A method, apparatus, device, and medium for voice call quality inspection

Abstract

This disclosure relates to a method, apparatus, device, and medium for voice call quality inspection. The method includes: determining a sample set of call data for a target customer service representative; segmenting each target text data to obtain corresponding blocks, and obtaining a call score corresponding to the target text data; determining the average of all call scores as the target score for the target customer service representative, and determining that the voice call quality of the target customer service representative is unqualified when the target score is less than or equal to a preset threshold; obtaining a second average score corresponding to blocks of the same text type among multiple target text data, and determining that the reason for the unqualified voice call quality is: the call content corresponding to the text type to which the lowest second average score belongs is unqualified. This disclosure, through the average score corresponding to blocks of the same text type, can determine the specific reasons for the unqualified voice call quality of the target customer service representative, thereby improving the service quality of customer service and enhancing the user experience in a targeted manner.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, and in particular to a method, apparatus, device, and medium for voice call quality inspection. Background Technology

[0002] In telephone sales scenarios, quality inspection of voice calls between customer service personnel and users can assess the quality of customer service personnel's voice calls, thereby effectively improving the overall level and quality of customer service personnel.

[0003] Current voice call quality inspection methods involve obtaining core keywords from the voice conversation manually by quality inspectors after the voice call is received, and then counting the number of core keywords to evaluate the quality of the voice conversation. This method can only determine whether the voice conversation quality inspection result is qualified, but it cannot determine which part of the voice conversation has a major problem if the quality inspection result is unqualified, thus failing to improve the quality of voice conversations in a targeted manner. Summary of the Invention

[0004] To address the aforementioned technical problems, this disclosure provides a method, apparatus, device, and medium for voice call quality inspection.

[0005] Firstly, this disclosure provides a method for voice call quality inspection, including:

[0006] A call data sample set for the target customer service representative is determined, wherein the call data sample set includes: multiple target text data corresponding to the multiple call data of the target customer service representative, and the multiple target text data correspond one-to-one with the multiple call data;

[0007] For each target text data, the target text data is segmented based on text type to obtain corresponding segments, and the call score corresponding to the target text data is obtained. Each segment belongs to a text type that corresponds to a score, and the call score is the first average score of the scores of the segments.

[0008] The average of the call scores of all target text data is determined as the target score corresponding to the target customer service representative, and when the target score is less than or equal to a preset threshold, the voice call quality of the target customer service representative is determined to be unqualified.

[0009] Obtain the second average score corresponding to the same text type segment in the multiple target text data, and determine that the reason for the unqualified voice call quality is: the call content corresponding to the text type to which the lowest second average score belongs is unqualified.

[0010] Optionally, before determining the target customer service call data sample set, the method further includes:

[0011] It was determined that none of the multiple target text data contained preset information.

[0012] Optionally, the above methods also include:

[0013] When a preset number of the target text data in the plurality of target text data contain preset information, it is determined that the voice call quality of the target customer service is unqualified, and the reason for the unqualified voice call quality is that the call record of the target customer service contains preset information.

[0014] Optionally, the above methods also include:

[0015] When the target score is greater than the preset threshold, the voice call quality of the target customer service representative is determined to be qualified.

[0016] Optionally, the step of determining the call data sample set of the target customer service representative includes:

[0017] Obtain the voice call dataset of the target customer service representative, which includes multiple dialogue data between the target customer service representative and different users;

[0018] The multiple dialogue data are processed by separate tracks to obtain the multi-call data corresponding to the target customer service and the user call data corresponding to different users;

[0019] For each call, speech recognition technology is used to identify the call data and obtain the corresponding target text data;

[0020] Based on the target text data, the call data sample set is formed.

[0021] Optionally, the text type includes at least one of the following: opening remarks, requirements analysis, product introduction, problem handling, invitation, time confirmation, and closing remarks.

[0022] Secondly, this disclosure provides a voice call quality inspection device, comprising:

[0023] The first determining module is used to determine the call data sample set of the target customer service representative, wherein the call data sample set includes: multiple target text data corresponding to the multiple call data of the target customer service representative, and the multiple target text data correspond one-to-one with the multiple call data.

[0024] The scoring module is used to segment each target text data based on its text type to obtain corresponding segments and to obtain the call score corresponding to the target text data. Each segment has a text type corresponding to a score, and the call score is the first average score of the scores of the segments.

[0025] The second determining module is used to determine the average of the call scores of all target text data as the target score corresponding to the target customer service, and to determine that the voice call quality of the target customer service is unqualified when the target score is less than or equal to a preset threshold.

[0026] The third determining module is used to obtain the second average score corresponding to the same text type in the multiple target text data, and determine the reason for the voice call quality being unqualified: the call content corresponding to the text type to which the lowest second average score belongs is unqualified.

[0027] Optionally, the above-mentioned device further includes:

[0028] The fourth determining module is used to determine, before determining the target customer service call data sample set, that none of the multiple target text data contain preset information.

[0029] Optionally, the above-mentioned device further includes:

[0030] The fifth determining module is used to determine that the voice call quality of the target customer service is unqualified when a preset number of the target text data in the plurality of target text data contains preset information, and the reason for the unqualified voice call quality is that the call record of the target customer service contains preset information.

[0031] Optionally, the above-mentioned device further includes:

[0032] The sixth determining module is used to determine that the voice call quality of the target customer service representative is qualified when the target score is greater than the preset threshold.

[0033] Optional, the first determining module is specifically used for:

[0034] Obtain the voice call dataset of the target customer service representative, which includes multiple dialogue data between the target customer service representative and different users;

[0035] The multiple dialogue data are processed by separate tracks to obtain the multi-call data corresponding to the target customer service and the user call data corresponding to different users;

[0036] For each call, speech recognition technology is used to identify the call data and obtain the corresponding target text data;

[0037] Based on the target text data, the call data sample set is formed.

[0038] Optionally, the text type includes at least one of the following: opening remarks, requirements analysis, product introduction, problem handling, invitation, time confirmation, and closing remarks.

[0039] Thirdly, this disclosure also provides an electronic device, including:

[0040] One or more processors;

[0041] Storage device for storing one or more programs.

[0042] When the one or more programs are executed by the one or more processors, the one or more processors implement any of the voice call quality inspection methods described in the embodiments of this disclosure.

[0043] Fourthly, this disclosure also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the voice call quality inspection methods described in the embodiments of this disclosure.

[0044] Compared with the prior art, the technical solution provided in this disclosure has the following advantages: First, a sample set of call data for the target customer service is determined. Then, for each target text data, the target text data is segmented based on text type to obtain corresponding segments. The call score corresponding to the target text data is obtained, wherein each segment belongs to a text type corresponding to a score, and the call score is the first average score of the segment scores. Then, the average of the call scores of all target text data is determined as the target score for the target customer service. When the target score is less than or equal to a preset threshold, the voice call quality of the target customer service is determined to be unqualified. Finally, the second average score corresponding to the segments of the same text type in multiple target text data is obtained, and the reason for the unqualified voice call quality is determined to be: the call content corresponding to the text type to which the lowest second average score belongs is unqualified. In the above technical solution, by segmenting each target text data to obtain corresponding segments and the average score corresponding to the segments of the same text type, the specific reason for the unqualified voice call quality of the target customer service can be determined, thereby improving the accuracy of quality inspection and improving the service quality of customer service and enhancing the user experience in a targeted manner. Attached Figure Description

[0045] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0046] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1AThis is a flowchart illustrating a voice call quality inspection method provided in an embodiment of this disclosure;

[0048] Figure 1B This is a schematic diagram of the structure of a target classification model provided in an embodiment of this disclosure;

[0049] Figure 2 This is a flowchart illustrating another voice call quality inspection method provided in this embodiment of the disclosure;

[0050] Figure 3 This is a schematic diagram of the structure of a voice call quality inspection device provided in an embodiment of this disclosure;

[0051] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0052] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0053] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0054] Figure 1A This is a flowchart illustrating a voice call quality inspection method provided in this embodiment. This embodiment is applicable to situations involving quality inspection of customer service voice call data. The method in this embodiment can be executed by a voice call quality inspection device, which can be implemented in hardware / software and configured in an electronic device. Figure 1A As shown, the method specifically includes the following:

[0055] S110, determine the call data sample set of the target customer service, wherein the call data sample set includes: multiple target text data corresponding to the multiple call data of the target customer service, and the multiple target text data correspond one-to-one with the multiple call data.

[0056] In this context, the target customer service representative is the customer service person whose voice call quality is to be inspected. The call data refers to the audio data corresponding to the target customer service representative during the voice call, which can be obtained through a recording module or other methods; this embodiment does not limit this. Customer service personnel can be understood as individuals who provide telephone answers, consultations, or telephone sales services, such as telephone sales personnel in the automotive industry.

[0057] During the quality inspection of voice calls, it is necessary to obtain multiple target text data corresponding to the multiple call data of the target customer service. Since the target customer service may make or receive many calls every day, countless call data will be generated. In order to improve the efficiency of quality inspection, this embodiment can extract representative multiple call data from the countless call data of the target customer service. The specific extraction method is not limited in this embodiment.

[0058] S120: For each target text data, the target text data is segmented based on text type to obtain corresponding segments, and the call score corresponding to the target text data is obtained. Each segment belongs to a text type that corresponds to a score, and the call score is the first average score of the segment scores.

[0059] The text type can be determined based on the scenario corresponding to the call data. The score corresponding to the text type can be preset or determined according to the specific situation; this embodiment does not impose any limitations on this.

[0060] After obtaining multiple target texts corresponding to the target customer service, due to the potentially large amount of content in the text data, for ease of analysis, each target text is segmented based on its text type to obtain corresponding segments. For example, the target text data is segmented according to the text type of each sentence, with sentences of the same type grouped as one segment, resulting in at least one segment. After obtaining the segments corresponding to each target text data, since each segment corresponds to a text type, and each text type corresponds to a score, the first average score obtained by adding up the scores of the segments corresponding to the same target text data and taking the average is the call score corresponding to that target text data.

[0061] For example, the first average score can be expressed by the following formula:

[0062]

[0063] Where Score represents the average score, n represents the total number of chunks, and S i Let i represent the score corresponding to the i-th slice, where i = 1, 2, ..., n.

[0064] In some embodiments, if the scenario corresponding to the call data is a car telemarketing scenario, the text type may include at least one of the following: opening remarks, needs analysis, product introduction, problem handling, invitation, time confirmation, and closing remarks. It may also include cognitive test drives, packaged test drives, or others, which are not limited in this embodiment.

[0065] In this context, a cognitive test drive can be understood as a simple test drive, while a packaged test drive can be understood as an in-depth test drive that allows users to experience the vehicle's technological features. For example, in a car telemarketing scenario, the correspondence between text type and score can be represented by Table 1 below:

[0066] Table 1

[0067]

[0068] Suppose the target text data is segmented into 7 segments: segment 1 is an opening statement, segment 2 is a requirements analysis, segment 3 is a product introduction, segment 4 is a test drive introduction, segment 5 is an invitation, segment 6 is a time announcement, and segment 7 is a closing statement. According to Table 1, segment 1 has a score of 1, segment 2 has a score of 2, segment 3 has a score of 2, segment 4 has a score of 4, segment 5 has a score of 4, segment 6 has a score of 3, and segment 7 has a score of 1.

[0069] S130, the average of the call scores of all target text data is determined as the target score corresponding to the target customer service, and when the target score is less than or equal to a preset threshold, the voice call quality of the target customer service is determined to be unqualified.

[0070] The target score can be understood as the call score corresponding to the target customer service representative. The preset threshold can be a pre-set value, such as 10, or it can be determined according to the specific situation; this embodiment does not limit this.

[0071] After determining the call score corresponding to each target text data, the call scores of all target text data are summed, and the average value is calculated. This average value is the target score for the target customer service representative. The target score is compared with a preset threshold. Based on the relationship between the target score and the preset threshold, it can be determined whether the voice call quality of the target customer service representative is acceptable. When the target score is less than or equal to the preset threshold, it indicates that the voice call quality of the target customer service representative is unacceptable.

[0072] S140, obtain the second average score corresponding to the same text type in multiple target text data, and determine that the reason for the unqualified voice call quality is: the call content corresponding to the text type to which the lowest second average score belongs is unqualified.

[0073] Since the multiple call data of the target customer service correspond to multiple target text data, and each target text data is divided into multiple segments, after determining that the voice call quality of the target customer service is unqualified, the scores corresponding to the segments of the same text type in the multiple target text data are added together, and the average value is calculated to obtain the second average score. By comparing the size of each second average score, it can be determined that the reason for the unqualified voice call quality is that the call content corresponding to the text type of the smallest second average score is unqualified.

[0074] In this embodiment, firstly, a sample set of call data for the target customer service is determined. Then, for each target text data, the target text data is segmented based on text type to obtain corresponding segments. The call score corresponding to the target text data is obtained, where each segment belongs to a text type corresponding to a score. The call score is the first average score of the segment scores. Then, the average of the call scores of all target text data is determined as the target score for the target customer service. When the target score is less than or equal to a preset threshold, the voice call quality of the target customer service is determined to be unqualified. Finally, the second average score corresponding to the segments of the same text type in multiple target text data is obtained. The reason for the unqualified voice call quality is determined to be: the call content corresponding to the text type of the lowest second average score is unqualified. In the above technical solution, by segmenting each target text data to obtain corresponding segments and the average score corresponding to the segments of the same text type, the specific reasons for the unqualified voice call quality of the target customer service can be determined. This can improve the accuracy of quality inspection, and improve the service quality of customer service and enhance the user experience in a targeted manner. At the same time, it can also effectively perform statistical analysis, helping enterprises to continuously optimize the customer service workflow.

[0075] In some embodiments, optionally, before segmenting the target text data based on text type to obtain corresponding segments for each target text data, the method may further include:

[0076] For each target text data, each sentence in the target text data is segmented using word segmentation technology to obtain the corresponding word segmentation results;

[0077] The word segmentation results are input into the target classification model to obtain the text type to which each sentence belongs.

[0078] The target classification model can be a pre-trained model, such as a Text Convolutional Neural Network (TextCNN), a Text Recurrent Neural Network (TextRNN), or a pre-trained BERT (Bidirectional Encoder Representation from Transformers) model. This embodiment does not limit the specific model.

[0079] Specifically, since the target text data may contain multiple sentences, each corresponding to a different text type, annotating the text type of each sentence individually would be time-consuming and labor-intensive. Therefore, for each target text data point, each sentence is segmented using the corresponding word segmentation technique to obtain the corresponding word segmentation result. After obtaining the word segmentation result for each sentence, these results are sequentially input into the target classification model to determine the text type of each sentence.

[0080] In this embodiment, the text type of the target text data is determined by the above method, which is simple, efficient, saves manpower and resources, and avoids errors.

[0081] In some embodiments, the target classification model may optionally include: a word embedding layer, a bidirectional long short-term memory network, a self-attention layer, a pooling layer, and a fully connected layer;

[0082] The word embedding layer is used to generate a representation vector based on the word segmentation result;

[0083] The bidirectional long short-term memory network is used to extract features from the representation vector to obtain a semantic vector;

[0084] The self-attention layer is used to extract keywords from the semantic vector to obtain the corresponding word vectors;

[0085] The pooling layer is used to convert the word vectors into sentence vectors;

[0086] The fully connected layer is used to predict the sentence vector to obtain the text type to which each sentence text belongs.

[0087] The word embedding layer can employ various encoding methods, such as token embedding, segment embedding, and position embedding; this embodiment does not impose any limitations on this. A Bidirectional Long Short-Term Memory (BiLSTM) network is a type of recurrent neural network used to extract contextual information from representation vectors to obtain semantic vectors. The BiLSTM, self-attention layer, and pooling layer constitute the encoder part. The fully connected layer, which can be a multi-layer perceptron neural network (MLP), serves as the decoder part.

[0088] In this embodiment, the target classification model described above can accurately classify the word segmentation results of each sentence text, obtain the text type to which each sentence text belongs, facilitate the smooth progress of subsequent processes, and improve work efficiency.

[0089] For example, Figure 1B This is a schematic diagram of the structure of a target classification model provided in an embodiment of this disclosure. For example... Figure 1B As shown, the input to this target classification model is the word segmentation result. The model includes a word embedding layer, a bidirectional long short-term memory network, a self-attention layer, a pooling layer, and a fully connected layer. Their functions have been described in the above embodiments, and will not be repeated here to avoid repetition.

[0090] In some embodiments, optionally, predicting the sentence vector to obtain the text type to which each sentence text belongs may specifically include:

[0091] Predict the sentence vector to obtain the probability value of the sentence vector under each text type;

[0092] The text type corresponding to the highest probability value among the probability values ​​is determined as the text type to which each sentence belongs.

[0093] Specifically, when the fully connected layer predicts the sentence vector, the output is the probability value of the sentence vector under each text type. By taking the text type corresponding to the highest probability value as the text type of each sentence text, the text type of each sentence text in the target text data can be determined.

[0094] In this embodiment, the text type of each sentence is determined by the above method, which is simple, quick, and can save time and improve work efficiency.

[0095] Figure 2 This is a flowchart illustrating another voice call quality inspection method provided in this embodiment. This embodiment is an optimization based on the above embodiment. Optionally, this embodiment provides a detailed explanation of the process prior to determining the call data sample set of the target customer service representative. Figure 2 As shown, the method specifically includes the following:

[0096] S210, determine the call data sample set of the target customer service representative, wherein the call data sample set includes: multiple target text data corresponding to the multiple call data of the target customer service representative, and the multiple target text data correspond one-to-one with the multiple call data.

[0097] S220, determine whether multiple target text data contain preset information.

[0098] If yes, execute S230; if no (i.e., none of the target text data contains the preset information), execute S240-S260.

[0099] The preset information can be understood as sensitive words, such as words that are abusive, pornographic, or politically sensitive.

[0100] After obtaining multiple target text data, it is necessary to determine whether the multiple target text data contain preset information. Specifically, each target text data can be compared with the vocabulary database corresponding to sensitive words to determine whether the target text data contains sensitive words.

[0101] S230, It is determined that the voice call quality of the target customer service representative is unqualified, and the reason for the unqualified voice call quality is that the target customer service representative's call record contains preset information.

[0102] The preset number can be a predetermined value, such as 1, or it can be determined according to the specific situation. This embodiment does not limit this.

[0103] Since preset information can be directly used as a condition to determine whether the voice call quality is qualified, when a preset number of target text data contain preset information in multiple target text data, it can be determined that the voice call quality of the target customer service is unqualified, and the reason for the unqualified voice call quality is that the target customer service's call record contains preset information.

[0104] S240, for each target text data, the target text data is segmented based on text type to obtain corresponding segments, and the call score corresponding to the target text data is obtained. Among them, the text type to which each segment belongs corresponds to a score, and the call score is the first average score of the segment scores.

[0105] S250, the average of the call scores of all target text data is determined as the target score for the target customer service representative, and when the target score is less than or equal to a preset threshold, the voice call quality of the target customer service representative is determined to be unqualified.

[0106] S260, obtain the second average score corresponding to the same text type in multiple target text data, and determine that the reason for the unqualified voice call quality is: the call content corresponding to the text type to which the lowest second average score belongs is unqualified.

[0107] In this embodiment, firstly, a sample set of call data for the target customer service representative is determined. Then, it is determined whether multiple target text data sets contain preset information. If a preset number of target text data sets contain preset information, the voice call quality of the target customer service representative is determined to be substandard, and the reason for the substandard voice call quality is that the call record of the target customer service representative contains preset information. If none of the multiple target text data sets contain preset information, for each target text data set, the target text data is segmented based on text type to obtain corresponding segments. The call score corresponding to the target text data is obtained, and the average of the call scores of all target text data sets is determined as the target score for the target customer service representative. If the target score is less than or equal to a preset threshold, the voice call quality of the target customer service representative is determined to be substandard. Finally, the second average score corresponding to the segments of the same text type among the multiple target text data sets is obtained, and the reason for the substandard voice call quality is determined to be: the lowest average score among the second average scores. The second average score indicates that the call content corresponding to the text type is unqualified. In the above technical solution, by determining whether multiple target text data contain preset information, it is possible to determine whether multiple target text data contain sensitive words. If a preset number of target text data contain preset information, it means that the quality inspection has failed, the voice call quality of the target customer service is unqualified, and the reason for the unqualification is that the call record of the target customer service contains preset information. This can directly determine the quality inspection result, saving time and work efficiency. If multiple target text data do not contain preset information, each target text data is segmented to obtain the corresponding segment. By using the average score corresponding to the segment of the same text type, the specific reason for the unqualified voice call quality of the target customer service can be determined. This can improve the accuracy of quality inspection, and improve the service quality of customer service and enhance the user experience in a targeted manner. At the same time, it can also be well statistically analyzed to help enterprises continuously optimize the customer service workflow.

[0108] In some embodiments, optionally, the method may further include:

[0109] When the target score is greater than the preset threshold, the voice call quality of the target customer service representative is determined to be qualified.

[0110] In this embodiment, if the target score is greater than the preset threshold, it means that the quality inspection has passed, that is, the voice call quality of the target customer service representative is qualified. Through the above process, the quality inspection result can be quickly determined, which is conducive to improving the service quality of customer service representatives.

[0111] In some embodiments, optionally, determining the call data sample set of the target customer service representative may specifically include:

[0112] Obtain the voice call dataset of the target customer service representative, which includes multiple dialogue data between the target customer service representative and different users;

[0113] The multiple dialogue data are processed by separate tracks to obtain the multi-call data corresponding to the target customer service and the user call data corresponding to different users;

[0114] For each call, speech recognition technology is used to identify the call data and obtain the corresponding target text data;

[0115] Based on the target text data, the call data sample set is formed.

[0116] Specifically, voice call datasets can be obtained through recording modules or information collection systems. After obtaining the dataset, models such as Transformer can be used to segment multiple dialogues within it. Specifically, call data corresponding to the target customer service representative can be labeled as 0, and user call data can be labeled as 1, thus distinguishing between target customer service and user call data, resulting in multiple call data corresponding to the target customer service representative and user call data corresponding to the user. Finally, each call data in the multiple call dataset is recognized using Automatic Speech Recognition (ASR) technology to obtain the corresponding target text data. Summarizing all the target text data into a single set yields the call data sample set.

[0117] It should be noted that the target text data can also be obtained through other recognition methods, which are not limited in this embodiment.

[0118] In this embodiment, the call data sample set is obtained through the above method, which is simple, efficient and fast.

[0119] In some embodiments, optionally, after obtaining the target customer service's voice call dataset, the process may further include:

[0120] After acquiring multiple dialogue data sets, target dialogue data with analytical value is selected based on the duration of each dialogue.

[0121] The preset time threshold can be a pre-set value, such as 2 minutes, or it can be determined according to the specific situation. This embodiment does not limit this.

[0122] Specifically, if the duration of a certain dialogue data exceeds a preset time threshold among multiple dialogue data, then that dialogue data is used as the target dialogue data. This facilitates subsequent track-based processing of the target dialogue data to obtain the call data corresponding to the target customer service representative and the user's call data. Additionally, each call data of the target customer service representative is identified using automatic speech recognition technology to obtain the corresponding target text data.

[0123] In this embodiment, filtering multiple dialogue data points helps improve the quality of the dialogue data and ensures the smooth progress of subsequent processes.

[0124] Figure 3 This is a schematic diagram of a voice call quality inspection device provided in an embodiment of this disclosure; the device is configured in an electronic device and can implement the voice call quality inspection method described in any embodiment of this application. The device specifically includes the following:

[0125] The first determining module 310 is used to determine the call data sample set of the target customer service representative, wherein the call data sample set includes: multiple target text data corresponding to the multiple call data of the target customer service representative, and the multiple target text data correspond one-to-one with the multiple call data.

[0126] The scoring module 320 is used to segment the target text data based on text type for each target text data to obtain corresponding segments and obtain the call score corresponding to the target text data. Each segment belongs to a text type that corresponds to a score, and the call score is the first average score of the scores of the segments.

[0127] The second determining module 330 is used to determine the average of the call scores of all target text data as the target score corresponding to the target customer service, and to determine that the voice call quality of the target customer service is unqualified when the target score is less than or equal to a preset threshold.

[0128] The third determining module 340 is used to obtain the second average score corresponding to the same text type of the multiple target text data, and determine that the reason for the voice call quality being unqualified is: the call content corresponding to the text type to which the lowest second average score belongs is unqualified.

[0129] Optionally, the above-mentioned device further includes:

[0130] The fourth determining module is used to determine, before determining the target customer service call data sample set, that none of the multiple target text data contain preset information.

[0131] Optionally, the above-mentioned device further includes:

[0132] The fifth determining module is used to determine that the voice call quality of the target customer service is unqualified when a preset number of the target text data in the plurality of target text data contains preset information, and the reason for the unqualified voice call quality is that the call record of the target customer service contains preset information.

[0133] Optionally, the above-mentioned device further includes:

[0134] The sixth determining module is used to determine that the voice call quality of the target customer service representative is qualified when the target score is greater than the preset threshold.

[0135] Optionally, the first determining module 310 is specifically used for:

[0136] Obtain the voice call dataset of the target customer service representative, which includes multiple dialogue data between the target customer service representative and different users;

[0137] The multiple dialogue data are processed by separate tracks to obtain the multi-call data corresponding to the target customer service and the user call data corresponding to different users;

[0138] For each call, speech recognition technology is used to identify the call data and obtain the corresponding target text data;

[0139] Based on the target text data, the call data sample set is formed.

[0140] Optionally, the text type includes at least one of the following: opening remarks, requirements analysis, product introduction, problem handling, invitation, time confirmation, and closing remarks.

[0141] The voice call quality inspection device provided in this embodiment first determines a sample set of call data for the target customer service representative. Then, for each target text data point, it segments the text data based on text type to obtain corresponding segments. A call score is obtained for each target text data point, where each segment's text type corresponds to a score. The call score is the first average score of the segment's scores. The average call scores of all target text data points are then determined as the target score for the target customer service representative. If the target score is less than or equal to a preset threshold, the voice call quality of the target customer service representative is determined to be unqualified. Finally, the second average score corresponding to segments of the same text type from multiple target text data points is obtained. The reason for the unqualified voice call quality is determined to be that the call content corresponding to the text type with the smallest second average score is unqualified. This technical solution, by segmenting each target text data point to obtain corresponding segments and the average score corresponding to segments of the same text type, can determine the specific reasons for the unqualified voice call quality of the target customer service representative. This improves the accuracy of quality inspection, allows for targeted improvement of customer service quality and user experience, and also enables effective statistical analysis, helping enterprises continuously optimize customer service workflows.

[0142] The voice call quality inspection device provided in this disclosure can execute the voice call quality inspection method provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects of the method execution.

[0143] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. For example... Figure 4 As shown, the electronic device includes a processor 410 and a storage device 420; the number of processors 410 in the electronic device can be one or more. Figure 4 Taking a processor 410 as an example; the processor 410 and the storage device 420 in the electronic device can be connected via a bus or other means. Figure 4 Taking the example of a connection between China and Israel via a bus.

[0144] Storage device 420, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the voice call quality inspection method in the embodiments of this disclosure. Processor 410 executes various functional applications and data processing of electronic devices by running the software programs, instructions, and modules stored in storage device 420, thereby implementing the voice call quality inspection method provided in the embodiments of this disclosure.

[0145] Storage device 420 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a given function; the data storage area may store data created based on terminal usage. Furthermore, storage device 420 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory, or other non-volatile solid-state storage device. In some instances, storage device 420 may further include memory remotely located relative to processor 410, which can be connected to electronic devices via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0146] The electronic device provided in this embodiment can be used to execute the voice call quality inspection method provided in any of the above embodiments, and has the corresponding functions and beneficial effects.

[0147] This disclosure also provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to implement the voice call quality inspection method provided in this disclosure.

[0148] Of course, the computer-executable instructions provided in the embodiments of this disclosure are not limited to the operation of the method described above, but can also perform related operations in the voice call quality inspection method provided in any embodiment of this disclosure.

[0149] Based on the above description of the implementation methods, those skilled in the art will clearly understand that this disclosure can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of this disclosure.

[0150] It is worth noting that in the embodiments of the above-mentioned voice call quality inspection device, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of this disclosure.

[0151] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0152] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for quality inspection of voice calls, characterized in that, include: A call data sample set for the target customer service representative is determined, wherein the call data sample set includes: multiple target text data corresponding to the multiple call data of the target customer service representative, and the multiple target text data correspond one-to-one with the multiple call data; For each target text data, the target text data is segmented based on text type to obtain corresponding segments, and the call score corresponding to the target text data is obtained. Each segment belongs to a text type that corresponds to a score, and the call score is the first average score of the scores of the segments. The average of the call scores of all target text data is determined as the target score corresponding to the target customer service representative, and when the target score is less than or equal to a preset threshold, the voice call quality of the target customer service representative is determined to be unqualified. Obtain the second average score corresponding to the same text type segment in the multiple target text data, and determine that the reason for the unqualified voice call quality is: the call content corresponding to the text type to which the lowest second average score belongs is unqualified.

2. The method according to claim 1, characterized in that, Before determining the target customer service call data sample set, the method further includes: It was determined that none of the multiple target text data contained preset information.

3. The method according to claim 2, characterized in that, Also includes: When a preset number of the target text data in the plurality of target text data contain preset information, it is determined that the voice call quality of the target customer service is unqualified, and the reason for the unqualified voice call quality is that the call record of the target customer service contains preset information.

4. The method according to claim 1, characterized in that, Also includes: When the target score is greater than the preset threshold, the voice call quality of the target customer service representative is determined to be qualified.

5. The method according to any one of claims 1-4, characterized in that, The call data sample set for determining the target customer service representative includes: Obtain the voice call dataset of the target customer service representative, which includes multiple dialogue data between the target customer service representative and different users; The multiple dialogue data are processed by separate tracks to obtain the multi-call data corresponding to the target customer service and the user call data corresponding to different users; For each call, speech recognition technology is used to identify the call data and obtain the corresponding target text data; Based on the target text data, the call data sample set is formed.

6. The method according to any one of claims 1-4, characterized in that, The text types include at least one of the following: opening remarks, requirements analysis, product introduction, problem handling, invitation, time confirmation, and closing remarks.

7. A voice call quality inspection device, characterized in that, The device includes: The first determining module is used to determine the call data sample set of the target customer service representative, wherein the call data sample set includes: multiple target text data corresponding to the multiple call data of the target customer service representative, and the multiple target text data correspond one-to-one with the multiple call data. The scoring module is used to segment each target text data based on its text type to obtain corresponding segments and to obtain the call score corresponding to the target text data. Each segment has a text type corresponding to a score, and the call score is the first average score of the scores of the segments. The second determining module is used to determine the average of the call scores of all target text data as the target score corresponding to the target customer service, and to determine that the voice call quality of the target customer service is unqualified when the target score is less than or equal to a preset threshold. The third determining module is used to obtain the second average score corresponding to the same text type in the multiple target text data, and determine the reason for the voice call quality being unqualified: the call content corresponding to the text type to which the lowest second average score belongs is unqualified.

8. The apparatus according to claim 7, characterized in that, Also includes: The fourth determining module is used to determine, before determining the target customer service call data sample set, that none of the multiple target text data contain preset information.

9. An electronic device, characterized in that, include: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-6.

Images