.

[0101] Figure 4is a flowchart of an AI customer service call-out process for implementing the scheme, and the main functions are integrated in the call-dp service. The step of sending the customer audio to the first speech recognition service corresponding to the first service identifier is initiated by the call-dp service. The call task is executed based on the FreeSWITCH application. After the call is connected, the call-dp scheme sends a command to Freeswitch, and Freeswitch pushes data to the call-dp application. In addition, the call-dp scheme supports dynamic silent duration judgment during the call process. The silent duration can be changed each time, that is, different lengths of silent duration can be set for different response audios. The specific setting is determined by the actual setting, which is not limited here. In addition, the call service (call-dp) is also used for dynamically switching ASR and TTS suppliers, directly solving the problem of rigid interaction and single manufacturer support in the prior art.

[0102] The above describes the voice call method based on artificial intelligence provided by the present application. To support the implementation of the above embodiments, the present application also provides a voice call device based on artificial intelligence. Please refer to Figure 5 An embodiment of the voice call device based on artificial intelligence provided by the present application includes:

[0103] The creating unit 501 is configured to create a call task, and the call task includes target customer information, a first service identifier, and a second service identifier.

[0104] The establishing unit 502 is configured to execute the call-out task through an AI call-out platform based on the target customer information, and establish a communication connection with the target customer.

[0105] The obtaining unit 503 is configured to obtain customer audio.

[0106] The first sending unit 504 is configured to send the customer audio to a first speech recognition service corresponding to the first service identifier, and obtain customer text.

[0107] The generating unit 505 is configured to send the customer text to an intelligent dialogue platform, and generate response text.

[0108] The second sending unit 506 is configured to send the response text to a second speech generation service corresponding to the second service identifier, and obtain response audio.

[0109] The playing unit 507 is configured to play the response audio.

[0110] The judging unit 508 is configured to judge whether the target customer speaks during the playing of the response audio.

[0111] The middle unit 509 is configured to, if there is speech, suspend the playing process of the response audio, and re-execute the step of acquiring the customer audio to the playing of the response audio.

[0112] According to the voice call device based on artificial intelligence provided in the second aspect of the embodiments of the present application, optionally, the middle unit is further configured to:

[0113] If there is speech, it is determined whether the response audio is an interruptible statement. If the response audio is the interruptible statement, the step of suspending the playing process of the response audio is not executed, and the step of re-acquiring the customer audio to the playing of the response audio is re-executed.

[0114] According to the voice call device based on artificial intelligence provided in the second aspect of the embodiments of the present application, optionally, the device further comprises a warning unit configured to:

[0115] The customer text is identified in real time. If the customer text is identified as a warning statement, a preset response statement is played in real time. The playing process of the preset response statement can overlap with the speaking process of the target customer.

[0116] According to the voice call device based on artificial intelligence provided in the second aspect of the embodiments of the present application, optionally, the determination unit is specifically configured to:

[0117] It is determined whether the target customer speaks for more than a preset time length during the playing process of the response audio.

[0118] If the target customer speaks for more than a preset time length, it is determined that the target customer has speech.

[0119] According to the voice call device based on artificial intelligence provided in the second aspect of the embodiments of the present application, optionally, different lengths of silence time can be set in advance for different response audios.

[0120] According to the voice call device based on artificial intelligence provided in the second aspect of the embodiments of the present application, optionally, the step of sending the customer audio to the first voice recognition service corresponding to the first service identifier is initiated by a call-dp service.

[0121] According to the voice call device based on artificial intelligence provided in the second aspect of the embodiments of the present application, optionally, the call task is executed based on a FreeSWITCH application.

[0122] In the embodiments, the processes performed by each unit of the voice call device based on artificial intelligence are similar to the method processes described in the foregoing Figure 1 corresponding embodiments, which will not be described here.

[0123] Figure 6 is a structural schematic diagram of a voice call device server based on artificial intelligence provided by an embodiment of the present application. The server 600 can include one or more central processing units (CPUs) 601 and a memory 605, and the memory 605 stores one or more application programs or data.

[0124] In the embodiment, the specific function module division in the central processing unit 601 can be similar to the function module division manner of the units described in the foregoing Figure 5 , and details are not repeated here.

[0125] The memory 605 can be volatile storage or persistent storage. The program stored in the memory 605 can include one or more modules, and each module can include a series of instruction operations in the server. Further, the central processing unit 601 can be configured to communicate with the memory 605 and execute the series of instruction operations in the memory 605 on the server 600.

[0126] The server 600 can also include one or more power supplies 602, one or more wired or wireless network interfaces 603, one or more input / output interfaces 604, and / or one or more operating systems, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.

[0127] The central processing unit 601 can perform the operations performed by the voice call method based on artificial intelligence in the foregoing Figure 1 embodiment, and details are not repeated here.

[0128] The embodiment of the present application also provides a computer storage medium for storing computer software instructions for the voice call method based on artificial intelligence, which includes a program designed for the voice call method based on artificial intelligence.

[0129] The voice call method based on artificial intelligence can be the voice call method based on artificial intelligence described in the foregoing Figure 1 Figure 2 .

[0130] The embodiment of the present application also provides a computer program product including computer software instructions, which can be loaded by a processor to implement the flow of the voice call method based on artificial intelligence of any one of the foregoing Figure 1 Figure 2 .

[0131] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the embodiments of the device described above are merely schematic, and the unit division is merely logical function division, and there can be other division manners in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some characteristics can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, or the among different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0132] The units described as separated components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purposes of the embodiments of the present application.

[0133] In addition, each functional unit in the embodiments of the present application can be integrated in a processing unit, or each unit can exist physically as a separate unit, or two or more units can be integrated in a unit. The integrated unit can be implemented in a form of hardware, or in a form of software functional unit.

[0134] The above only provides preferred embodiments of the present application, and is not used to limit the present application, and any modification, equivalent replacement or improvement made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A voice call method based on artificial intelligence, characterized in that, The method comprises the following steps: creating a call task, wherein the call task comprises target customer information, a first service identifier and a second service identifier; executing the call task through an AI outbound call platform based on the target customer information to establish a communication connection with the target customer; obtaining customer audio; sending the customer audio to a first speech recognition service corresponding to the first service identifier to obtain customer text; sending the customer text to an intelligent dialogue platform to generate response text; sending the response text to a second speech generation service corresponding to the second service identifier to obtain response audio; playing the response audio; judging whether the target customer speaks during the playing of the response audio; if the target customer speaks, stopping the playing of the response audio and re-executing the steps of obtaining customer audio to playing response audio. 2.The AI-based voice call method of claim 1, wherein, Before the step of stopping the playing of the response audio, the method further comprises the following steps: 3.The AI-based voice call method of claim 1, wherein, if the target customer speaks, judging whether the response audio is an unbreakable sentence, and if the response audio is the unbreakable sentence, not executing the step of stopping the playing of the response audio and re-executing the steps of obtaining customer audio to playing response audio. The method further comprises the following steps: 4.The AI-based voice call method of claim 1, wherein, identifying the customer text in real time, and if the customer text is a warning sentence, playing a preset response sentence in real time, wherein the playing of the preset response sentence can overlap with the speaking of the target customer. The step of judging whether the target customer speaks during the playing of the response audio comprises the following steps: judging whether the target customer speaks for more than a preset time length during the playing of the response audio; 5.The AI-based voice call method of claim 1, wherein, if the target customer speaks for more than the preset time length, it is determined that the target customer speaks. Different lengths of silence time can be preset for different response audios.

6. The AI-based voice call method according to claim 1, wherein the step of sending the customer audio to the first speech recognition service corresponding to the first service identifier is initiated by a call-dp service.

7. The AI-based voice call method according to claim 1, wherein the call task is executed based on a FreeSWITCH application. The method comprises the following steps:

8. An artificial intelligence-based voice call device, characterized by, a creating unit configured to create a call task, wherein the call task comprises target customer information, a first service identifier and a second service identifier; an establishing unit configured to execute the call task through an AI outbound call platform based on the target customer information to establish a communication connection with the target customer; an obtaining unit configured to obtain customer audio; a first sending unit configured to send the customer audio to a first speech recognition service corresponding to the first service identifier to obtain customer text; a generating unit configured to send the customer text to an intelligent dialogue platform to generate response text; a second sending unit configured to send the response text to a second speech generation service corresponding to the second service identifier to obtain response audio; a playing unit configured to play the response audio; a judging unit configured to judge whether the target customer speaks during the playing of the response audio; ​ A middle unit is configured to, if there is a speech, interrupt the playing process of the response audio and re-perform the step of obtaining the customer audio to the playing of the response audio. 9.A voice call device based on artificial intelligence, characterized in that, The method comprises the steps of: A central processing unit, a memory, an input / output interface, a wired or wireless network interface and a power supply; The memory is a volatile memory or a persistent memory; The central processing unit is configured to communicate with the memory, execute the instruction operation in the memory on the device to perform the method as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium comprises instructions, when the instructions are run on the computer, the computer is caused to perform the method as claimed in any one of claims 1 to 7.

Images