Voice service method based on voice automatic grading and voice customer service platform
By acquiring users' voice and text information, extracting and analyzing key information, and determining the level of communication information, the problem of customer service platforms being unable to accurately analyze user needs has been solved, thus achieving efficient and accurate voice services.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- STATE GRID JIANGSU ELECTRIC POWER CO LIANYUNGANG POWER SUPPLY CO
- Filing Date
- 2022-04-18
- Publication Date
- 2026-07-14
AI Technical Summary
The existing customer service platform is unable to accurately analyze user communication information, resulting in an inability to provide efficient and accurate voice services.
By acquiring users' voice and text information, key information, including key voices, keywords, and key derived information, is extracted, analyzed to determine the level of the communication information, and the type and level of the voice service task are determined based on the level.
It enables detailed and accurate classification and demand identification of user communication information, providing efficient and accurate voice services.
Smart Images
Figure CN114783416B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of computer technology, and in particular to a voice service method and system based on automatic voice classification. Background Technology
[0002] Currently, for user communication and interviews in various scenarios of electricity services (such as consultation, coordination, complaints, etc.), the customer service platform usually first processes the user's voice information, analyzes the user's needs, records them, and provides corresponding solutions.
[0003] However, the processing capabilities of existing customer service platforms are relatively limited, which may prevent them from accurately analyzing the needs expressed by users and thus from ensuring that they can provide corresponding voice services efficiently and accurately based on the information exchanged between users.
[0004] Therefore, it is particularly important to provide users with efficient and accurate voice services by accurately analyzing their communication information. Summary of the Invention
[0005] The purpose of this invention is to provide a voice service method and system based on automatic voice classification. This application embodiment can provide users with efficient and accurate voice services. The specific technical solution is as follows:
[0006] In a first aspect of the present invention, a voice service method based on automatic voice classification is provided, comprising: acquiring user communication information; wherein the communication information includes voice information and text information during the user's communication process; processing the communication information to extract key information; wherein the key information includes key voices, keywords, and key derived information; analyzing the key information to obtain a first level corresponding to the communication information; and determining a voice service task for the communication information of the first level based on the communication information of the first level, wherein the voice service task includes a voice service type and a second level.
[0007] Optionally, processing the communication information and extracting key information includes:
[0008] The communication information is input into the key information extraction model to obtain the intermediate classification results of the communication information;
[0009] Based on the intermediate classification results, extract the key information;
[0010] Wherein, if the key information is key speech, the classification criteria for the classification result include the speech emotion and speech rate corresponding to the communication information;
[0011] If the key information is a keyword, then the classification criteria for the classification result include the number of emotional and degree words corresponding to the communication information.
[0012] Optionally, the key derived information includes key speaker information of the communication information, which is determined based on one or more of the following: speaking frequency, speaking percentage, and number of interruptions to customer service voice in the key voice.
[0013] Optionally, the step of analyzing the key information to obtain the first level corresponding to the communication information includes:
[0014] Based on the key information and combined with the first preset conditions, the primary level of the communication information is determined; wherein, the primary level includes the urgency level and importance level of the communication information;
[0015] Based on the primary level of the communication information and the key information, and in conjunction with the second preset condition, the sub-level corresponding to the communication information under the primary level is determined; wherein, the level of the sub-level is positively correlated with the level of urgency and the level of importance.
[0016] Optionally, the method further includes:
[0017] If the sub-level of the communication information is at the highest, obtain the degree of deviation between the key information and the second preset condition;
[0018] If the degree of deviation meets the third preset condition, the primary level of the communication information will be adjusted.
[0019] Optionally, based on the communication information of the first level, a voice service task for the communication information of the first level is determined, including:
[0020] The type of the voice service task is determined based on the urgency level in the first level of the communication information, and the type includes human voice service and intelligent voice service.
[0021] Optionally, the method further includes:
[0022] Based on the importance level in the first level of the communication information, a second level matching the first level is determined, the second level including the artificial voice service level and the intelligent voice service level;
[0023] If the voice service task is the human voice service, then according to the human voice service level, match at least one of the following screening criteria among the positive review rate, rating and years of service of the corresponding human voice service personnel, and determine the human voice service personnel who meet the screening criteria as the executors of the human voice service.
[0024] If the voice service task is the intelligent voice service, then the corresponding intelligent voice library is matched according to the intelligent voice service level, so as to provide voice services to the user based on the voice content of the intelligent voice library.
[0025] Optionally, the method further includes:
[0026] Based on the voice service task for the first level of communication information, the communication outline is updated; wherein, the communication outline is used to formulate the communication content between the voice service customer service representative and the user.
[0027] In another aspect of the present invention, a voice service platform based on automatic voice classification is provided, the service platform comprising:
[0028] The communication information acquisition module is used to acquire the user's communication information; wherein, the communication information includes voice information and text information during the user's communication process;
[0029] A key information extraction module is used to process the communication information and extract key information; wherein, the key information includes key voice, keywords, and key derived information;
[0030] The key information analysis module is used to analyze the key information to obtain the first level corresponding to the communication information;
[0031] The task creation module is used to determine a voice service task for the first level of communication information based on the first level of communication information. The voice service task includes the type of voice service and the second level.
[0032] In another aspect of the present invention, a computer-readable storage medium is provided, the storage medium storing computer instructions, and when a computer reads the computer instructions in the storage medium, the computer executes the method described above. Attached Figure Description
[0033] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0034] Figure 1 This is a schematic diagram illustrating an application scenario of the voice service platform based on automatic voice classification provided in this application embodiment;
[0035] Figure 2 This is a flowchart illustrating the voice service method based on automatic voice classification provided in an embodiment of this application;
[0036] Figure 3 This is a schematic diagram of the structure of the voice service platform provided in the embodiments of this application;
[0037] Figure 4 This is an internal structural diagram of the computer device provided in the embodiments of this application. Detailed Implementation
[0038] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this specification. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.
[0039] It should be understood that the terms “system,” “device,” “unit,” and / or “module” used herein are one way to distinguish different components, elements, parts, sections, or assemblies at different levels. However, if other terms can achieve the same purpose, they may be replaced by other expressions.
[0040] As indicated in this specification and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of expressly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.
[0041] Flowcharts are used in this specification to illustrate the operations performed by the system according to embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the steps can be processed in reverse order or simultaneously. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.
[0042] Figure 1 This is a schematic diagram illustrating an application scenario of a voice service platform (hereinafter referred to as "voice service platform") based on automatic speech classification, according to some embodiments of this application. Figure 1 As shown, the voice service platform 100 may include a server 110, a network 120, a first user terminal 130, a second user terminal 140, and a storage device 150.
[0043] Server 110 can process data and / or information obtained from at least one component of voice service platform 100 (e.g., first user terminal 130, second user terminal 140, and storage 150) or from external data sources (e.g., cloud data centers). For example, server 110 can obtain interactive instructions from first user terminal 130 (e.g., passenger terminal). As another example, server 110 can also obtain historical data from storage 150.
[0044] In some embodiments, server 110 may include processing device 112. Processing device 112 may process information and / or data related to the human-computer interaction system to perform one or more functions described herein. For example, processing device 112 may determine a voice service based on interactive instructions and / or historical data. In some embodiments, processing device 112 may include at least one processing unit (e.g., a single-core processing engine or a multi-core processing engine). In some embodiments, processing device 112 may be part of a first client 130 and / or a second client 140.
[0045] Network 120 can provide a channel for information exchange. In some embodiments, network 120 may include one or more network access points. One or more components of voice service platform 100 may connect to network 120 through the access points to exchange data and / or information. In some embodiments, at least one component of voice service platform 100 may access data or instructions stored in memory 150 via network 120.
[0046] The owner of the first user terminal 130 can be the user or someone else. For example, owner A of the first user terminal 130 can use the first user terminal 130 to send a service request to user B. In some embodiments, the first user terminal 130 may include various devices with information receiving and / or sending functions. The first user terminal 130 can process information and / or data. In some embodiments, the first user terminal 130 may be a device with positioning function. The first user terminal 130 may be a device with display function, displaying the text content of the voice service fed back to the first user terminal 130 by the server 110, and the display method may be an interface, pop-up window, floating window, small window, text, etc. The first user terminal 130 may be a device with voice function, thereby playing the voice service fed back to the first user terminal 130 by the server 110.
[0047] The second user terminal 140 can communicate with the first user terminal 130. In some embodiments, the first user terminal 130 and the second user terminal 140 can communicate via a short-range communication device. In some embodiments, the type of the second user terminal 140 can be the same as or different from that of the first user terminal 130. For example, the first user terminal 130 and the second user terminal 140 can be, but are not limited to, tablet computers, laptop computers, mobile devices, desktop computers, and any combination thereof.
[0048] In some embodiments, memory 150 may store data and / or instructions that processing device 112 may execute or use to perform the exemplary methods described herein. For example, memory 150 may store historical data, models for determining voice services, audio files for voice services, text files, etc. In some embodiments, memory 150 may be directly connected to server 110 as back-end storage. In some embodiments, memory 150 may be part of server 110, first client 130, and / or second client 140.
[0049] Figure 2 This document illustrates a flowchart of a voice service method based on automatic voice classification, as provided in an embodiment of this application. Figure 2 As shown, a voice service method based on automatic voice classification includes the following steps:
[0050] Step 210: Obtain the user's communication information.
[0051] In some embodiments, image data or video data of the user's hand can be collected by an image acquisition device, such as a camera, light field camera, or dashcam, and the collected data can be uploaded to the system.
[0052] The communication information may include voice and text messages exchanged during the user interaction. In some scenarios, users can communicate with voice customer service representatives; for example, users can log into a voice service platform through a terminal to make inquiries, coordinate, or file complaints. It is understood that the communication process can also take the form of interviews, such as conducting interviews with multiple users simultaneously.
[0053] Text information can be obtained by converting speech information, for example, by extracting text information from speech information. In some embodiments, a pre-trained speech recognition model or acoustic model can be used to recognize the content of speech information, and then the recognized content can be recorded in text form to obtain the text information corresponding to the communication information. For example, text information can be extracted using algorithms or models, such as LSTM, BERT, one-hot encoding, bag-of-words models, term frequency–inverse document frequency (TF-IDF) models, vocabulary models, etc.
[0054] Step 220: Process the communication information and extract key information.
[0055] The key information can include key speech, keywords, and key derived information. Specifically, key speech can be extracted from speech information, keywords can be extracted from text information, and key derived information can be extracted by combining speech and text information.
[0056] Optionally, step 220 may also include the following steps:
[0057] The communication information is input into the key information extraction model to obtain the intermediate classification results of the communication information;
[0058] Based on the intermediate classification results, extract the key information;
[0059] Wherein, if the key information is key speech, the classification criteria for the classification result include the speech emotion and speech rate corresponding to the communication information;
[0060] If the key information is a keyword, then the classification criteria for the classification result include the number of emotional and degree words corresponding to the communication information.
[0061] Voice emotion refers to the emotion expressed in voice information, such as positive, neutral, and negative emotions. Specifically, the classification function of the key information extraction model can be used to extract key voice words containing user emotions and provide classification results; that is, the key information extraction model can be used as an emotion classifier.
[0062] The voice emotions can include, but are not limited to, feelings of loss, calmness, enthusiasm, or passion, and can be customized as needed in real-world scenarios. For example, in some embodiments, voice emotions can also include joy, sadness, pain, relief, excitement, etc., without exhaustive list.
[0063] Specifically, the classification result can be the emotional probability of the voice information, where the emotion indicated by the voice emotion classification result is the emotion with the highest probability. For example, if the emotion classifier outputs the following emotion classification results: disappointment 2%, calmness 20%, enthusiasm 80%, passion 60%, then the emotion indicated by this emotion recognition result is: enthusiasm. In other embodiments, the classification result can also identify the final result by outputting an emotion label. For example, if the classification result is: disappointment 1, calmness 0, enthusiasm 0, then the emotion labeled 1, "disappointment," is identified as the final classification result of the voice emotion. It should be noted that the specific details of extracting text emotions can be found in the relevant instructions for extracting voice emotions, which will not be repeated in this embodiment.
[0064] Optionally, the acoustic model can be calculated using a trained acoustic model, which can be designed as a sub-model within the key information extraction model, or the key information extraction model can possess the functionality of the acoustic model. The user's speech rate during communication can be obtained by calculating the average duration of each sentence / segment in the user's speech information; for example, the shorter the average duration of each sentence / segment in the speech information, the faster the user's speech rate.
[0065] Degree words can be keywords in textual information that indicate degree, such as: very, extremely, quite, as much as possible, etc. It can be understood that when a user's communication contains a large number of degree words, it highly likely reflects the urgency and importance of their need. Therefore, the classification results can provide the number of degree words, such as 1, 5, 10, etc. It can be understood that the more degree words there are, the higher the urgency, importance, and need of the user's communication.
[0066] The key information extraction model itself or its sub-classification models may include, but are not limited to, neural networks (NN), convolutional neural networks (CNN), deep neural networks (DNN), recurrent neural networks (RNN), or any combination thereof.
[0067] The key derived information includes key speaker information of the communication information, which is determined based on one or more of the following: speaking frequency, speaking percentage, and number of interruptions to customer service voice in the key voice messages.
[0068] In some embodiments, a threshold for speaking frequency can be preset (e.g., 3 times per minute). When a user's speaking frequency reaches or exceeds this threshold, the user can be identified as a key speaker, and their communication information can be identified as key speaker information. In some embodiments, a threshold for speaking percentage can also be preset, such as 30%, 40%, etc. Speaking percentage refers to the proportion of a user's speaking time to the total speaking time of all users participating in the communication and the voice customer service representative. For example, if the user's total speaking time is 5 minutes, and the total speaking time of all users participating in the communication and the voice customer service representative is 10 minutes, then the user's speaking percentage is 50%. When a respondent's speaking time reaches or exceeds this threshold, the user can be identified as a key speaker, and their communication information can be identified as key speaker information. In some embodiments, a threshold for the number of times customer service voice can be preset (e.g., 5 interruptions within 10 minutes). Interrupting customer service voice refers to the user interrupting the intelligent customer service representative's voice output during the process of the intelligent customer service representative providing voice services. It is understandable that when a respondent's speech is interrupted a number of times that threshold is reached or exceeded, it indicates that the user has objections or misunderstandings about the content or direction of the voice service. Therefore, the voice customer service platform should focus on the user's communication information to identify the user as a key speaker, and their communication information can be identified as key speaker information.
[0069] In some embodiments, to ensure accurate extraction of key information from user communication, the key information extraction model can also be a voice recognition model capable of distinguishing different timbres (e.g., distinguishing the timbres of the user and the intelligent customer service). In some embodiments, it can be trained based on the GMM-SVM algorithm, where the Gaussian Mixture Model (GMM) is used to depict the feature space distribution of the speech speaker using a weighted mixture of multiple Gaussian distributions, and the Support Vector Machine (SVM) is used for classification to determine whether the speaking voice comes from the intelligent customer service or the user.
[0070] In some embodiments, the recognition accuracy can be tested by inputting the voice data of users and intelligent customer service as a test set into a trained key information extraction model.
[0071] The formula for the GMM model can be expressed as:
[0072]
[0073] In formula (1), p(x) is the feature space distribution value, x is the test sample, k is the sample size, and Π k For weighting coefficients, μ k ,∑ k Representing user voice and intelligent customer service voice respectively, 'a' is a hyperparameter, N(x|μ) k ,∑k ) represents the k-th component in the model.
[0074] The formula for the SVM model can be expressed as:
[0075]
[0076] Among them, in formula (2) Let represent the feature vector of sample x after mapping, w and b are the hyperplane parameters to be classified, st is the conditional constraint symbol, and T is the transpose symbol.
[0077] For the sample x mentioned above, the BERT model can be used to obtain the classification probability of each sample; the entropy value of each classification probability is calculated, and the smaller the entropy value, the better the stability of the information contained in the sample data. The entropy value of each classification probability is calculated using the following formula:
[0078]
[0079] In formula (3), H is the corresponding entropy value, n is the total number of samples, and x i For each sample classification, p(x) i ) is the classification probability for each sample.
[0080] Step 230: Analyze the key information to obtain the first level corresponding to the communication information.
[0081] Optionally, step 230 may also include:
[0082] Based on the key information and combined with the first preset conditions, the primary level of the communication information is determined; wherein, the primary level includes the urgency level and importance level of the communication information;
[0083] Based on the primary level of the communication information and the key information, and in conjunction with the second preset condition, the sub-level corresponding to the communication information under the primary level is determined; wherein, the level of the sub-level is positively correlated with the level of urgency and the level of importance.
[0084] The first preset condition can include various pre-set conditions related to the urgency and importance level of the communication information. For example, the first preset condition can include a negative emotional tone and a fast speaking speed. It can be understood that key information that meets this first preset condition reflects that the user's emotions are relatively negative and the speaking speed is relatively fast, which greatly increases the probability that the inquiry is about a relatively urgent and important matter. Therefore, the first level can correspond to a high urgency level and a high importance level.
[0085] Optionally, after obtaining the primary level of key information, the primary level can be further subdivided according to its sub-levels. Therefore, the second preset condition is similar to the first preset condition and can also include various preset conditions related to the urgency and importance of the communication information, but it is more detailed than the first preset condition. For example, the threshold information used to determine key speakers can be obtained as the second preset condition to subdivide the primary level into sub-levels. For instance, multiple threshold intervals can be set and combined with the proportion of user speech in the communication information to determine which sub-level the user's speech proportion falls into, thus identifying the corresponding sub-level. Alternatively, the primary level can be subdivided into sub-levels based on the intensity of positive and negative emotions corresponding to voice / text emotions. For example, a sub-level satisfying strong negative emotions is 1, and a sub-level satisfying general message emotions is 2, etc.
[0086] Of course, both the first and second levels can be represented by text, numbers, special symbols, etc. For example, the urgency level is 1, the sub-level is 1, the urgency level is ※, the sub-level is ※, etc. This embodiment does not impose any restrictions.
[0087] Optionally, the method in this application embodiment may further include:
[0088] If the sub-level of the communication information is at the highest, obtain the degree of deviation between the key information and the second preset condition;
[0089] If the degree of deviation meets the third preset condition, the primary level of the communication information will be adjusted.
[0090] The degree of deviation between the key information and the second preset condition can be represented by a deviation value. For example, if the second preset condition is that the number of degree words is greater than 5, but the user's corresponding key information contains 10 degree words, the deviation value is 5. However, the proportion of this deviation value to the threshold of the second preset condition has reached a 1:1 ratio, indicating that the deviation between the key information and the second preset condition is relatively large. The primary level of the key information can be adjusted, for example, by raising the primary level of the key information by one level.
[0091] Step 240: Based on the communication information of the first level, determine the voice service task for the communication information of the first level. The voice service task may include the type of voice service and the second level.
[0092] Optionally, step 240 may further include: determining the type of the voice service task based on the urgency level in the first level of the communication information. The type of voice task may include human voice service and intelligent voice service.
[0093] Optionally, the method in this application embodiment may further include:
[0094] Based on the importance level in the first level of the communication information, a second level matching the first level is determined, the second level including the artificial voice service level and the intelligent voice service level;
[0095] If the voice service task is the human voice service, then according to the human voice service level, match at least one of the following screening criteria among the positive review rate, rating and years of service of the corresponding human voice service personnel, and determine the human voice service personnel who meet the screening criteria as the executors of the human voice service.
[0096] If the voice service task is the intelligent voice service, then the corresponding intelligent voice library is matched according to the intelligent voice service level, so as to provide voice services to the user based on the voice content of the intelligent voice library.
[0097] It is understandable that after determining the first level of the communication information, the urgency and importance of the communication information can be obtained. The higher the urgency and importance level, the more important the user's communication information is. For communication information with a higher first level, the user can be transferred to human voice service or a higher-level intelligent voice customer service to improve the user experience and the efficiency of providing voice services to the user.
[0098] Optionally, the method in this application embodiment further includes:
[0099] The communication outline is updated based on the voice service task for the first level of communication information.
[0100] The communication outline is used to formulate communication content between voice service customer service representatives and users. For example, when the communication topic or main content is determined to be product consultation based on the user's communication information, the interview outline can be updated with relevant interview content such as product parameters and core product advantages. In some alternative embodiments, the interview outline can also be updated based on content manually entered by staff, with the staff manually judging the tendency of the communication information to update the interview content.
[0101] As can be seen from the above, after obtaining the user's communication information, this application embodiment can extract key information and analyze the key information to classify and identify the needs of the communication information in a detailed and accurate manner. Based on the urgency and importance of the user's communication information, it can efficiently and accurately determine the type and level of voice service for the user, so as to provide the user with efficient and accurate voice service.
[0102] To implement the above-described method embodiments, this application also provides a voice service platform based on automatic speech classification. Figure 3 This illustration shows a structural diagram of a voice service platform based on automatic speech classification provided in an embodiment of this application. The system includes:
[0103] The communication information acquisition module 301 is used to acquire the user's communication information; wherein, the communication information includes voice information and text information during the user's communication process;
[0104] The key information extraction module 302 is used to process the communication information and extract key information; wherein, the key information includes key voice, keywords and key derived information;
[0105] The key information analysis module 303 is used to analyze the key information to obtain the first level corresponding to the communication information;
[0106] The task creation module 304 is used to determine a voice service task for the first level of communication information based on the first level of communication information. The voice service task includes the type of voice service and the second level.
[0107] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the modules / units / subunits / components in the above-described device can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0108] In some embodiments, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 4 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores user communication information and relevant data from the intelligent voice customer service platform. The network interface communicates with external terminals via a network. When the computer program is executed by the processor, it implements a voice-based automatic classification voice service method and a voice customer service platform.
[0109] In some embodiments, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 4As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a voice-based automatic hierarchical voice service method and a voice customer service platform. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.
[0110] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0111] In some embodiments, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.
[0112] In some embodiments, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above method embodiments.
[0113] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.
[0114] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0115] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.
[0116] In summary, the speech service method based on automatic speech classification provided in this application is characterized by comprising:
[0117] Obtain user communication information; wherein, the communication information includes voice information and text information during the user's communication process;
[0118] The communication information is processed to extract key information; wherein, the key information includes key speech, keywords, and key derived information;
[0119] The key information is analyzed to obtain the first level corresponding to the communication information;
[0120] Based on the communication information of the first level, a voice service task is determined for the communication information of the first level, wherein the voice service task includes the type of voice service and the second level.
[0121] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.
[0122] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0123] In addition, the functional units in the embodiments provided in this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0124] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0125] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0126] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application. All should be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.
Claims
1. A voice service method based on automatic speech classification, characterized in that, include: Obtain user communication information; wherein, the communication information includes voice information and text information during the user's communication process; The communication information is processed to extract key information, including: inputting the communication information into a key information extraction model to obtain intermediate classification results of the communication information; extracting the key information based on the intermediate classification results; wherein the key information includes key speech, keywords, and key derived information; obtaining the classification probability of each sample used to train the key information extraction model through a BERT model, and calculating the entropy value of each classification probability to measure the stability of the information included in each sample; The key information is analyzed to obtain the first level corresponding to the communication information; Based on the communication information of the first level, a voice service task is determined for the communication information of the first level, wherein the voice service task includes the type of voice service and the second level. The analysis of the key information to obtain the first level corresponding to the communication information includes: Based on the key information and combined with the first preset conditions, the primary level of the communication information is determined; wherein, the primary level includes the urgency level and importance level of the communication information; the first preset conditions include pre-set conditions related to the urgency and importance level of the communication information. If the key information is key speech, then the classification criteria for the classification result include the speech emotion and speech rate corresponding to the communication information; If the key information is a keyword, then the classification criteria for the classification result include the number of emotional and degree words corresponding to the communication information; The key derived information includes key speaker information of the communication information, which is determined based on one or more of the following: speaking frequency, speaking percentage, and number of interruptions to customer service voice in the key voice messages; Based on the primary level of the communication information and the key information, and in conjunction with the second preset conditions, the sub-level corresponding to the communication information under the primary level is determined; wherein, the level of the sub-level is positively correlated with the level of urgency and the level of importance. If the sub-level of the communication information is at the highest, obtain the degree of deviation between the key information and the second preset condition; If the degree of deviation meets the third preset condition, the primary level of the communication information will be adjusted. Meanwhile, the key information extraction model is trained based on the GMM-SVM algorithm. The Gaussian Mixture Model (GMM) is used to depict the feature space distribution of the speaker by using a weighted mixture of multiple Gaussian distributions, and the Support Vector Machine (SVM) is used for classification and discrimination to determine whether the voice comes from the intelligent customer service or the user.
2. The method according to claim 1, characterized in that, The step of determining the voice service task for the first level of communication information based on the first level of communication information includes: The type of the voice service task is determined based on the urgency level in the first level of the communication information, and the type includes human voice service and intelligent voice service.
3. The method according to claim 2, characterized in that, The method further includes: Based on the importance level in the first level of the communication information, a second level matching the first level is determined, the second level including the artificial voice service level and the intelligent voice service level; If the voice service task is the human voice service, then according to the human voice service level, match at least one of the following screening criteria among the positive review rate, rating and years of service of the corresponding human voice service personnel, and determine the human voice service personnel who meet the screening criteria as the executors of the human voice service. If the voice service task is the intelligent voice service, then the corresponding intelligent voice library is matched according to the intelligent voice service level, so as to provide voice services to the user based on the voice content of the intelligent voice library.
4. The method according to claim 1, characterized in that, The method further includes: Based on the voice service task for the first level of communication information, the communication outline is updated; wherein, the communication outline is used to formulate the communication content between the voice service customer service representative and the user.
5. A voice service platform based on automatic speech classification, characterized in that, The platform includes: The communication information acquisition module is used to acquire the user's communication information; wherein, the communication information includes voice information and text information during the user's communication process; A key information extraction module is used to process the communication information, extract key information, and input the communication information into a key information extraction model to obtain intermediate classification results of the communication information; extract the key information based on the intermediate classification results; wherein, the key information includes key speech, keywords, and key derived information; obtain the classification probability of each sample used to train the key information extraction model through a BERT model, and calculate the entropy value of each classification probability to measure the stability of the information included in each sample; The key information analysis module is used to analyze the key information to obtain the first level corresponding to the communication information; The task creation module is used to determine a voice service task for the first level of communication information based on the first level of communication information. The voice service task includes the type of voice service and the second level. The analysis of the key information to obtain the first level corresponding to the communication information includes: Based on the key information and combined with the first preset conditions, the primary level of the communication information is determined; wherein, the primary level includes the urgency level and importance level of the communication information; the first preset conditions include pre-set conditions related to the urgency and importance level of the communication information; if the key information is key speech, the classification criteria of the classification result include the speech emotion and speech rate corresponding to the communication information. If the key information is a keyword, then the classification criteria for the classification result include the number of emotional and degree words corresponding to the communication information; The key derived information includes key speaker information of the communication information, which is determined based on one or more of the following: speaking frequency, speaking percentage, and number of interruptions to customer service voice in the key voice messages; Based on the primary level of the communication information and the key information, and in conjunction with the second preset conditions, the sub-level corresponding to the communication information under the primary level is determined; wherein, the level of the sub-level is positively correlated with the level of urgency and the level of importance. If the sub-level of the communication information is at the highest, obtain the degree of deviation between the key information and the second preset condition; If the degree of deviation meets the third preset condition, the primary level of the communication information will be adjusted. Meanwhile, the key information extraction model is trained based on the GMM-SVM algorithm. The Gaussian Mixture Model (GMM) is used to depict the feature space distribution of the speaker by using a weighted mixture of multiple Gaussian distributions, and the Support Vector Machine (SVM) is used for classification and discrimination to determine whether the voice comes from the intelligent customer service or the user.
6. A computer-readable storage medium, characterized in that, The storage medium stores computer instructions. When the computer reads the computer instructions in the storage medium, the computer executes the method as described in any one of claims 1-4.