A voice quality inspection method, device, computer device, and storage medium

Abstract

This application relates to a voice quality inspection method, device, computer device, and storage medium. The voice quality inspection method includes: converting voice data into text data, performing sentence segmentation processing on the text data to obtain text segments; configuring keyword types according to task types, retrieving keywords corresponding to the keyword types, and updating the keyword library; selecting keyword texts from the keyword library, comparing them with the text segments to obtain matching information of the keyword texts; comparing the matching information of the keyword texts with the number of selected keyword texts to obtain a matching coefficient of the keyword type; comparing the matching coefficient of the keyword type with a preset matching threshold to obtain a quality inspection result of the task type, which can avoid the defects of manual quality inspection methods, reduce enterprise costs, improve voice quality inspection efficiency, and realize the systematization of customer service performance evaluation and customer service satisfaction.

Description

Technical Field

[0001] The present invention relates to the technical field of information processing, and particularly to a voice quality inspection method, apparatus, computer device, and storage medium. Background Art

[0002] With the development of information technology, customers usually communicate with enterprise customer service via voice for operations such as business consultation and complaint. To improve voice service quality and customer satisfaction, enterprises need to evaluate and assess the performance of voice customer service. Currently, mainly through manual means, the voice communication records generated between customer service and customers within a certain period are sorted out and quality inspected. However, with the increase in business volume, the manual method has defects such as cumbersome operation steps, low execution efficiency, poor quality inspection, high labor costs, and excessive subjectivity, resulting in problems such as low accuracy of voice quality inspection results and unreasonable performance assessment results of voice customer service. Summary of the Invention

[0003] Based on this, it is necessary to provide a voice quality inspection method, apparatus, computer device, and storage medium for improving the poor effect of manual voice quality inspection in view of the above technical problems.

[0004] On the one hand, a voice quality inspection method is provided, and the voice quality inspection method includes:

[0005] Convert voice data into text data, perform sentence segmentation on the text data, and obtain text segments;

[0006] Configure keyword types according to the task type, retrieve keywords corresponding to the keyword types, and update the keyword library;

[0007] Select keyword texts from the keyword library, compare them with the text segments, and obtain matching information of the keyword texts;

[0008] Compare the matching information of the keyword texts with the number of selected keyword texts to obtain a matching coefficient of the keyword type;

[0009] Compare the matching coefficient of the keyword type with a preset matching threshold to obtain a quality inspection result of the task type.

[0010] In one embodiment, the step of configuring keyword types according to the task type, retrieving keywords corresponding to the keyword types, and updating the keyword library includes:

[0011] Configure keyword types according to the task type, where the keyword types include: standard keywords, prohibited word keywords, and sentiment keywords;

[0012] Retrieve the meanings and application scenarios of each keyword type to obtain the corresponding keywords;

[0013] Update the keyword library according to the keywords.

[0014] In one embodiment, the steps of selecting keyword texts from the keyword library, comparing them with the text fragment, and obtaining the matching information of the keyword texts include:

[0015] According to the keyword library, select at least one keyword text for each keyword type to form a single configuration for each keyword type;

[0016] Perform content matching and traversal on the single configuration starting from the first clause of the text fragment. If there is a matching relationship between the clause and the keyword text in the single configuration, record the serial number of the clause in the text fragment to obtain the matching position of the keyword text;

[0017] Record the number of keyword texts in the matching relationship between the clause and the single configuration. If the same keyword text in the single configuration appears multiple times in the same clause, only count the number of times of that keyword text once. According to the number of keyword texts, obtain the matching times of the keyword text;

[0018] Obtain the matching information of the keyword text according to the matching position and the matching times of the keyword text.

[0019] In one embodiment, the steps of comparing the matching information of the keyword text with the number of selected keyword texts to obtain the matching coefficient of the keyword type include:

[0020] Sort the matching times according to the matching information of the keyword text, and obtain the maximum matching times according to the largest matching times in the sorting;

[0021] Divide the maximum matching times by the number of selected keyword texts to obtain the matching coefficient of the single configuration;

[0022] Set a sampling weight for each single configuration, and obtain the sampling coefficient of the single configuration through the sampling weight and the matching coefficient of the single configuration. The mathematical expression of the sampling coefficient sp of the single configuration is:

[0023] sp = w * p

[0024] Where sp is the sampling coefficient of the single configuration, w is the sampling weight of the single configuration, and p is the matching coefficient of the single configuration;

[0025] Compare the single - configuration sampling coefficient with a preset single - configuration matching threshold to obtain a matching coefficient S for the keyword type. The mathematical expression of the keyword - type matching coefficient S is as follows:

[0026]

[0027] Where S is the keyword - type matching coefficient, sp is the single - configuration sampling coefficient, t is the single - configuration matching threshold, max(·) represents taking the maximum value, and d(·) is the differential operator.

[0028] In one embodiment, the step of comparing the matching coefficient of the keyword type with a preset matching threshold to obtain the quality - inspection result of the task type includes:

[0029] Compare the matching coefficients of each keyword type with the preset matching thresholds respectively to obtain the quality - inspection result.

[0030] In one embodiment, the step of comparing the matching coefficient of the keyword type with a preset matching threshold to obtain the quality - inspection result of the task type further includes:

[0031] Set sampling weights for each keyword type, and obtain sampling coefficients according to the sampling weights and the matching coefficients of the keyword types;

[0032] Compare the sampling coefficients with a preset matching threshold to obtain the quality - inspection result.

[0033] In one embodiment, the steps of converting voice data into text data and performing sentence - segmentation processing on the text data to obtain text segments include:

[0034] Separate the silent content and voice content in the voice data, and obtain a time - separation label;

[0035] Perform sentence - segmentation on the text data according to the time - separation label, and add punctuation marks at the end of the sentence according to the color - related vocabulary at the end of the sentence;

[0036] Query the number of characters of the text without punctuation marks in the text data. When the text without punctuation marks exceeds a preset character - number threshold, add punctuation marks.

[0037] On the other hand, a voice quality - inspection device is provided. The voice quality - inspection device includes:

[0038] A voice - to - text conversion module, configured to convert voice data into text data, perform sentence - segmentation processing on the text data, and obtain text segments;

[0039] A task parameter configuration module, which is used to configure keyword types according to task types, retrieve keywords corresponding to the keyword types, and update the keyword library;

[0040] A matching information acquisition module, which is used to select keyword texts from the keyword library, compare them with the text fragments, and obtain the matching information of the keyword texts;

[0041] A matching coefficient acquisition module, which is used to compare the matching information of the keyword texts with the number of selected keyword texts to obtain the matching coefficient of the keyword type;

[0042] A quality inspection result acquisition module, which is used to compare the matching coefficient of the keyword type with a preset matching threshold to obtain the quality inspection result of the task type.

[0043] On the other hand, a voice quality inspection device is provided. The voice quality inspection device includes a quality inspection result acquisition module, and the quality inspection result acquisition module includes:

[0044] A first acquisition unit, which is used to compare the matching coefficients of each keyword type with the preset matching thresholds respectively to obtain the quality inspection result.

[0045] On the other hand, a voice quality inspection device is provided. The voice quality inspection device includes a quality inspection result acquisition module, and the quality inspection result acquisition module further includes:

[0046] A first acquisition unit, which is used to set sampling weights for each keyword type, and obtain sampling coefficients according to the sampling weights and the matching coefficients of the keyword types;

[0047] A second acquisition unit, which is used to compare the sampling coefficients with a preset matching threshold to obtain the quality inspection result.

[0048] On yet another aspect, a computer device is provided, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented:

[0049] Convert voice data into text data, perform sentence segmentation processing on the text data to obtain text fragments;

[0050] Configure keyword types according to task types, retrieve keywords corresponding to the keyword types, and update the keyword library;

[0051] Select keyword texts from the keyword library, compare them with the text fragments, and obtain the matching information of the keyword texts;

[0052] Based on the matching information of the keyword text, compare it with the number of selected keyword texts to obtain the matching coefficient of the keyword type;

[0053] Based on the matching coefficient of the keyword type, compare it with a preset matching threshold to obtain the quality inspection result of the task type.

[0054] On the other hand, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0055] Convert the voice data into text data, perform sentence segmentation processing on the text data to obtain text fragments;

[0056] Configure the keyword type according to the task type, retrieve the keywords corresponding to the keyword type, and update the keyword library;

[0057] Select keyword texts from the keyword library, compare them with the text fragments to obtain the matching information of the keyword texts;

[0058] Based on the matching information of the keyword text, compare it with the number of selected keyword texts to obtain the matching coefficient of the keyword type;

[0059] Based on the matching coefficient of the keyword type, compare it with a preset matching threshold to obtain the quality inspection result of the task type.

[0060] The above-mentioned voice quality inspection method, device, computer device and storage medium set the keyword type according to the task type, select keyword texts to compare with the customer service voice text, obtain the matching information and matching coefficient, and judge the customer service voice quality inspection result, which can avoid the defects of the manual quality inspection method, reduce the enterprise cost, improve the voice quality inspection efficiency, and realize the systematization of customer service performance appraisal and customer service satisfaction. Description of the Drawings

[0061] Figure 1 It is an application flow chart of a voice quality inspection method in an embodiment;

[0062] Figure 2 It is an application environment diagram of a voice quality inspection method in an embodiment;

[0063] Figure 3 It is a flow schematic diagram of a voice quality inspection method in an embodiment;

[0064] Figure 4 It is a flow schematic diagram of the step of obtaining text fragments in an embodiment;

[0065] Figure 5Schematic flowchart of the steps for updating the keyword library in an embodiment;

[0066] Figure 6 Schematic flowchart of the steps for obtaining the matching position in an embodiment;

[0067] Figure 7 Schematic flowchart of the steps for obtaining the matching coefficient in an embodiment;

[0068] Figure 8 Schematic flowchart of the steps for obtaining the quality inspection result in an embodiment;

[0069] Figure 9 Schematic flowchart of the steps for obtaining the quality inspection result in another embodiment;

[0070] Figure 10 Structural block diagram of the quality inspection result acquisition module in an embodiment;

[0071] Figure 11 Structural block diagram of the voice quality inspection device in an embodiment;

[0072] Figure 12 Internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0073] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0074] A voice quality inspection method provided by the present application has an application process as Figure 1 shown. For example, the voice quality inspection method provided by the present application can be applied to the detection of the quality of customer service voice services. By converting the voice data 100 into text data 101 and obtaining the quality inspection result 103 through the matching step 102 for the text data 101, the defects existing in the manual quality inspection method can be avoided, the enterprise cost can be reduced, the voice quality inspection efficiency can be improved, and the systematic evaluation of customer service performance and customer service satisfaction can be realized.

[0075] A voice quality inspection method provided by the present application can be applied to such as Figure 2In the application environment shown. Among them, the terminal 200 communicates with the server 201 through the network. For example, the voice quality inspection method provided in this application converts voice data into text data and performs sentence segmentation processing, configures keyword types according to the task type, updates the keyword library, selects keyword texts from the keyword library, compares them with text segments, then obtains the matching information of the keyword texts, compares according to the matching information and the number of selected keyword texts, obtains the matching coefficient of the keyword type, and according to the matching coefficient, compares with a preset matching threshold to obtain the quality inspection result of the task type. Among them, the terminal 200 can be but is not limited to various personal computers, laptop computers, smart phones, tablet computers, portable wearable devices or sub-servers, and the server 201 can be implemented by an independent server or a server cluster or cloud computing platform composed of multiple servers.

[0076] In one embodiment, as Figure 3 shown, a voice quality inspection method is provided, including the following steps:

[0077] S1: Convert voice data into text data, perform sentence segmentation processing on the text data, and obtain text segments;

[0078] S2: Configure keyword types according to the task type, and retrieve the keywords corresponding to the keyword types to update the keyword library;

[0079] S3: Select keyword texts from the keyword library, compare them with the text segments, and obtain the matching information of the keyword texts;

[0080] S4: According to the matching information of the keyword texts, compare with the number of selected keyword texts to obtain the matching coefficient of the keyword type;

[0081] S5: According to the matching coefficient of the keyword type, compare with a preset matching threshold to obtain the quality inspection result of the task type.

[0082] Through the above steps, it is possible to improve the problems of cumbersome operation steps, low execution efficiency, poor quality inspection, high labor cost, and excessive subjectivity when manually sorting and inspecting the voice communication records generated between the customer service and the customer. Convert the voice data into text data, compare the text data with the keyword texts in the keyword library, set the matching rules, and obtain the voice quality inspection result, which can avoid the defects of the manual quality inspection method, reduce the enterprise cost, improve the voice quality inspection efficiency, and realize the systematization of customer service performance evaluation and customer service satisfaction.

[0083] Since the voice data includes two-way voice records generated by the customer and the enterprise customer service through voice communication within a period of time, and when detecting the quality of the customer service voice service, it is necessary to distinguish the different voice contents of the customer and the customer service, and there may be noise in the voice content, so it is impossible to detect the text data directly converted from the voice data into a single text. In step S1, by way of example, the voice data can be converted into text data, the text data can be segmented to obtain text fragments. For example, after obtaining the two-way voice records generated by the customer and the enterprise customer service through voice communication within a period of time, the voice data is preprocessed such as removing noise and interference, and the voice activation detection and other training methods are used to separate the silent content and the voice content in the voice data to obtain time separation tags. The preprocessed voice data is recognized through voice recognition algorithms such as the Viterbi algorithm to obtain text data. The automatic speech recognition technology such as the automatic rapid recognition technology is used to separate the roles of the text data to obtain customer text data and customer service text data. The customer text data and the customer service text data are segmented according to the time separation tags and punctuation marks are added. The customer text data and the customer service text data are segmented according to the color vocabulary of the customer text data and the customer service text data and punctuation marks are added. The number of words of the text of the customer text data and the customer service text data without punctuation marks is queried. When the text without punctuation marks exceeds the preset word count threshold, punctuation marks are added to obtain text fragments.

[0084] When detecting the quality of the voice customer service, in order to select newer or more appropriate keywords, it is necessary to update the keyword library. In step S2, by way of example, one or more task types can be set, and then the keyword types are configured according to the task types, and the common words corresponding to the meanings and application scenarios of the keyword types are retrieved, and then the words are combined to update the keyword library. For example, according to different task types, such as business consultation, marketing outbound call, complaint and suggestion, etc., different keyword types are configured, the common words corresponding to the meanings, application scenarios and other contents of different keyword types are retrieved, and the common words are combined to update the keyword library. New keywords can also be added to the keyword library, and inappropriate, infrequently used or misclassified keywords in the keyword library can be deleted to update the keyword library.

[0085] In order to detect whether the content of the voice communication between the customer service and the customer meets the requirements, it is necessary to perform keyword matching on the text data. In step S3, by way of example, keyword texts can be selected from the keyword library and compared with the text segment to obtain the matching information of the keyword texts. For example, according to different task types and different keyword types, one or more keyword texts are selected from the keyword library, and content matching and traversal of the keyword texts are performed starting from the first clause of the text segment. If there is a matching relationship between the clause and the keyword text, record the serial number of the clause in the text segment, obtain the matching position, and record the matching times of the keyword text.

[0086] After obtaining the matching information of the keyword texts, it is necessary to count the number of matches between each clause in the text segment and a single keyword text. In step S4, by way of example, according to the matching information, the maximum number of keyword texts with a matching relationship in each clause is divided by the number of selected keyword texts to obtain the matching coefficient of the keyword texts.

[0087] After obtaining the matching coefficient of the keyword texts, in order to determine whether the voice quality inspection is qualified, in step S5, by way of example, according to the matching coefficient, it is compared with a preset matching threshold to obtain the quality inspection result of the task type. For example, according to the task type, sampling weights are set for each keyword type, and the sampling weights are used to represent the importance of each keyword type in different task types. For example, for the task type of business consultation, an enterprise can increase the weight of the standard keywords in the keyword type, while for the task type of complaint and suggestion, an enterprise can increase the weights of the prohibited word keywords and emotional keywords in the keyword type. Then, according to the matching coefficient and the sampling weights, it is compared with the preset matching threshold to obtain the quality inspection result.

[0088] Before performing quality inspection on the voice communication content between the customer and the customer service, it is necessary to convert the voice data into text data. As Figure 4 shown, the steps for converting the voice data into text data and performing sentence segmentation processing on the text data to obtain text segments include:

[0089] S11: Separate the silent content and the voice content in the voice data and obtain a time separation label;

[0090] S12: Perform sentence segmentation on the text data according to the time separation label, and add punctuation marks at the end of the sentence according to the color words at the end of the sentence;

[0091] S13: Query the number of characters of the text without punctuation marks in the text data. When the text without punctuation marks exceeds a preset character threshold, add punctuation marks.

[0092] Through the above steps, the voice communication content between the customer service and the customer within a period of time can be segmented into silent content and voice content, while distinguishing the different voice content of the customer and the customer service, eliminating the possible noise in the voice content, and adding punctuation marks to the converted text data for sentence segmentation, so as to improve the applicability of the text data.

[0093] As Figure 4 shown, in step S11, by way of example, to illustrate the method of separating the silent content and the voice content in the voice data, the voice data can be detected by using a voice activity detection training method. When the detected silent duration exceeds the silent threshold, such as 3 seconds, record this time point, obtain a time separation label, and use the time separation label to separate the silent content and the voice content. For example, at 5 seconds, record this time point, obtain a time separation label, and use the time separation label to separate the silent content and the voice content. Different silent thresholds can be set according to different speech rates to make the separation of the silent content and the voice content in the voice data more reasonable, and at the same time take into account the respective pause habits of different customers and customer services during voice communication, so that the separated voice content is more suitable for the subsequent voice quality inspection steps.

[0094] As Figure 4 shown, in step S12, by way of example, when an exclamation emotion-colored word, such as: "ah", "ba", "ma", etc. is retrieved at the time separation label, add an exclamation mark at the end of the sentence. When a question emotion-colored word, such as: "ma", "what", etc. is retrieved at the time separation label, add a question mark at the end of the sentence. By this means, appropriate punctuation marks can be added to the text according to the emotion color of the customer or the customer service, which is convenient for the subsequent implementation of the voice quality inspection step.

[0095] As Figure 4 shown, in step S13, by way of example, query the number of characters of the text without punctuation marks in the text data. For each sentence segment in the text data, if its number of characters exceeds the preset character threshold, such as 20 characters, add a full stop for sentence segmentation, otherwise add a comma for sentence segmentation. The last sentence segment in the text data is directly added with a full stop for sentence segmentation. For example, when it is 30 characters, add a full stop for sentence segmentation, otherwise add a comma for sentence segmentation. The last sentence segment in the text data is directly added with a full stop for sentence segmentation.

[0096] Before selecting the keyword text from the keyword library, the keyword library needs to be updated to facilitate the selection of appropriate keyword text. As Figure 5 shown, in some embodiments, the voice quality inspection method further includes:

[0097] S21: Configure keyword types according to the task type, where the keyword types include: standard keywords, prohibited words, and emotion keywords;

[0098] S22: Retrieve the meaning and application scenario of each keyword type and obtain the corresponding keyword;

[0099] S23: Update the keyword library according to the keyword.

[0100] Through the above steps, different task types are analyzed, and different keyword types can be configured according to the task types. After further analyzing the meaning of each keyword type, the corresponding common words can be queried in combination with their application scenarios, and the common words can be combined to form a keyword library. The keyword library is then updated to facilitate the subsequent selection of appropriate keyword texts from the keywords and improve the applicability of the selected keyword texts.

[0101] like Figure 5 As shown, in step S21, the keyword types configured according to the task type include: standard keywords, banned keywords, and emotional keywords. For each keyword type, a single or multiple single configurations can be set, and one or more keyword texts are set in each single configuration to meet the different considerations and quality inspection requirements of the enterprise when dealing with different task types.

[0102] like Figure 5 As shown, in step S22, it is exemplarily explained that after configuring each keyword type, the meaning and application scenario of the keyword type are parsed, and the corresponding keywords are retrieved from the Internet. For example, for the task type of marketing outbound calls, its standard keywords can retrieve the following keywords: "Hello", "Sir", "Ms.", "Product", "Price", etc., so as to meet the diversity and richness of the available keywords under different task types.

[0103] like Figure 5 As shown, in step S23, it is exemplified that after the keywords are retrieved, they are manually reviewed and the keyword library is added with new keywords and deleted with new keywords. For example, when the keywords "product" and "price" are retrieved, they are added to the keyword library after being manually reviewed and queried. If banned words such as illegal, irregular, negative, and neglectful are found, they are deleted from the keyword library after being manually reviewed and queried, so as to ensure the real-time nature of the keyword library and provide protection for the selection of subsequent keyword texts.

[0104] After updating the keyword library, you need to select specific keyword texts to form each single configuration of each keyword type, and compare each single configuration with the text fragment, such as Figure 6 As shown, in some implementations, a method for obtaining keyword text matching information is provided:

[0105] S31: According to the keyword library, for each keyword type, select at least one keyword text respectively to form a single configuration for each keyword type.

[0106] S32: Start content matching and traversal from the first clause of the text segment for the single configuration. If there is a matching relationship between the clause and the keyword text in the single configuration, record the serial number of the clause in the text segment and obtain the matching position of the keyword text.

[0107] S33: Record the number of keyword texts in the clause that form a matching relationship with the single configuration. If the same keyword text in the single configuration appears multiple times in the same clause, only count this keyword text once. According to the number of keyword texts, obtain the matching times of the keyword text.

[0108] S34: Obtain the matching information of the keyword text based on the matching position and the matching times of the keyword text.

[0109] Through the above steps, the matching rules of keyword texts in the voice quality inspection process can be set in detail and accurately, the matching positions and matching times of each single configuration and the text segment can be obtained and recorded, the convenience of querying the matching text information of keyword texts can be improved, which helps to analyze the matching situation of each keyword text in the voice quality inspection process and improve the accuracy of voice quality inspection at the same time.

[0110] As Figure 6 shown, in step S31, it is exemplarily illustrated that for each keyword type, at least one keyword text is selected respectively to form a single configuration for each keyword type. For example, when the task type is marketing outbound call, three keyword types are configured, namely standard keywords, prohibited words, and sentiment keywords. For standard keywords, three single configurations are set. For prohibited words, two single configurations are set. For sentiment keywords, two single configurations are set. Among them, the keywords selected by standard keyword single configuration 1 are: "Hello", "Sir", "Madam". The keywords selected by standard keyword single configuration 2 are: "Function", "Price". The keyword selected by standard keyword single configuration 3 is: "Answer". The keyword selected by prohibited word single configuration 1 is: "Fraud". The keyword selected by prohibited word single configuration 2 is: "Unclear". The keywords selected by sentiment keyword single configuration 1 are: "Thank you", "Angry". In this way, the enterprise's requirement of setting different single configurations for different keyword types and the selected keyword texts being diverse and rich can be met.

[0111] As Figure 6As shown, in step S32, by way of example, the keyword text in the single configuration is regarded as a whole, starting from the first clause of the text segment, content matching and traversing are performed on the single configuration. If there is a matching relationship between the clause and the keyword text in the single configuration, record the serial number of the clause in the text segment, and obtain the matching position of the keyword text. For example, when the nth clause in the text segment is "You're talking nonsense. I don't care what you think. It has nothing to do with me", and the keyword text selected for a single configuration of the prohibited word keyword is: "casually", "garbage", then it is considered that the nth clause has a matching relationship with the keyword text in a single configuration of the prohibited word keyword, and the serial number of the clause in the text segment is n. Therefore, the matching position of the keyword text of the single configuration is n, where n is a non-negative integer.

[0112] As Figure 6 shown, in step S33, by way of example, when there is a matching relationship between the clause and the keyword text in the single configuration, record the number of keyword texts in the matching relationship between the clause and the single configuration. However, if the same keyword text within the single configuration appears multiple times in the clause, the number of this keyword text is only counted once. For example, the single configuration consists of N keyword texts, and among the N keywords, M keyword texts all have a matching relationship with the clause content, and among the M keyword texts, Q keyword texts appear K times in the clause content. At this time, although the P keyword texts appear K times in the clause content, the Q keyword texts are only counted once. Then it is considered that there are only M keyword texts in the matching relationship between the clause and the single configuration. Therefore, the number of times the keyword text of the single configuration is matched is M, where N, M, Q, K are non-negative integers, and N≥M≥Q.

[0113] As Figure 6As shown, in step S34, by way of example, after obtaining the matching position and the number of matches of a single configuration, the matching position and the number of matches are combined. For example, if a clause has a matching relationship with a single configuration, the serial number of the clause in the text segment is n, and when the single configuration matches the clause, the number of matches is M, then for the clause, the matching information of the single configuration can be expressed as [n, M]. For another clause, which also has a matching relationship with the single configuration, the serial number of the other clause in the text segment is m, and when the single configuration matches the other clause, the number of matches is Q, then for the other clause, the matching information of the single configuration can be expressed as [m, Q]. Additionally, when a single configuration has a matching relationship with X clauses, the matching information of the single configuration is expressed as a matrix of X rows and Y columns, that is, in the form of X*Y, where X represents the number of clauses having a matching relationship with the single configuration, and Y includes the matching positions and the number of matches obtained when the single configuration matches each of the X clauses. Here, n, m, M, Q, X, and Y are non-negative integers.

[0114] After obtaining the matching information of a single configuration, since a single configuration may have a matching relationship with multiple clauses, there are multiple sets of matching positions and the number of matches. Before obtaining the quality inspection result, it is necessary to process the matching information and obtain more matching information, such as Figure 7 As shown, in some implementation processes, a method for obtaining is provided:

[0115] S41: According to the matching information of the keyword text, sort the number of matches, and obtain the maximum number of matches according to the numerically largest number of matches in the sorting;

[0116] S42: Divide the maximum number of matches by the number of selected keyword texts to obtain the matching coefficient of a single configuration;

[0117] S43: Set a sampling weight for each single configuration, and obtain the sampling coefficient of a single configuration through the sampling weight and the matching coefficient of the single configuration. The mathematical expression of the sampling coefficient sp of a single configuration is:

[0118] sp = w * p

[0119] where sp is the sampling coefficient of the single configuration, w is the sampling weight of the single configuration, and p is the matching coefficient of the single configuration;

[0120] S44: Compare the sampling coefficient of the single configuration with a preset matching threshold of the single configuration to obtain the matching coefficient of the keyword type. The mathematical expression of the matching coefficient S of the keyword type is:

[0121]

[0122] Wherein, S is the keyword type matching coefficient, sp is the single configuration sampling coefficient, t is the single configuration matching threshold, max(·) represents taking the maximum value, and d(·) is the differential operator.

[0123] Through the above steps, the specific information of the matching positions and the number of matches of each single configuration when there is a matching relationship with different clauses in the text segment can be analyzed. At the same time, a suitable number of matches is selected for calculation in the subsequent quality inspection result acquisition step, improving the accuracy of the voice quality inspection result.

[0124] Such as Figure 7 As shown, in step S41, by way of example, the multiple sets of match counts in the match information are sorted in ascending or descending order, and the maximum match count among the multiple sets of match counts is obtained. For example, when the single configuration information expression is {[m,M],[n,N],[q,Q]}, where M > N > Q, then the maximum match count is considered to be M, where n, m, q, N, M, Q are non-negative integers.

[0125] In step S42, by way of example, after obtaining the maximum match count, the maximum match count is divided by the number of keyword texts selected for the current single configuration to obtain the matching coefficient. For example, when the number of keyword texts selected for the single configuration is N and the maximum match count is M, then the value of the matching coefficient p for the single configuration at this time is p = M / N, where N, M are non-negative integers.

[0126] In some embodiments, for step S42, when the keyword type is a prohibited word keyword, after obtaining the maximum match count, the number of keyword texts selected for the current single configuration minus the maximum match count is then divided by the number of keyword texts selected for the current single configuration to obtain the matching coefficient. For example, when the number of keyword texts selected for the single configuration is N and the maximum match count is M, then the value of the matching coefficient for the single configuration at this time is p = (N - M) / N, where N, M are non-negative integers.

[0127] In step S43, by way of example, a sampling weight is set for each single configuration corresponding to the keyword type, and the sampling weights are multiplied by the matching coefficients of the respective single configurations and then accumulated to obtain the matching coefficient of the keyword type. For example, the keyword type has i single configurations, and the matching coefficients of the i single configurations are expressed as {p1, p2,..., pi}, and the sampling weights set for the i single configurations are expressed as {w1, w2,..., wi}, then the calculation formula for the single configuration sampling coefficient sp is:

[0128] In step S44, a single configuration threshold is set to t. If sp is greater than t, it is considered that a keyword type formed by each single configuration is qualified. Then, the calculation formula for the matching coefficient S of the keyword type is: where k and i are positive integers, max(·) represents taking the maximum value, and d(·) is the differential operator.

[0129] In some other implementation processes, for steps S43 and S44, the calculation method for the matching coefficient of the keyword type further includes:

[0130] For a keyword type, there are i single configurations. The matching coefficients of the i single configurations are expressed as {p1, p2,..., pi}, and sampling weights are set for the i single configurations, expressed as {w1, w2,..., wi}. Then, the sampling coefficients of the i single configurations are sp = {sp1, sp2,…, spi} = {p1*w1, p2*w2,…, pi*wi}. Set the thresholds of the i single configurations as t = {t1, t2,…, ti}. If the sampling coefficient of the kth single configuration is greater than the matching threshold of the kth single configuration, it is considered that the kth single configuration passes the quality inspection. When all i single configurations pass the quality inspection, it is considered that a keyword type formed by the i single configurations passes the quality inspection. Then, the calculation formula for the matching coefficient S of a keyword type is: where k and i are positive integers, max(·) represents taking the maximum value, and d(·) is the differential operator.

[0131] After obtaining the matching coefficients of each keyword type, it is necessary to compare them with the preset matching thresholds of each keyword type to obtain the quality inspection results. As Figure 8 shown, in some implementation processes, a method for obtaining is provided:

[0132] S51: According to the matching coefficients of each keyword type, compare them with the preset matching thresholds respectively to obtain the quality inspection results.

[0133] In step S51, by way of example, after obtaining the matching coefficients of each keyword type, directly compare the matching coefficients of each keyword type with the preset matching thresholds of each keyword type, and obtain the quality inspection results according to the comparison results. For example, there are three keyword types, namely standard keywords, prohibited words, and sentiment keywords. The matching coefficients of the three keyword types are S1, S2, and S3 respectively, and the preset matching thresholds of the three keyword types are T1, T2, and T3 respectively. When S1>T1 and S2>T2 and S3>T3, it is considered that the three keyword types are all qualified, and then the quality inspection result of this time is qualified.

[0134] In some implementation processes, a method for obtaining is also provided:

[0135] S52: Set sampling weights for each keyword type, and obtain a keyword type sampling coefficient according to the matching coefficient between the sampling weight and the keyword type. The mathematical expression of the keyword type sampling coefficient SP is:

[0136] SP = S * W

[0137] where SP is the keyword type sampling coefficient, S is the matching coefficient of the keyword type, and W is the sampling weight of the keyword type;

[0138] S53: Compare the sampling coefficient with a preset matching threshold to obtain a quality inspection result.

[0139] In step S52, by way of example, after obtaining the matching coefficients of each keyword type, preset the sampling weights of each keyword type, multiply the matching coefficients of each keyword type by the corresponding sampling coefficients and then sum them up to obtain the sampling coefficient. For example, there are three keyword types, namely standard keywords, prohibited words, and sentiment keywords. The matching coefficients of the three keyword types are S1, S2, and S3 respectively, and the sampling weights of the three keyword types are W1, W2, and W3 respectively. Then the calculation formula for the sampling coefficient SP is: SP = S1 * W1 + S2 * W2 + S3 * W3.

[0140] Through the above steps, when an enterprise has different degrees of emphasis on different task types, the needs for comprehensive evaluation and judgment of the quality inspection results of each task type can be met, so as to obtain a speech quality inspection result with higher accuracy and stronger rationality.

[0141] In step S53, by way of example, after obtaining the sampling coefficient, perform a numerical comparison with a preset matching threshold to obtain a quality inspection result. For example: there are three keyword types, namely standard keywords, prohibited words, and sentiment keywords. The matching coefficients of the three keyword types are S = {S1, S2, S3}, and the sampling weights of the three keyword types are W = {W1, W2, W3}. Then the calculation formula for the sampling coefficient SP is: SP = S1 * W1 + S2 * W2 + S3 * W3. The expression of the preset matching threshold is T. If SP is greater than T, it is considered that the quality inspection result is qualified.

[0142] In some implementation processes, as Figure 9 shown, for step S52, the method for obtaining the sampling coefficient further includes:

[0143] After obtaining the matching coefficients of each keyword type, preset the sampling weights of each keyword type, multiply the matching coefficients of each keyword type by the corresponding sampling coefficients of each keyword type to obtain sampling coefficients, and keep the vector form of the sampling coefficients of each keyword type. For example, there are three keyword types, namely standard keywords, prohibited words, and sentiment keywords. The matching coefficients of the three keyword types are {S1, S2, S3}, and the sampling weights of the three keyword types are {W1, W2, W3}. Then the calculation formula for the sampling coefficient SP is: SP = {SP1, SP2, SP3} = {S1*W1, S2*W2, S3*W3}.

[0144] In some implementation processes, for step S53, the steps of obtaining the quality inspection result further include:

[0145] After obtaining the sampling coefficients in vector form, preset the matching threshold in vector form, compare the sampling coefficients in vector form with the matching threshold in vector form to obtain the quality inspection result. For example, there are three keyword types, namely standard keywords, prohibited words, and sentiment keywords. The matching coefficients of the three keyword types are {S1, S2, S3}, and the sampling weights of the three keyword types are {W1, W2, W3}. Then the calculation formula for the sampling coefficient SP is:

[0146] SP = {SP1, SP2, SP3} = {P1*W1, P2*W2, P3*W3}, and the preset matching threshold expression is {T1, T2, T3}. If SP1 > T1 and SP2 > T2 and SP3 > T3, it is considered that the quality inspection result is qualified.

[0147] It should be understood that although Figure 2-9 the steps in the flowchart of Figure 2-9 are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover,

[0148] In one embodiment, as Figure 10 described, a voice quality inspection device is provided. The voice quality inspection device includes a quality inspection result acquisition module, and the quality inspection result acquisition module includes:

[0149] The first acquisition unit is configured to set sampling weights for each keyword type, and obtain a sampling coefficient according to the matching coefficient between the sampling weight and the keyword type;

[0150] The second acquisition unit compares the sampling coefficient with a preset matching threshold to obtain a quality inspection result.

[0151] Through the quality inspection result acquisition module, an enterprise can analyze and evaluate the importance of different keyword types, set sampling weights for each keyword type according to different requirements, then preset a matching threshold, and comprehensively evaluate the matching results of each keyword type.

[0152] In the first acquisition unit, by way of example, after obtaining the matching coefficients of each keyword type, preset the sampling weights of each keyword type, multiply the matching coefficients of each keyword type by the corresponding sampling coefficients of each keyword type and then sum them up to obtain a sampling coefficient. For example, there are three keyword types, namely standard keywords, forbidden words, and sentiment keywords. The matching coefficients of the three keyword types are S1, S2, and S3 respectively, and the sampling weights of the three keyword types are W1, W2, and W3 respectively. Then the calculation formula for the sampling coefficient SP is: SP = S1 * W1 + S2 * W2 + S3 * W3.

[0153] In the second acquisition unit, by way of example, after obtaining the sampling coefficient, perform a numerical comparison with a preset matching threshold to obtain a quality inspection result. For example: there are three keyword types, namely standard keywords, forbidden words, and sentiment keywords. The matching coefficients of the three keyword types are S = {S1, S2, S3}, and the sampling weights of the three keyword types are W = {W1, W2, W3}. Then the calculation formula for the sampling coefficient SP is: SP = S1 * W1 + S2 * W2 + S3 * W3. The preset matching threshold expression is T. If SP is greater than T, it is considered that the quality inspection result is qualified.

[0154] In some implementation processes, the first acquisition unit further includes: after obtaining the matching coefficients of each keyword type, preset the sampling weights of each keyword type, multiply the matching coefficients of each keyword type by the corresponding sampling coefficients of each keyword type to obtain a sampling coefficient, and keep the vector form of the sampling coefficients of each keyword type. For example, there are three keyword types, namely standard keywords, forbidden words, and sentiment keywords. The matching coefficients of the three keyword types are S = {S1, S2, S3}, and the sampling weights of the three keyword types are W = {W1, W2, W3}. Then the calculation formula for the sampling coefficient SP is: SP = {S1 * W1, S2 * W2, S3 * W3}.

[0155] In some implementation processes, the second acquisition unit further includes: after acquiring the sampling coefficients in vector form, presetting the matching threshold in vector form, comparing the sampling coefficients in vector form with the matching threshold in vector form to obtain the quality inspection result. For example, there are three keyword types, namely standard keywords, prohibited words, and sentiment keywords. The matching coefficients of the three keyword types are S = {S1, S2, S3}, and the sampling weights of the three keyword types are W = {W1, W2, W3}. Then the sampling coefficient calculation formula is: SP = {SP1, SP2, SP3} = {S1*W1, S2*W2, S3*W3}. The preset matching threshold expression is T = {T1, T2, T3}. If SP1>T1 and SP2>T2 and SP3>T3, it is considered that the quality inspection result is qualified.

[0156] As Figure 11 described, a voice quality inspection device further includes a voice conversion text module, a task parameter configuration module, a matching information acquisition module, a matching coefficient acquisition module, and a quality inspection report acquisition module. After converting voice data into text data, performing sentence segmentation processing on the text data to obtain text segments, then configuring keyword types according to the task type, retrieving the keywords corresponding to the keyword types, updating the keyword library, and selecting keyword texts from the keyword library to compare with the text segments to obtain the matching information of the keyword texts. Then, according to the matching information of the keyword texts, comparing with the number of selected keyword texts to obtain the matching coefficient of the keyword type. Finally, according to the matching coefficient of the keyword type, comparing with the preset matching threshold to obtain the quality inspection result of the task type.

[0157] Through the above device, it is possible to improve the problems of cumbersome operation steps, low execution efficiency, poor quality inspection, high labor cost, and excessive subjectivity when manually sorting and inspecting the voice communication records generated between the customer service and the customers. Convert the voice data into text data, compare the text data with the keyword texts in the keyword library, set the matching rules, and obtain the voice quality inspection result, which can avoid the defects of the manual quality inspection method, reduce the enterprise cost, improve the voice quality inspection efficiency, and realize the systematization of customer service performance evaluation and customer service satisfaction.

[0158] In some embodiments, the steps of the task parameter configuration module include:

[0159] Configuring keyword types according to the task type, where the keyword types include: standard keywords, prohibited words, and sentiment keywords;

[0160] Retrieving the meanings and application scenarios of each keyword type to obtain the corresponding keywords;

[0161] Update the keyword library according to the keywords.

[0162] In some embodiments, the steps of the quality inspection result acquisition module include:

[0163] Compare the matching coefficients of each keyword type with the preset matching thresholds respectively to obtain the quality inspection results.

[0164] In some embodiments, the steps of the quality inspection result acquisition module include:

[0165] Set a sampling weight for each keyword type, and obtain a sampling coefficient according to the sampling weight and the matching coefficient of the keyword type;

[0166] Compare the sampling coefficient with the preset matching threshold to obtain the quality inspection result.

[0167] In some embodiments, after obtaining the matching coefficients of each single configuration, set a sampling weight for each single configuration, multiply the matching coefficient of each single configuration by the sampling weight of each single configuration and then accumulate to obtain the matching coefficient of each keyword type, then set a sampling weight for each keyword type, multiply the matching coefficient of each keyword type by the sampling weight of each keyword type and then accumulate to obtain the sampling coefficient, and then compare the obtained sampling coefficient with the preset matching threshold. If the sampling coefficient is greater than the matching threshold, it is considered that the voice quality inspection result is qualified this time, otherwise it is unqualified.

[0168] In some embodiments, after obtaining the matching coefficients of each single configuration of the standard keyword type, compare the matching coefficients of each single configuration of the standard keyword type with the preset thresholds of each single configuration of the standard keyword type. If the coefficients of each single configuration of the standard keyword type are all greater than the preset thresholds of each single configuration of the standard keyword type, it is considered that the standard keyword type is qualified. Then obtain the matching coefficients of each single configuration of the prohibited word keyword type, compare the matching coefficients of each single configuration of the prohibited word keyword type with the preset thresholds of each single configuration of the prohibited word keyword type. If the coefficients of each single configuration of the prohibited word keyword type are all greater than the preset thresholds of each single configuration of the prohibited word keyword type, it is considered that the prohibited word keyword type is qualified. Finally, obtain the matching coefficients of each single configuration of the emotion keyword type, compare the matching coefficients of each single configuration of the emotion keyword type with the preset thresholds of each single configuration of the emotion keyword type. If the coefficients of each single configuration of the emotion keyword type are all greater than the preset thresholds of each single configuration of the emotion keyword type, it is considered that the emotion keyword type is qualified. If the standard keyword type, the prohibited word keyword type, and the emotion keyword type are all qualified, it is considered that the voice quality inspection result is qualified this time. If at least one of the standard keyword type, the prohibited word keyword type, and the emotion keyword type is unqualified, it is considered that the voice quality inspection result is unqualified.

[0169] For the specific limitations of the voice quality inspection device, reference can be made to the limitations of the voice quality inspection method in the above text, which will not be elaborated here. Each module in the above voice quality inspection device can be implemented in whole or in part by software, hardware, and their combinations. Each of the above modules can be embedded in the processor of the computer device in hardware form or be independent of it, or be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each of the above modules.

[0170] In one embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 12 shown. The computer device includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data for account management. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements an account management method.

[0171] Those skilled in the art can understand that Figure 12 the structure shown in

[0172] is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0173] Convert voice data into text data, perform sentence segmentation processing on the text data, and obtain text fragments;

[0174] Configure keyword types according to the task type, retrieve keywords corresponding to the keyword types, and update the keyword library;

[0175] Select keyword texts from the keyword library, compare them with the text fragments, and obtain matching information of the keyword texts;

[0176] According to the matching information of the keyword texts, compare with the number of selected keyword texts to obtain the matching coefficient of the keyword type;

[0177] Compare the matching coefficient of the keyword type with a preset matching threshold to obtain the quality inspection result of the task type.

[0178] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0179] Convert the voice data into text data, perform sentence segmentation processing on the text data to obtain text segments;

[0180] Configure the keyword type according to the task type, retrieve the keywords corresponding to the keyword type, and update the keyword library;

[0181] Select keyword texts from the keyword library, compare them with the text segments to obtain the matching information of the keyword texts;

[0182] Compare the matching information of the keyword texts with the number of selected keyword texts to obtain the matching coefficient of the keyword type;

[0183] Compare the matching coefficient of the keyword type with a preset matching threshold to obtain the quality inspection result of the task type.

[0184] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0185] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, 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, it should be considered as the scope recorded in this specification.

[0186] The above-described embodiments merely represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.

Claims

1. A voice quality inspection method, characterized in that, Including: Converting voice data into text data, performing sentence segmentation processing on the text data, and obtaining text segments; Configuring keyword types according to the task type, retrieving keywords corresponding to the keyword types, and updating the keyword library; According to the keyword library, for each keyword type, at least one keyword text is selected respectively to form a single configuration for each keyword type; Starting from the first clause of the text segment, perform content matching and traversal on the single configuration. If the clause has a matching relationship with the keyword text in the single configuration, record the serial number of the clause in the text segment to obtain the matching position of the keyword text; Record the number of keyword texts in the clause that form a matching relationship with the single configuration. If the same keyword text in the single configuration appears multiple times in the same clause, only count this keyword text once. According to the number of keyword texts, obtain the matching times of the keyword text; According to the matching position and the matching times of the keyword text, combine the matching position and the matching times to obtain the matching information of the keyword text; According to the matching information of the keyword text, sort the matching times, and according to the largest matching times in the sorting, obtain the maximum matching times; Divide the maximum matching times by the number of selected keyword texts to obtain the matching coefficient of the single configuration; Set a sampling weight for each single configuration. Through the sampling weight and the matching coefficient of the single configuration, obtain the sampling coefficient of the single configuration. The mathematical expression of the sampling coefficient sp of the single configuration is: sp = w * p Where sp is the sampling coefficient of the single configuration, w is the sampling weight of the single configuration, and p is the matching coefficient of the single configuration; Compare the sampling coefficient of the single configuration with a preset matching threshold of the single configuration to obtain the matching coefficient of the keyword type. The mathematical expression of the matching coefficient S of the keyword type is: Where S is the matching coefficient of the keyword type, sp is the sampling coefficient of the single configuration, t is the matching threshold of the single configuration, max(·) represents taking the maximum value, and d(·) is the differential operator; According to the matching coefficient of the keyword type, compare it with a preset matching threshold to obtain the quality inspection result of the task type.

2. The voice quality inspection method according to claim 1, characterized in that, The step of configuring keyword types according to the task type, retrieving keywords corresponding to the keyword types, and updating the keyword library includes: Configuring keyword types according to the task type, where the keyword types include: standard keywords, prohibited word keywords, and sentiment keywords; According to the meanings and application scenarios of each keyword type, obtain the corresponding keywords; Update the keyword library according to the keywords.

3. The voice quality inspection method according to claim 1, wherein The step of comparing the matching coefficient of the keyword type with a preset matching threshold to obtain the quality inspection result of the task type includes: Compare the matching coefficients of each keyword type with the preset matching thresholds respectively to obtain the quality inspection result.

4. The voice quality inspection method according to claim 1 or 3, characterized in that The step of comparing the matching coefficient of the keyword type with a preset matching threshold to obtain the quality inspection result of the task type further includes: Set sampling weights for each keyword type, and obtain the keyword type sampling coefficient according to the matching coefficient between the sampling weight and the keyword type. The mathematical expression of the keyword type sampling coefficient SP is as follows: SP = S * W where SP is the keyword type sampling coefficient, S is the matching coefficient of the keyword type, and W is the sampling weight of the keyword type; Compare the sampling coefficient with a preset matching threshold to obtain the quality inspection result.

5. The voice quality inspection method according to claim 1, wherein The steps of converting voice data into text data and performing sentence segmentation on the text data to obtain text fragments include: Separate the silent content and voice content in the voice data, and obtain the time separation label; Perform sentence segmentation on the text data according to the time separation label, and add punctuation marks at the end of the sentence according to the color words at the end of the sentence; Query the number of words in the text data without punctuation marks. When the text without punctuation marks exceeds the preset word count threshold, add punctuation marks.

6. An apparatus for implementing the voice quality inspection method according to claim 1, characterized in that, The device includes: A voice-to-text conversion module for converting voice data into text data, performing sentence segmentation on the text data, and obtaining text fragments; A task parameter configuration module for configuring keyword types according to the task type, retrieving keywords corresponding to the keyword types, and updating the keyword library; A matching information acquisition module for selecting keyword texts from the keyword library, comparing them with the text fragments, and obtaining the matching information of the keyword texts; A matching coefficient acquisition module for comparing the matching information of the keyword texts with the number of selected keyword texts to obtain the matching coefficient of the keyword type; A quality inspection result acquisition module for comparing the matching coefficient of the keyword type with a preset matching threshold to obtain the quality inspection result of the task type.

7. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the voice quality inspection method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the voice quality inspection method according to any one of claims 1 to 5.

Images