Recording data quality inspection method, device, equipment, storage medium and program product
By using quality inspection models combined with video and audio recognition sub-models for automated inspection in recorded data quality inspection, the problem of small audit volume and easy to miss is solved in the existing quality inspection methods, and efficient and accurate quality inspection of recorded data is achieved, reducing the quality inspection of efficient and accurate data. cost.
Patent Information
- Application Number
- CN202210271387.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-03-18
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2042-03-18
AI Technical Summary
The existing recording data quality inspection methods have small audit volume and are prone to errors and omissions, which cannot meet the rapidly growing sales business needs.
The quality inspection model is adopted, and the recorded data is detected and processed in combination with the video recognition sub-model and the audio recognition sub-model, and the quality inspection results are generated, and confirmed in the quality inspection background to ensure the accuracy and efficiency of the quality inspection results.
Through automated quality inspection processes, the efficiency and accuracy of quality inspection are greatly improved, the workload and personnel costs in the quality inspection backend are reduced, and the compliance of recorded data is ensured.
Smart Images

Figure CN114679557B_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of artificial intelligence technology, and in particular, relates to a recording data quality inspection method, device, equipment, storage medium and program product. Background Art
[0002] At present, with the economic development of society, people's income level is gradually increasing, residents' financial management awareness is gradually strengthening, and people's willingness to manage assets is also gradually increasing. Financial products such as funds, financial management and insurance policies have become more common investment options.
[0003] In order to protect the rights and interests of customers, regulatory agencies have clearly stipulated that the sales process must be recorded through audio and video recording (double recording) when sales occur for subsequent review. Therefore, in order to ensure the compliance of the recorded data and avoid related risks of the recorded data, it is necessary to conduct quality inspection on the recorded data.
[0004] The existing recording data quality inspection method is mainly that the back-end quality inspection personnel extract a part of the recording data from the database storing the recording data, and use manual identification to conduct quality inspection and review of this part of the recording data to confirm the compliance of the recording data. However, with the substantial increase in sales business, the audit volume of this extraction audit method cannot meet the requirements. In addition, the quality inspection personnel are prone to errors and omissions during the long audit process. Summary of the invention
[0005] The embodiments of the present application provide a recording data quality inspection method, device, equipment, storage medium and program product, which can solve the problem that the existing recording data quality inspection method has a small audit volume and is prone to errors and omissions.
[0006] In a first aspect, an embodiment of the present application provides a recording data quality inspection method, wherein the recording data includes audio data and video data, and the method includes:
[0007] Obtain the recorded data to be inspected;
[0008] Inputting the recorded data to be quality inspected into a quality inspection model, wherein the quality inspection model includes a video recognition sub-model and an audio recognition sub-model;
[0009] Performing detection and processing on the recorded data according to the video recognition sub-model and the audio recognition sub-model respectively to obtain a first quality inspection result of the recorded data, wherein the first quality inspection result includes a sound recording quality inspection result and a video recording quality inspection result;
[0010] When at least one of the audio quality inspection result and the video quality inspection result of the first quality inspection result is an abnormal quality inspection result, abnormal recording data is sent to the quality inspection background, and the abnormal recording data is the recording data corresponding to the abnormal quality inspection result;
[0011] When the quality inspection confirmation result generated by the quality inspection background according to the abnormal recording data is inconsistent with the first quality inspection result, the quality inspection confirmation result is used as the final quality inspection result.
[0012] In some embodiments, the detecting and processing the recorded data according to the video recognition sub-model and the audio recognition sub-model to obtain a first quality inspection result of the recorded data includes:
[0013] Determine the video data corresponding to the video recognition sub-model and the audio data corresponding to the audio recognition sub-model from the recorded data to be quality inspected;
[0014] Processing the video data according to the video recognition sub-model to obtain a video quality inspection result of the video data;
[0015] Processing the recording data according to the audio recognition sub-model to obtain a recording quality inspection result of the recording data;
[0016] The first quality inspection result is generated according to the video quality inspection result and the audio quality inspection result.
[0017] In some embodiments, the processing of the video data according to the video recognition sub-model to obtain a video quality inspection result of the video data includes:
[0018] Determine each service node according to the service type corresponding to the recorded data;
[0019] Dividing the video data into a plurality of video sub-data according to each service node, wherein the plurality of video sub-data correspond to each service node one by one;
[0020] The video content is identified on the multiple video sub-data respectively according to the video identification sub-model to obtain a video quality inspection result.
[0021] In some embodiments, dividing the video data into a plurality of video sub-data according to each service node includes:
[0022] Obtain the video features corresponding to each business node;
[0023] Determine the video frames corresponding to the respective video features from the video data;
[0024] The video data is divided according to each video frame to obtain a plurality of video sub-data.
[0025] In some embodiments, the processing of the recording data according to the audio recognition sub-model to obtain a recording quality inspection result of the recording data includes:
[0026] Determine a business keyword according to the business type corresponding to the recorded data, wherein the business keyword includes necessary keywords and illegal keywords in the business handling process;
[0027] Performing speech recognition processing on the recorded data to obtain speech text information;
[0028] Matching the voice text information with the business keywords to obtain a matching result;
[0029] The recording quality inspection result is generated according to the matching result.
[0030] In some embodiments, after taking the quality inspection confirmation result as the final quality inspection result, the method further includes:
[0031] Acquire a plurality of recording data whose quality inspection confirmation results are inconsistent with the first quality inspection results;
[0032] Arranging and counting the multiple recording data, and determining the problematic service nodes corresponding to the multiple recording data respectively;
[0033] According to the problem business node, the model parameters of the quality inspection model are adjusted and optimized to update the quality inspection model.
[0034] In some embodiments, adjusting and optimizing the model parameters of the quality inspection model according to the problematic service node to update the quality inspection model includes:
[0035] Adjusting the model parameters according to the problem business node to obtain an adjusted quality inspection model;
[0036] Inputting a plurality of recording data into the adjusted quality inspection model to obtain second quality inspection results respectively corresponding to the plurality of recording data;
[0037] Calculating a first matching degree between the quality inspection confirmation results of the plurality of recording data and the plurality of first quality inspection results, and a second matching degree between the quality inspection confirmation results of the plurality of recording data and the plurality of second quality inspection results;
[0038] Determining an optimized quality inspection result from the first quality inspection result and the second quality inspection result according to the first matching degree and the second matching degree;
[0039] The quality inspection model corresponding to the optimized quality inspection result is used as the updated quality inspection model.
[0040] In a second aspect, an embodiment of the present application provides a recording data quality inspection device, the device comprising:
[0041] Acquisition module, to obtain the recorded data to be inspected;
[0042] An input module, inputting the recorded data to be quality inspected into a quality inspection model, wherein the quality inspection model includes a video recognition sub-model and an audio recognition sub-model;
[0043] A detection module, which performs detection processing on the recorded data according to the video recognition sub-model and the audio recognition sub-model to obtain a first quality inspection result of the recorded data;
[0044] An abnormality module, when the first quality inspection result is an abnormal quality inspection result, sends quality inspection abnormality information to the quality inspection background;
[0045] The confirmation module receives the quality inspection confirmation result sent by the quality inspection background, and replaces the abnormal quality inspection result with the quality inspection confirmation result.
[0046] In a third aspect, an embodiment of the present application provides a recording data quality inspection device, the device comprising: a processor and a memory storing computer program instructions;
[0047] The processor implements the above recorded data quality inspection method when executing the computer program instructions.
[0048] In a fourth aspect, an embodiment of the present application provides a computer storage medium, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the above-mentioned recording data quality inspection method is implemented.
[0049] In a fifth aspect, an embodiment of the present application provides a computer program product, which includes computer program instructions, and when the computer program instructions are executed by a processor, the above-mentioned recording data quality inspection method is implemented.
[0050] In all embodiments of the present application, the acquisition, storage, use, and processing of data are in compliance with relevant provisions of national laws and regulations.
[0051] The recording data quality inspection method, device, equipment, storage medium and program product provided by the embodiment of the present application, by inputting the recording data to be quality inspected into the quality inspection model, the recording data and the video data in the recording data can be detected and processed respectively by the sub-model in the quality inspection model to obtain the first quality inspection result of the recording data. When the first quality inspection result includes at least one abnormal quality inspection result, the abnormal recording data can be sent to the quality inspection background to confirm the quality inspection of the abnormal recording data through the quality inspection background. When the quality inspection confirmation result is the same as the abnormal quality inspection result, the first quality inspection result can be used as the final quality inspection result; when the quality inspection confirmation result is inconsistent with the abnormal quality inspection result, the quality inspection confirmation result can be used as the final quality inspection result. Through the quality inspection model, a large amount of recording data can be batch-automated for quality inspection. For recording data with no abnormal quality inspection results, the first quality inspection result can be directly used as the final quality inspection result. For recording data with abnormalities, secondary quality inspection confirmation is performed through the quality inspection background to obtain the final quality inspection result. Quality inspection through the quality inspection model can greatly improve the efficiency and accuracy of quality inspection. In addition, the personnel in the quality inspection background only need to conduct quality inspection and confirmation on the recorded data that have been identified as abnormal, which reduces the quality inspection workload and personnel costs of the quality inspection background. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0053] Figure 1 It is a flowchart of a recording data quality inspection method provided in an embodiment of the present application;
[0054] Figure 2 is a flowchart of a recording data quality inspection method provided by another embodiment of the present application;
[0055] Figure 3 It is a detailed flow chart of step S220 provided in an embodiment of the present application;
[0056] Figure 4 is a detailed flow chart of step S220 provided in another embodiment of the present application;
[0057] Figure 5 It is a detailed flow chart of step S230 provided in an embodiment of the present application;
[0058] Figure 6 This is a partial flow chart of a recording data quality inspection method provided in an embodiment of the present application;
[0059] Figure 7 It is a partial flow chart of a recording data quality inspection method provided by another embodiment of the present application;
[0060] Figure 8 It is a structural schematic diagram of a recording data quality inspection device provided in an embodiment of the present application;
[0061] Fig. 9 It is a schematic diagram of the hardware structure of a recording data quality inspection device provided in one embodiment of the present application. DETAILED DESCRIPTION
[0062] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by illustrating the examples of the present application.
[0063] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.
[0064] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The embodiments will be described in detail below in conjunction with the accompanying drawings.
[0065] At present, with the economic development of society, people's income level is gradually increasing, residents' financial management awareness is gradually strengthening, and they are more willing to invest part of their assets to earn returns. The characteristic of funds being managed by professional practitioners is particularly popular among ordinary residents, and buying funds has become a more common investment choice. In addition, financial products such as financial management and insurance are also gradually becoming popular.
[0066] Due to the wide variety of fund product categories, it is difficult for ordinary customers to identify the differences and risks of each product. According to the requirements of regulatory authorities, when banks sell funds to customers, they must provide them with necessary information in a clear and noticeable manner in accordance with regulations, and fully inform customers of the legal risks, policy risks and market risks involved in the products sold. They must also strictly implement the requirements for audio and video recording in sales areas.
[0067] However, there are inevitably omissions in the sales process, such as unclear reading of the sales script leading to unclear customer understanding, missing necessary sales links leading to non-compliant sales processes, etc. Therefore, in order to ensure the compliance of the sales process, it is necessary to conduct quality inspection on the recorded data. The existing quality inspection method is mainly for quality inspectors to manually check the recorded data from the database and use manual recognition to conduct quality inspection and review of the data. However, with the continuous increase in sales business volume, the quality inspection efficiency and accuracy of the existing quality inspection method cannot meet the needs, and additional personnel costs are required to match the increase in business volume.
[0068] In order to solve the problems of the prior art, the embodiments of the present application provide a recording data quality inspection method, device, equipment, storage medium and program product. The recording data quality inspection method provided by the embodiments of the present application is first introduced below. In all the following implementation methods of the present application, the acquisition, storage, use and processing of data are in compliance with the relevant provisions of national laws and regulations.
[0069] Figure 1 A flow chart of a recording data quality inspection method provided by an embodiment of the present application is shown. The method comprises the following steps:
[0070] S110, obtaining recorded data to be inspected;
[0071] S120, inputting the recorded data to be quality inspected into a quality inspection model, wherein the quality inspection model includes a video recognition sub-model and an audio recognition sub-model;
[0072] S130, performing detection processing on the recorded data according to the video recognition sub-model and the audio recognition sub-model respectively to obtain a first quality inspection result of the recorded data, wherein the first quality inspection result includes a sound recording quality inspection result and a video recording quality inspection result;
[0073] S140, when at least one of the audio quality inspection result and the video quality inspection result of the first quality inspection result is an abnormal quality inspection result, sending abnormal recording data to the quality inspection background, the abnormal recording data being the recording data corresponding to the abnormal quality inspection result;
[0074] S150: When the quality inspection confirmation result generated by the quality inspection background according to the abnormal recording data is inconsistent with the first quality inspection result, the quality inspection confirmation result is used as the final quality inspection result.
[0075] In this embodiment, by inputting the recorded data to be quality inspected into the quality inspection model, the sub-model in the quality inspection model can detect and process the recorded data and the video data in the recorded data respectively to obtain the first quality inspection result of the recorded data. When the first quality inspection result includes at least one abnormal quality inspection result, the abnormal recorded data can be sent to the quality inspection background to confirm the abnormal recorded data through the quality inspection background. When the quality inspection confirmation result is the same as the abnormal quality inspection result, the first quality inspection result can be used as the final quality inspection result; when the quality inspection confirmation result is inconsistent with the abnormal quality inspection result, the quality inspection confirmation result can be used as the final quality inspection result. The quality inspection model can be used to perform batch automatic quality inspection on a large amount of recorded data. For recorded data with no abnormal quality inspection results, the first quality inspection result can be directly used as the final quality inspection result. For recorded data with abnormalities, a secondary quality inspection confirmation is performed through the quality inspection background to obtain the final quality inspection result. Quality inspection through the quality inspection model can greatly improve the efficiency and accuracy of quality inspection. In addition, the personnel in the quality inspection background only need to conduct quality inspection and confirmation on the recorded data that have been identified as abnormal, which reduces the quality inspection workload and personnel costs of the quality inspection background.
[0076] When handling relevant business for customers, business handling personnel can record audio and video of the handling area to generate recorded data of the business handling process, and the recorded data may include audio data and video data. The recording device can upload the recorded data to the database for storage at the end of the business handling. When the recorded data is quality-checked, the recorded data can be obtained from the database and quality-checked. It should be noted that the acquisition, storage, use, and processing of data in the following embodiments all comply with the relevant provisions of national laws and regulations.
[0077] The following uses the recording data quality inspection device as the execution body to introduce the specific implementation methods of the above steps.
[0078] In S110, when the quality inspection device performs quality inspection on the recorded data, it can obtain the recorded data to be inspected from the database. The recorded data in the database may include quality inspection parameters, and the quality inspection parameters of the recorded data before and after the quality inspection are different. That is, the quality inspection parameters of the recorded data will change after the quality inspection. The quality inspection device can determine whether the recorded data is the recorded data that has been inspected or the recorded data to be inspected based on the quality inspection parameters of a certain recorded data in the database.
[0079] In S120, a quality inspection model is pre-set in the quality inspection device. After the recorded data to be quality inspected is obtained, the recorded data can be input into the quality inspection model so as to perform quality inspection on the recorded data through the quality inspection model. The quality inspection model can include a video recognition sub-model and an audio recognition sub-model. The video recognition sub-model can perform quality inspection on the video data part of the recorded data, and the audio recognition sub-model can perform quality inspection on the audio data part of the recorded data.
[0080] In S130, in the quality inspection model, the video recognition sub-model and the audio recognition sub-model are used to inspect and process the recorded data respectively, and a first quality inspection result of the recorded data can be obtained. The first quality inspection result includes a recording quality inspection result and a video quality inspection result. The recording quality inspection result is a quality inspection result generated after the audio recognition sub-model inspects the recorded data; the video quality inspection result is a quality inspection result generated after the video recognition sub-model inspects the recorded data.
[0081] As an alternative embodiment, please refer to Figure 2 In order to, the above S130 may include:
[0082] S210, determining the video data corresponding to the video recognition sub-model and the audio data corresponding to the audio recognition sub-model from the recorded data to be quality inspected;
[0083] S220, processing the video data according to the video recognition sub-model to obtain a video quality inspection result of the video data;
[0084] S230, processing the recording data according to the audio recognition sub-model to obtain a recording quality inspection result of the recording data;
[0085] S240: Generate the first quality inspection result according to the video quality inspection result and the audio quality inspection result.
[0086] In this embodiment, the two sub-models in the quality inspection model can be used to perform quality inspection on the audio data and the video data in the recorded data, and generate corresponding video quality inspection results and audio quality inspection results, and the first quality inspection result can be generated according to the video quality inspection results and audio quality inspection results. By performing quality inspection on the recorded data to be inspected by the quality inspection model, a large amount of recorded data can be automatically inspected and the corresponding first quality inspection result can be obtained, thereby improving the efficiency of quality inspection.
[0087] In S210, after recording the business handling process, the recording device may store the generated audio data and video data separately in the database, or store them together in the database.
[0088] When the audio data and the video data are stored in a database in a separate storage manner, the quality inspection device can determine the audio data and the video data according to the read position in the database when obtaining the audio data to be inspected. When the audio data and the video data are stored in a database in a combined storage manner, the quality inspection device needs to separate the audio data after obtaining the audio data to be inspected to obtain the audio data and the video data. For example, the quality inspection device can use the FFmpeg (Fast Forward Mpeg) tool to separate the audio data.
[0089] After the quality inspection device determines the video data and the audio data from the recorded data to be inspected, it can correspond the video data to the video recognition sub-model and correspond the audio data to the audio recognition sub-model.
[0090] In S220, the quality inspection device processes the video data through the video recognition sub-model to obtain the video quality inspection result of the video data. The video quality inspection result can be that the video quality inspection passes or fails. When the video recognition sub-model performs video quality inspection on the video data, it can identify and process the video data from multiple processing nodes according to the type of business to determine whether the corresponding multiple processing nodes in the video data meet the quality inspection rules. When the data corresponding to each processing node in the video data meets the quality inspection rules, the video quality inspection result is passed, otherwise the video quality inspection result is failed.
[0091] As an alternative embodiment, please refer to Figure 3 In order to, the above S220 may include:
[0092] S310, determining each service node according to the service type corresponding to the recorded data;
[0093] S320, dividing the video data into a plurality of video sub-data according to each service node, wherein the plurality of video sub-data correspond to each service node one by one;
[0094] S330: Perform video content recognition on the plurality of video sub-data respectively according to the video recognition sub-model to obtain a video quality inspection result.
[0095] In this embodiment, the service type corresponding to the recorded data can be used to determine the service nodes corresponding to the service type. For each service node, corresponding video sub-data can be divided from the recorded data, and video content recognition can be performed on the recorded sub-data to determine whether the video content meets the requirements of the service node. After performing video content recognition on each recorded sub-data, the video quality inspection result of the recorded data can be integrated.
[0096] In S310, after obtaining the recorded data to be inspected, the quality inspection device can determine the corresponding business nodes according to the business type corresponding to the recorded data. For example, when handling the sales business of financial products such as funds, financial management or insurance, the business nodes may include a precaution display node, a certificate display node, a user identity verification node, an information verification reminder node, a customer opinion solicitation node, and a customer signature node.
[0097] In S320, the video data may be divided according to each service node corresponding to the service type, so as to divide the video data into video sub-data corresponding to the service node one by one. By detecting the video data, it may be determined whether the video data includes the video feature corresponding to the service node, and the corresponding video sub-data may be divided from the video data according to the corresponding video feature.
[0098] For example, taking the certificate display node as an example, in the process of detecting the video data, if a video frame containing a business processing personnel showing a qualification certificate to a customer is detected, it means that the video frame contains the video feature corresponding to the certificate display node. Based on the video frame, a time period containing the video frame can be used as the video sub-data corresponding to the certificate display node.
[0099] It is understandable that after determining the video feature from the video frame in the video data, the video data of a period of time before and after the video frame can be segmented as video sub-data according to the video frame. The video duration of the video sub-data can be a preset fixed duration, or a duration set according to the corresponding business node. Since two or more video features may be included at the same time during the business process, or the time interval between two adjacent video features is small, the video sub-data divided according to the video data may overlap with other video sub-data. For example, a certain video sub-data may be the fifth to fifteenth minute of the video data, and another video sub-data may be the tenth to twentieth minute of the video data.
[0100] As an alternative embodiment, please refer to Figure 4 , in order to, the above S320 may include:
[0101] S410, obtaining video features corresponding to each service node;
[0102] S420, determining video frames corresponding to respective video features from the video data;
[0103] S430, dividing the video data according to each video frame to obtain a plurality of video sub-data.
[0104] In this embodiment, after determining the corresponding service node according to the service type, the video features corresponding to each service node can be obtained, and the video frames corresponding to each video feature can be determined from the video data. According to the video frames corresponding to each video feature, the video sub-data can be divided from the video data according to the range before and after the video frame. For example, after determining the first frame and the last frame of the video corresponding to the video feature, the video sub-data can be divided from the video data according to the timestamps of the first frame and the last frame.
[0105] In S410, different video features are respectively corresponding to different service nodes. After determining the corresponding multiple service nodes according to the service type corresponding to the recorded data, the corresponding video features can be respectively determined according to each service node. For example, the video feature can be that a certain video frame includes a specific behavior feature, or that the duration of a certain behavior in continuous video frames reaches a preset duration, etc.
[0106] In S420, by performing image analysis on the video data, the video frames corresponding to each video feature can be determined from the video data. When the video feature is a specific behavior feature, the video frames having the behavior feature can be screened out from the video data. When the duration of a certain behavior with a video feature reaches a preset duration, the first and last video frames of the behavior can be screened out from the image data, and it can be determined whether the duration of the behavior reaches the preset duration based on the timestamps corresponding to the first and last video frames.
[0107] In S430, after determining the video frames corresponding to the respective video features in the video data, the video data may be divided according to the multiple video frames to divide the video data into multiple video sub-data. Each video sub-data includes the video features required by the corresponding service node.
[0108] In S330, the quality inspection device can perform video content recognition on multiple video sub-data respectively through the video recognition sub-model to determine whether the video content in each video sub-data meets the relevant regulations of the business handling process. For example, taking the certificate display node as an example, the video recognition sub-model can perform video content recognition on the video sub-data corresponding to the certificate display node, determine the user's location, user orientation, and the location of the certificate presented by the business handling personnel from the video content, and confirm whether the user can see the certificate presented by the business handling personnel at his position. When it is determined that the user is in a position where he can see the certificate presented by the business handling personnel, it can be determined that the certificate display node meets the process regulations for business handling, otherwise it can be determined that the certificate display node does not meet the process regulations for business handling.
[0109] After the video recognition sub-model recognizes the video content of each video sub-data, it can determine whether the video content recorded in the video sub-data meets the process regulations of the corresponding business node. When the video content recorded in each video sub-data can meet the relevant regulations, the video quality inspection result can be determined to be passed.
[0110] In S230, the quality inspection device processes the recording data through the audio recognition sub-model to obtain a recording quality inspection result of the recording data. The recording quality inspection result can be that the recording quality inspection passes or fails.
[0111] When the audio recognition sub-model performs audio quality inspection on the audio recording data, it can determine the keywords in the business handling process according to the type of business being handled, and convert the audio recording data into text data by performing voice recognition on the audio recording data. By matching the text data with the keywords, the corresponding matching results can be obtained, and the audio quality inspection results can be determined based on the matching results. For example, in the business handling process, it is stipulated that the business handling personnel need to conduct information verification, willingness inquiry, and risk warning processes with the customer, and the user needs to conduct willingness confirmation, risk confirmation and other processes. By matching the text data with the keywords, it can be determined whether the business handling personnel and the user perform the above processes during the business handling process.
[0112] When the matching degree between the text data and the keywords meets the requirements, the recording quality inspection result can be determined to be passed. If the text data and the keywords do not match or the matching degree is low, it can be determined that there is a missing or abnormality in the corresponding process, and the recording quality inspection result is failed.
[0113] As an alternative embodiment, please refer to Figure 5 In order to, the above S230 may include:
[0114] S510, determining a business keyword according to the business type corresponding to the recorded data, wherein the business keyword includes necessary keywords and illegal keywords in the business handling process;
[0115] S520, performing speech recognition processing on the recorded data to obtain speech text information;
[0116] S530, matching the voice text information with the business keyword to obtain a matching result;
[0117] S540: Generate the recording quality inspection result according to the matching result.
[0118] In this embodiment, the corresponding business keywords can be determined by the business type corresponding to the recorded data. After the recorded data is processed by voice recognition, the corresponding voice text information can be obtained, and the voice text information is matched with the business keywords to obtain a matching result. The recording quality inspection result of the recorded data can be generated based on the matching result. The recording quality inspection is realized by voice recognition processing of the quality inspection model, which can avoid the personnel in the quality inspection background from manually listening to the audio content in the recorded data for quality inspection, thereby improving the efficiency and accuracy of quality inspection.
[0119] In S510, the quality inspection device can determine the business keywords according to the business type corresponding to the recorded data. The business keywords can include necessary keywords and illegal keywords in the business handling process. Necessary keywords refer to some words included in the words that the business handling personnel need to read to the user during the business handling process. Illegal keywords refer to some words that the business handling personnel should not say to the user during the handling process. For example, illegal keywords may include capital preservation, risk-free, guaranteed returns or high returns.
[0120] In S520, after determining the service keyword corresponding to the service type, the audio recognition sub-model may perform speech recognition processing on the recording data to convert the recording data into speech text information.
[0121] In S530, by matching the voice text information with the business keywords, a matching result of the voice text information and the business keywords can be obtained. For example, when the business keywords include necessary keywords and illegal keywords, the matching value increases when the voice text information successfully matches the business keywords; the matching value decreases when the voice text information successfully matches the illegal keywords. After the voice text information and the business keywords are matched, the corresponding matching result can be obtained.
[0122] In S540, after the voice-to-text information obtained by voice conversion of the recording data is matched with the business keywords and a matching result is obtained, a recording quality inspection result can be generated according to the matching result. For example, a matching threshold is pre-set in the audio recognition sub-model, and when the value of the matching result is greater than the matching threshold, it can be determined that the recording quality inspection result is passed; when the value of the matching result is less than the matching threshold, it can be determined that the recording quality inspection result is not passed.
[0123] It is understandable that the illegal keywords may also include prohibited keywords. When the voice text information successfully matches the prohibited keywords, the recording quality inspection result can be directly determined as failed.
[0124] In S240, after the video data and the audio data are processed by the video recognition sub-model and the audio recognition sub-model respectively, the video quality inspection result and the audio quality inspection result can be obtained respectively, and the first quality inspection result is generated according to the video quality inspection result and the audio quality inspection result. For example, during the business handling process, the business personnel completed the corresponding specific behavior, but did not inform the user of the risk warning information language when giving the risk warning to the user. Then the first quality inspection result generated is that the video quality inspection result passes, and the audio quality inspection result fails.
[0125] In S140, after the recorded data is detected and processed and a first quality inspection result is obtained, if one or both of the recording quality inspection result and the video quality inspection result in the first quality inspection result are abnormal quality inspection results, it can be determined that the business handling process corresponding to the recorded data is an abnormal business process. The abnormal quality inspection result may include that the recording quality inspection result fails or the video quality inspection result fails.
[0126] The quality inspection device can send abnormal recording data to the quality inspection background so that the quality inspection background can conduct quality inspection and confirmation on the abnormal recording data. The abnormal recording data is the recording data corresponding to the abnormal quality inspection result. After the quality inspection background receives the abnormal recording data, the quality inspection personnel can conduct manual quality inspection on the abnormal recording data and generate the corresponding quality inspection confirmation result.
[0127] It is understandable that when the quality inspection device performs batch quality inspection on the recorded data to be inspected, when it is determined that the first quality inspection result corresponding to a piece of recorded data contains an abnormal quality inspection result, the recorded data can be treated as abnormal recorded data and sent to the quality inspection background. The quality inspection device can also send multiple abnormal recorded data containing abnormal quality inspection results in the first quality inspection result within the preset period to the quality inspection background at every preset period.
[0128] When the quality inspection device sends abnormal recording data to the quality inspection background, it can also send the corresponding first quality inspection result to the quality inspection background, so that the quality inspection personnel of the quality inspection background can determine the abnormal part of the abnormal recording data based on the first quality inspection result, and conduct quality inspection and confirmation on the part of the data to improve the efficiency of quality inspection and confirmation. For example, the quality inspection personnel of the quality inspection background determine that the video quality inspection result is an abnormal quality inspection result based on the first quality inspection result, and the abnormal quality inspection result is that the video content recorded by one or more video sub-data in the video data does not meet the process regulations for business handling. The quality inspection personnel can then confirm the one or more video sub-data and generate the corresponding quality inspection confirmation result.
[0129] In S150, after the quality inspection background receives the abnormal recording data and generates the corresponding quality inspection confirmation result, the quality inspection confirmation result can be sent to the quality inspection device. After the quality inspection device receives the quality inspection confirmation result, it can determine the abnormal recording data corresponding to the quality inspection confirmation result and obtain the first quality inspection result corresponding to the abnormal recording data. When the quality inspection confirmation result is inconsistent with the first quality inspection result, the quality inspection device can use the quality inspection confirmation result made by the quality inspection background as the final quality inspection result.
[0130] After the quality inspection device performs quality inspection on the recorded data to be inspected through the quality inspection model and obtains the first abnormal result, it can automatically perform quality inspection on the massive amount of recorded data, and screen the first quality inspection results containing abnormal quality inspection results from multiple first quality inspection results. For the screened first quality inspection results with abnormal quality inspection results, the quality inspection device can send the corresponding abnormal recorded data to the quality inspection background, so that the abnormal recorded data can be quality inspected and confirmed by the quality inspectors, and the results of the quality inspection confirmation can be used as the final quality inspection results. The quality inspectors do not need to perform quality inspection on massive amounts of recorded data, but only need to perform secondary quality inspection and confirmation on the abnormal recorded data with abnormalities screened out by the quality inspection model, which greatly improves the quality inspection efficiency and accuracy of the recorded data, reduces the workload of the quality inspectors, and thus reduces the corresponding personnel costs.
[0131] As an alternative embodiment, please refer to Figure 6 , in order to, after the above S140, further include:
[0132] S610, obtaining a plurality of recording data whose quality inspection confirmation result is inconsistent with the first quality inspection result;
[0133] S620, sorting and counting the multiple recording data to determine the problematic service nodes corresponding to the multiple recording data respectively;
[0134] S630: According to the problematic service node, adjust and optimize the model parameters of the quality inspection model to update the quality inspection model.
[0135] In this embodiment, after a plurality of recording data to be quality inspected are quality inspected by the quality inspection model, the first quality inspection result and the quality inspection confirmation result of each recording data can be obtained. For the recording data whose first quality inspection result and the quality inspection confirmation result are inconsistent, they can be sorted, counted and analyzed to determine the problem business nodes where the first quality inspection result and the quality inspection confirmation result are inconsistent, and the model parameters of the quality inspection model can be adjusted and optimized according to these problem business nodes to update the quality inspection model and improve the quality inspection accuracy of the quality inspection model.
[0136] In S610, after receiving the quality inspection confirmation result returned by the quality inspection background, if the quality inspection confirmation result is inconsistent with the first quality inspection result, the quality inspection confirmation result is used as the final quality inspection result. After quality inspection of multiple recorded data to be quality inspected, the quality inspection device can determine multiple recorded data whose quality inspection confirmation results are inconsistent with the first quality inspection result from the recorded data that have been quality inspected.
[0137] In S620, after determining a plurality of recorded data whose quality inspection confirmation results are inconsistent with the first quality inspection result, the quality inspection model may sort and count the plurality of recorded data, and determine the problematic service nodes respectively corresponding to the plurality of recorded data.
[0138] It is understandable that, for a certain recording data, the first quality inspection result may include anomalies of multiple business nodes, and in the quality inspection confirmation process for multiple business nodes in the quality inspection confirmation result, the quality inspection results of some of the multiple business nodes are different from the first quality inspection results. That is, when the first quality inspection result and the quality inspection confirmation result are inconsistent, there may be inconsistencies in the quality inspection results of some business nodes, and the quality inspection results of other business nodes are consistent. In this case, the problem business node corresponding to the recording data is the business node whose quality inspection confirmation result is different from the first quality inspection result. For business nodes whose quality inspection confirmation results are the same as the first quality inspection results, they may not be considered as problem business nodes.
[0139] In S630, after determining the problem business node corresponding to each recorded data, the reason why the quality inspection confirmation result is inconsistent with the first quality inspection result can be determined based on the problem business node, and the model parameters of the quality inspection model can be adjusted and optimized based on the determined problem cause to update the quality inspection model and improve the quality inspection efficiency and quality inspection accuracy of the quality inspection model.
[0140] Taking the problem business node as the node for soliciting customer opinions as an example, when the quality inspection model performs video content recognition on the video sub-data and fails to recognize the behavioral feature of the user nodding or gesturing in agreement, the first quality inspection result obtained by the quality inspection model is a failed video quality inspection result. When the quality inspection background performs quality inspection confirmation on the recorded data, it is found that the user has expressed his agreement through voice, and the video quality inspection result in the quality inspection confirmation result is a passed result. According to the above-mentioned cause of the problem, the quality inspection model can be adjusted. When the video content recognition fails to identify the corresponding behavioral feature, the recorded sub-data in the time period corresponding to the video sub-data is determined from the recorded data, and the voice text information corresponding to the recorded sub-data is used to determine whether there is a voice feature corresponding to the behavioral feature. If there is a corresponding voice feature, it is determined that the business node meets the regulations for business handling.
[0141] As an alternative embodiment, please refer to Figure 7 , in order to, the above S630, may include:
[0142] S710, adjusting the model parameters according to the problematic service node to obtain an adjusted quality inspection model;
[0143] S720, inputting a plurality of recording data into the adjusted quality inspection model to obtain second quality inspection results respectively corresponding to the plurality of recording data;
[0144] S730, calculating a first matching degree between the quality inspection confirmation results of the plurality of recording data and the plurality of first quality inspection results, and a second matching degree between the quality inspection confirmation results of the plurality of recording data and the plurality of second quality inspection results;
[0145] S740, determining an optimized quality inspection result from the first quality inspection result and the second quality inspection result according to the first matching degree and the second matching degree;
[0146] S750: Use the quality inspection model corresponding to the optimized quality inspection result as the updated quality inspection model.
[0147] In this embodiment, after determining the problem business node where the first quality inspection result and the quality inspection confirmation result are inconsistent, the model parameters can be adjusted according to the problem business node to obtain an adjusted quality inspection model. By performing quality inspection on the same batch of recorded data using the adjusted quality inspection model, multiple second quality inspection results can be obtained. According to the first matching degree of the first quality inspection result and the quality inspection confirmation result, the difference between the quality inspection model before adjustment and the quality inspection performed by the quality inspection background can be determined; according to the second matching degree of the second quality inspection result and the quality inspection confirmation result, the difference between the quality inspection model after adjustment and the quality inspection performed by the quality inspection background can be determined. When the second matching degree corresponding to the adjusted quality inspection model is higher, it means that adjusting the model parameters can improve the quality inspection accuracy of the quality inspection model, and the adjusted quality inspection model can be used as the updated model. Otherwise, the quality inspection model before adjustment is still used.
[0148] In S710, the quality inspection device may adjust the model parameters of the quality inspection model according to the determined problematic service nodes to obtain an adjusted quality inspection model.
[0149] In S720, the quality inspection device may input a plurality of recording data into the adjusted quality inspection model, so as to generate corresponding second quality inspection results according to the plurality of recording data respectively through the adjusted quality inspection model.
[0150] In S730, after determining the second quality inspection results respectively corresponding to the plurality of recording data, the quality inspection device may determine the first quality inspection results and the quality inspection confirmation results respectively corresponding to the plurality of recording data.
[0151] The quality inspection device can calculate a first matching degree between the quality inspection confirmation results of the plurality of recording data and the first quality inspection confirmation result, and calculate a second matching degree between the quality inspection confirmation results of the plurality of recording data and the second quality inspection confirmation result.
[0152] In S740, the quality inspection device may determine the one with a higher matching degree from the first matching degree and the second matching degree, and determine the quality inspection result corresponding to the higher matching degree as the optimized quality inspection result. For example, when the first matching degree is higher than the second matching degree, the first quality inspection result may be determined as the optimized quality inspection result; otherwise, the second quality inspection result may be determined as the optimized quality inspection result.
[0153] It is understandable that for multiple recording data, the first quality inspection result of a part of the recording data is inconsistent with the quality inspection confirmation result, and the second quality inspection result of another part of the recording data is inconsistent with the quality inspection confirmation result. When these two parts of the recording data are completely consistent, it means that the quality inspection result generated by the quality inspection model has not changed before and after the adjustment of the model parameters. If only part of the data in the two parts of the recording data overlaps or there is no overlapping part, it means that the quality inspection result generated by the quality inspection model has changed by adjusting the model parameters. At this time, the quality inspection model with higher quality inspection accuracy can be determined based on the differences between the first quality inspection result and the second quality inspection result and the quality inspection confirmation result.
[0154] In S750, after the optimized quality inspection result is determined from the first quality inspection result and the second quality inspection result, the quality inspection model corresponding to the optimized quality inspection result may be used as the updated quality inspection model.
[0155] Based on the recording data quality inspection method provided in the above embodiment, the present application also provides a specific implementation of the recording data quality inspection device. Please refer to the following embodiment.
[0156] See also Figure 8 The recording data quality inspection device 800 provided in the embodiment of the present application includes the following modules:
[0157] Acquisition module 801, acquiring recorded data to be quality inspected;
[0158] An input module 802 inputs the recorded data to be quality inspected into a quality inspection model, wherein the quality inspection model includes a video recognition sub-model and an audio recognition sub-model;
[0159] The detection module 803 performs detection processing on the recorded data according to the video recognition sub-model and the audio recognition sub-model to obtain a first quality inspection result of the recorded data;
[0160] An abnormality module 804 sends abnormal quality inspection information to a quality inspection backend when the first quality inspection result is an abnormal quality inspection result;
[0161] The confirmation module 805 receives the quality inspection confirmation result sent by the quality inspection background, and replaces the abnormal quality inspection result with the quality inspection confirmation result.
[0162] By inputting the recorded data to be quality inspected into the quality inspection model, the sub-models in the quality inspection model can be used to detect and process the recorded data and the video data in the recorded data respectively to obtain the first quality inspection result of the recorded data. When the first quality inspection result includes at least one abnormal quality inspection result, the abnormal recorded data can be sent to the quality inspection background to confirm the abnormal recorded data through the quality inspection background. When the quality inspection confirmation result is the same as the abnormal quality inspection result, the first quality inspection result can be used as the final quality inspection result; when the quality inspection confirmation result is inconsistent with the abnormal quality inspection result, the quality inspection confirmation result can be used as the final quality inspection result. The quality inspection model can be used to perform batch automatic quality inspection on a large amount of recorded data. For recorded data with no abnormal quality inspection results, the first quality inspection result can be directly used as the final quality inspection result. For recorded data with abnormalities, a secondary quality inspection confirmation is performed through the quality inspection background to obtain the final quality inspection result. Quality inspection through the quality inspection model can greatly improve the efficiency and accuracy of quality inspection. In addition, the personnel in the quality inspection background only need to conduct quality inspection and confirmation on the recorded data that have been identified as abnormal, which reduces the quality inspection workload and personnel costs of the quality inspection background.
[0163] The recording data quality inspection device provided by the embodiment of the present invention can achieve Figures 1 to 7 To avoid repetition, the various processes in the method embodiment are not described here.
[0164] Fig. 9 A schematic diagram of the hardware structure of the recording data quality inspection device provided in an embodiment of the present application is shown.
[0165] The recorded data quality inspection device may include a processor 901 and a memory 902 storing computer program instructions.
[0166] Specifically, the processor 901 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.
[0167] The memory 902 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 902 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 902 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 902 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 902 is a non-volatile solid-state memory.
[0168] The memory may include read-only memory (ROM), random access memory (RAM), magnetic disk storage media devices, optical storage media devices, flash memory devices, electrical, optical or other physical / tangible memory storage devices. Thus, typically, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to an aspect of the present disclosure.
[0169] The processor 901 implements any one of the recording data quality inspection methods in the above embodiments by reading and executing computer program instructions stored in the memory 902 .
[0170] In one example, the recording data quality inspection device may further include a communication interface 903 and a bus 910. Fig. 9 As shown, the processor 901, the memory 902, and the communication interface 903 are connected via a bus 910 and communicate with each other.
[0171] The communication interface 903 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.
[0172] Bus 910 includes hardware, software or both, and the parts of recording data quality inspection equipment are coupled to each other.For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 910 may include one or more buses. Although the present application embodiment describes and shows a specific bus, the application considers any suitable bus or interconnection.
[0173] The recording data quality inspection device can be based on, thereby achieving the combination of Figures 1 to 8 The described recording data quality inspection method and device.
[0174] In addition, in combination with the recording data quality inspection method in the above embodiment, the embodiment of the present application can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by the processor, any one of the recording data quality inspection methods in the above embodiment is implemented.
[0175] In addition, an embodiment of the present application further provides a computer program product, including computer program instructions, which can implement the steps and corresponding contents of the aforementioned method embodiment when the computer program instructions are executed by a processor.
[0176] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.
[0177] The functional blocks shown in the above block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.
[0178] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.
[0179] Aspects of the present disclosure are described above with reference to the flowchart and / or block diagram of the method, device and computer program product according to the embodiment of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.
[0180] The above are only specific implementation methods of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.
Claims
1. A recording data quality inspection method, wherein the recording data includes audio data and video data, characterized in that: The method comprises: Obtain the recorded data to be inspected; Inputting the recorded data to be quality inspected into a quality inspection model, wherein the quality inspection model includes a video recognition sub-model and an audio recognition sub-model; Performing detection and processing on the recorded data according to the video recognition sub-model and the audio recognition sub-model respectively to obtain a first quality inspection result of the recorded data, wherein the first quality inspection result includes a sound recording quality inspection result and a video recording quality inspection result; When at least one of the audio quality inspection result and the video quality inspection result of the first quality inspection result is an abnormal quality inspection result, abnormal recording data is sent to the quality inspection background, and the abnormal recording data is the recording data corresponding to the abnormal quality inspection result; When the quality inspection confirmation result generated by the quality inspection background according to the abnormal recording data is inconsistent with the first quality inspection result, the quality inspection confirmation result is used as the final quality inspection result; Acquire a plurality of recording data whose quality inspection confirmation results are inconsistent with the first quality inspection results; Arranging and counting the multiple recording data, and determining the problematic service nodes corresponding to the multiple recording data respectively; According to the problem business node, adjusting and optimizing the model parameters of the quality inspection model to update the quality inspection model; The adjusting and optimizing the model parameters of the quality inspection model according to the problem business node to update the quality inspection model includes: Adjusting the model parameters according to the problem business node to obtain an adjusted quality inspection model; Inputting a plurality of recording data into the adjusted quality inspection model to obtain second quality inspection results respectively corresponding to the plurality of recording data; For multiple recording data, when a first quality inspection result of a part of the recording data is inconsistent with the quality inspection confirmation result, and a second quality inspection result of another part of the recording data is inconsistent with the quality inspection confirmation result, if the two parts of the recording data only partially overlap or do not overlap, the steps of: calculating a first matching degree between the quality inspection confirmation results of the multiple recording data and the multiple first quality inspection results, and a second matching degree between the quality inspection confirmation results of the multiple recording data and the multiple second quality inspection results; Determining an optimized quality inspection result from the first quality inspection result and the second quality inspection result according to the first matching degree and the second matching degree; The quality inspection model corresponding to the optimized quality inspection result is used as the updated quality inspection model.
2. The recording data quality inspection method according to claim 1, characterized in that: The detecting and processing the recorded data according to the video recognition sub-model and the audio recognition sub-model to obtain a first quality inspection result of the recorded data includes: Determine the video data corresponding to the video recognition sub-model and the audio data corresponding to the audio recognition sub-model from the recorded data to be quality inspected; Processing the video data according to the video recognition sub-model to obtain a video quality inspection result of the video data; Processing the recording data according to the audio recognition sub-model to obtain a recording quality inspection result of the recording data; The first quality inspection result is generated according to the video quality inspection result and the audio quality inspection result.
3. The recording data quality inspection method according to claim 2, characterized in that: The step of processing the video data according to the video recognition sub-model to obtain a video quality inspection result of the video data includes: Determine each service node according to the service type corresponding to the recorded data; Dividing the video data into a plurality of video sub-data according to each service node, wherein the plurality of video sub-data correspond to each service node one by one; The video content is identified on the multiple video sub-data respectively according to the video identification sub-model to obtain a video quality inspection result.
4. The recording data quality inspection method according to claim 3, characterized in that: The video data is divided into a plurality of video sub-data according to each service node, including: Obtain the video features corresponding to each business node; Determine the video frames corresponding to the respective video features from the video data; The video data is divided according to each video frame to obtain a plurality of video sub-data.
5. The recording data quality inspection method according to claim 2, characterized in that: The processing of the recording data according to the audio recognition sub-model to obtain a recording quality inspection result of the recording data includes: Determine a business keyword according to the business type corresponding to the recorded data, wherein the business keyword includes necessary keywords and illegal keywords in the business handling process; Performing speech recognition processing on the recorded data to obtain speech text information; Matching the voice text information with the business keywords to obtain a matching result; The recording quality inspection result is generated according to the matching result.
6. A recording data quality inspection device, characterized in that: The recording data quality inspection device comprises: Acquisition module, to obtain the recorded data to be inspected; An input module, inputting the recorded data to be quality inspected into a quality inspection model, wherein the quality inspection model includes a video recognition sub-model and an audio recognition sub-model; A detection module, which performs detection processing on the recorded data according to the video recognition sub-model and the audio recognition sub-model to obtain a first quality inspection result of the recorded data; An abnormality module, when the first quality inspection result is an abnormal quality inspection result, sends quality inspection abnormality information to the quality inspection background; A confirmation module receives the quality inspection confirmation result sent by the quality inspection background, and replaces the abnormal quality inspection result with the quality inspection confirmation result; The recorded data quality inspection device is further used to obtain multiple recorded data whose quality inspection confirmation results are inconsistent with the first quality inspection results; organize and count the multiple recorded data to determine the problem business nodes corresponding to the multiple recorded data; and adjust and optimize the model parameters of the quality inspection model according to the problem business nodes to update the quality inspection model; The recorded data quality inspection device is specifically used to adjust the model parameters according to the problem business node to obtain an adjusted quality inspection model; input multiple recorded data into the adjusted quality inspection model to obtain second quality inspection results corresponding to the multiple recorded data respectively; for multiple recorded data, when the first quality inspection result of a part of the recorded data is inconsistent with the quality inspection confirmation result, and the second quality inspection result of another part of the recorded data is inconsistent with the quality inspection confirmation result, if the two parts of the recorded data only partially overlap or do not overlap, then execute the steps: calculate the first matching degree between the quality inspection confirmation results of the multiple recorded data and the multiple first quality inspection results, and the second matching degree between the quality inspection confirmation results of the multiple recorded data and the multiple second quality inspection results; determine the optimized quality inspection result from the first quality inspection result and the second quality inspection result according to the first matching degree and the second matching degree; and use the quality inspection model corresponding to the optimized quality inspection result as the updated quality inspection model.
7. A recording data quality inspection device, characterized in that: The recorded data quality inspection device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the recording data quality inspection method according to any one of claims 1 to 5 is implemented.
8. A computer storage medium, characterized in that: The computer storage medium stores computer program instructions, and when the computer program instructions are executed by the processor, the recording data quality inspection method according to any one of claims 1 to 5 is implemented.
9. A computer program product, characterized in that The computer program product includes computer program instructions, and when the computer program instructions are executed by a processor, the recording data quality inspection method according to any one of claims 1 to 5 is implemented.
