Method, device and equipment for carrying out quality inspection on call recording based on large model

Through the automatic quality inspection method based on large models, the problem of high and untimely inspection of call recording quality is solved, efficient and accurate quality inspection results are achieved, and the company's operational efficiency and customer satisfaction are improved.

CN118982292APending Publication Date: 2024-11-19BEIJING RENKE INTERACTIVE NETWORK TECH CO LTD
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411142805.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-20
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

The existing call recording quality inspection technology relies on manual review, which leads to high costs and untimely impacting the operational efficiency of the company.

Method used

Automatic quality inspection method based on large models is adopted to generate quality inspection results by obtaining call recording files, historical quality inspection cases and preset quality inspection standards, reducing manual intervention and improving quality inspection efficiency and accuracy.

Benefits of technology

It reduces the cost of quality inspection, improves the efficiency and accuracy of quality inspection, achieves timely feedback on quality inspection results, and improves the quality of customer service.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118982292A_ABST
    Figure CN118982292A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of artificial intelligence, and provides a method, device and equipment for carrying out quality inspection on call recording based on a large model, and the method comprises the steps: obtaining a call recording file, and converting the call recording file into text data to be subjected to quality inspection; obtaining a historical quality inspection case, and determining a case prompt word according to the historical quality inspection case; and performing quality inspection on the text data to be subjected to quality inspection based on the case cue word through the large model, and generating a quality inspection result according to a preset quality inspection standard. According to the invention, the defects of high manual quality inspection cost and untimely quality inspection in the prior art are overcome, the quality inspection cost is reduced, and the quality inspection efficiency is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method, device and equipment for quality inspection of call recordings based on a large model. Background Art

[0002] In today's business environment, telephone sales (telemarketing) and 400 hotlines are important channels for C-end enterprises (enterprises that provide products or services directly to consumers) to directly connect with customers. Their importance is self-evident. Telemarketing has brought a large number of business opportunities to enterprises through the characteristics of real-time communication, thereby effectively increasing sales. At the same time, as a window for customer feedback and after-sales service, the 400 hotline has greatly improved the customer's communication experience and enhanced customer satisfaction. However, with the surge in the number of telemarketing and after-sales calls, how to ensure the quality of communication has become a major challenge facing enterprises.

[0003] At present, there are many call recording quality inspection tools on the market. These tools mainly rely on recording-to-text technology. After the recording files are converted into text, they are manually reviewed by quality inspectors. Although this method has improved the efficiency of quality inspection to a certain extent, it still has many limitations. First, manual review requires a lot of human resources, resulting in high quality inspection costs. Second, due to the lag of manual review, quality inspection data often cannot be fed back in time, affecting the operational efficiency of the enterprise.

[0004] The existing call recording quality inspection technology is costly and has poor timeliness. Massive call recordings require a large quality inspection team to conduct manual review, which increases the company's operating costs and reduces the overall quality of customer service. It is an important issue that needs to be urgently addressed in the industry. Summary of the invention

[0005] The present invention provides a method, device and equipment for quality inspection of call recordings based on a large model, so as to solve the defects of high cost and untimely manual quality inspection in the prior art, reduce the quality inspection cost and improve the quality inspection efficiency.

[0006] The present invention provides a method for quality inspection of call recordings based on a large model, comprising the following steps: Obtaining a call recording file, and converting the call recording file into text data to be inspected; Obtain historical quality inspection cases, and determine case prompt words according to the historical quality inspection cases; The text data to be inspected is quality inspected based on the case prompt words through the large model, and a quality inspection result is generated according to a preset quality inspection standard.

[0007] According to a method for quality inspection of call recordings based on a large model provided by the present invention, the call recording file is obtained and the call recording file is converted into text data to be quality inspected, which specifically includes: obtaining a real-time call recording or obtaining an existing call recording from a historical record to form a call recording file; determining an automatic speech recognition tool and configuring the parameters of the automatic speech recognition tool; uploading the call recording file to a server or application programming interface of the configured automatic speech recognition tool and initiating a conversion request; receiving the converted raw text data, and preprocessing the raw text data to obtain the text data to be quality inspected.

[0008] According to a method for quality inspection of call recordings based on a big model provided by the present invention, the obtaining of historical quality inspection cases and determining case prompt words based on the historical quality inspection cases specifically include: screening out historical quality inspection cases related to the text data to be quality inspected from a case library; analyzing the historical quality inspection cases and constructing a big model prompt word framework based on the analysis results; filling the big model prompt word framework based on the text data to be quality inspected to determine the big model prompt words.

[0009] According to a method for quality inspection of call recordings based on a big model provided by the present invention, the big model is used to quality inspect the text data to be quality inspected based on the case prompt words, and a quality inspection result is generated according to a preset quality inspection standard. The method specifically includes: integrating the case prompt words, the text data to be quality inspected, and the preset quality inspection standard into a prompt word as a quality inspection request; sending the quality inspection request as input to the big model for processing, and obtaining an output result after the big model processes the quality inspection request; parsing the output result to extract key information; and generating a quality inspection result in JSON format according to the preset quality inspection standard and the key information.

[0010] According to a method for quality inspection of call recordings based on a large model provided by the present invention, the raw text data is preprocessed to obtain text data to be quality inspected, specifically including: performing text cleaning on the raw text data to remove noise, irrelevant characters or special characters in the raw text data; after text cleaning, performing text segmentation on the raw text data to determine conversation turns; performing text tagging on the segmented data to determine the identity of the interlocutor; and using the data after text cleaning, text segmentation and text tagging as the text data to be quality inspected.

[0011] According to a method for quality inspection of call recordings based on a large model provided by the present invention, after generating the quality inspection results, the method also includes: calling the application programming interface of the business system to convert the quality inspection results in JSON format into business data in the business system.

[0012] The present invention also provides a device for quality inspection of call recordings based on a large model, comprising the following modules: A module for acquiring text data to be inspected, used to acquire call recording files and convert the call recording files into text data to be inspected; A case prompt word determination module is used to obtain historical quality inspection cases and determine case prompt words according to the historical quality inspection cases; The large model quality inspection module is used to perform quality inspection on the text data to be inspected based on the case prompt words, and generate quality inspection results according to preset quality inspection standards.

[0013] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for quality inspection of call recordings based on a large model as described in any one of the above is implemented.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for quality inspection of call recordings based on a large model as described in any of the above is implemented.

[0015] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for quality inspection of call recordings based on a large model.

[0016] The present invention provides a method, device and equipment for quality inspection of call recordings based on a large model. The large model is used to automatically inspect the text data to be inspected, thereby improving the efficiency and accuracy of quality inspection and reducing the cost and time of manual quality inspection. The call recording files are converted into text data, which is convenient for the large model to process and analyze, and also makes the quality inspection process more intuitive and easy to manage. Case prompt words are determined through historical quality inspection cases. These prompt words can guide the large model to more accurately identify problems in the text to be inspected, thereby improving the accuracy of quality inspection. The quality inspection method based on the large model can improve the accuracy and efficiency of quality inspection by continuously learning and optimizing the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 It is a flowchart of the method for quality inspection of call recordings based on a large model provided by the present invention.

[0019] Figure 2 It is a structural schematic diagram of the device for quality inspection of call recordings based on a large model provided by the present invention.

[0020] Figure 3 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] Some terms involved in the present invention are explained below.

[0023] Large Model: Foundation Model in the field of artificial intelligence usually refers to a type of widely used basic model (or infrastructure model), which is a general and highly versatile deep learning model trained on the basis of massive data and computing resources. These models are widely used in various tasks in the fields of natural language processing, computer vision, speech recognition, etc. Foundation Models are usually developed by large technology companies, research institutions or open source communities. The purpose of these models is to provide a shared infrastructure to provide better opportunities and resources for a wider range of applications and developers. The training of these models requires a lot of data and computing resources, so only some large institutions or enterprises can undertake such work. In the field of natural language processing, the most famous Foundation Model is the GPT (Generative Pre-trained Transformer) series of models developed by OpenAI. The GPT model adopts a pre-training and fine-tuning method. The model trained by a large-scale corpus can perform well in a variety of natural language processing (NLP) tasks, such as text classification, machine translation, summary generation, etc.

[0024] Prompt: As the name implies, prompt means "prompt". Prompt is simply an instruction to AI. Prompt can be a piece of text, such as a conversation between you and ChatGPT, or it can be a parameter description in a certain format, such as AI drawing software, which uses parameters more often.

[0025] LangChain: is an open source application development framework that currently supports two programming languages: Python and TypeScript. It gives LLM two core capabilities: data awareness, connecting the language model with other data sources; and proxy capabilities, allowing the language model to interact with its environment.

[0026] JSON (JavaScript Object Notation): is a lightweight data exchange format based on a subset of ECMAScript (js specification developed by the European Computer Association) and uses a text format that is completely independent of programming languages ​​to store and represent data. The simple and clear hierarchical structure makes JSON an ideal data exchange language. It is easy for people to read and write, and it is also easy for machines to parse and generate, and it effectively improves network transmission efficiency. JSON data has two main structures: objects (called "objects" in JSON, and may be called "dictionaries", "hash tables" or "associative arrays" in other programming languages) and arrays.

[0027] Combine the following Figure 1-Figure 3 Embodiments of the present invention are described.

[0028] For C-end enterprises such as e-commerce, education, and finance, telephone sales and 400 hotlines are very important channels for direct connection with customers. Although policies have been suppressing marketing calls, the real-time communication characteristics of telephone sales (telemarketing) can still bring huge business opportunities to enterprises, thereby increasing sales. When customers encounter problems during the purchase and use process, they can communicate with the manufacturer's after-sales department through the 400 hotline, which can also greatly improve the communication experience and increase customer satisfaction.

[0029] However, there are a large number of telemarketing and after-sales calls. In order to ensure the quality of communication, companies often hire a large number of customer service personnel who can start working after a short training. As a result, there are problems such as non-compliance with the Standard Operating Procedure (SOP) and incorrect response to the words. In order to better measure and improve the service quality of customer service, companies often introduce recording quality inspection. However, traditional quality inspection requires a lot of manpower and has defects such as high cost, delay, and poor scoring consistency.

[0030] 1. High cost: A telemarketer often makes 200 calls a day, and the number of 400 hotlines is even higher. Massive call recordings require a large number of quality inspection personnel.

[0031] 2: Untimely: Since it takes time for quality inspectors to be involved, quality inspection data is often delayed for several days or a week, which will greatly affect operational efficiency.

[0032] 3. Poor scoring consistency: Manual quality inspection requires high ability and experience of quality inspectors, and short-term training is often difficult to provide. This leads to large deviations in the evaluation standards of different quality inspectors, making it difficult to quantify the assessment.

[0033] If call recordings can be automatically inspected based on big model technology, it will greatly reduce the cost of enterprise quality inspection and improve the quality of operation. Therefore, the present invention develops a method for automated quality inspection of call recordings based on big model technology, which becomes an effective way to solve the above problems and improve the efficiency and accuracy of quality inspection. This method uses the powerful capabilities of big models in the field of natural language processing to automatically inspect recordings, which not only reduces the cost of quality inspection, but also improves the timeliness and consistency of quality inspection, providing enterprises with a more efficient and accurate quality inspection solution.

[0034] Figure 1 is one of the flow charts of the method for quality inspection of call recordings based on a large model provided by the present invention, such as Figure 1 As shown, the method includes the following: S110: Obtain a call recording file, and convert the call recording file into text data to be inspected.

[0035] According to a method for quality inspection of call recordings based on a large model provided by the present invention, a call recording file is obtained, and the call recording file is converted into text data to be quality inspected, which specifically includes: obtaining a real-time call recording or obtaining an existing call recording from a historical record to form a call recording file; determining an automatic speech recognition tool, and configuring the parameters of the automatic speech recognition tool; uploading the call recording file to the server or application programming interface of the configured automatic speech recognition tool, and initiating a conversion request; receiving the converted original text data, and preprocessing the original text data to obtain the text data to be quality inspected.

[0036] Specifically, if it is a real-time call, the call is recorded with high quality through special recording equipment or software, and the clarity and accuracy of the recording are ensured; if it is a historical call recording, the required call recording file is retrieved and downloaded from the storage system.

[0037] Traditional Automatic Speech Recognition (ASR) products (speech-to-text) can convert recording files into text, which is convenient for large-scale model processing. According to the language and content characteristics of the call recording, select an efficient and accurate automatic speech recognition tool, and configure the corresponding parameters according to the requirements of the tool, such as recognition language, audio format, sampling rate, etc., to ensure the accuracy of the conversion.

[0038] Upload the call recording file to the server of the configured automatic speech recognition tool through a secure network connection or use the application programming interface (API) provided by it to initiate a conversion request, and wait for the automatic speech recognition tool to complete the audio-to-text conversion.

[0039] According to a method for quality inspection of call recordings based on a large model provided by the present invention, raw text data is preprocessed to obtain text data to be inspected, specifically including: performing text cleaning on the raw text data to remove noise, irrelevant characters or special characters in the raw text data; after text cleaning, performing text segmentation on the raw text data to determine conversation turns; performing text tagging on the segmented data to determine the identity of the interlocutor; and using the data after text cleaning, text segmentation and text tagging as the text data to be inspected.

[0040] Specifically, once the conversion is completed, the system receives the raw text data returned by the automatic speech recognition tool and performs text cleaning on the raw text data, including removing repeated, erroneous or unrecognizable characters, and irrelevant noise text; the cleaned text is segmented, and the conversation rounds are reasonably divided according to the pauses and turns of the conversation for subsequent analysis; the segmented text is marked to clearly identify the identity of the interlocutor of each paragraph of text, so as to facilitate the understanding of the conversation flow and the interaction between the participants.

[0041] Integrate the cleaned, segmented and labeled text data to form structured text data to be inspected. Store the text data to be inspected for use in subsequent quality inspection processes.

[0042] By converting call recordings into text using automated tools, the time cost of manual listening and transcription is greatly reduced, thereby improving the efficiency of quality inspection. Compared with manual transcription, the use of professional automatic speech recognition tools for conversion can more accurately capture the content of the call and reduce human errors. Through text cleaning, segmentation and tagging, the generated text data for quality inspection has a clear structure, which is easy to perform subsequent data analysis and quality inspection.

[0043] S120: Obtain historical quality inspection cases, and determine case prompt words according to the historical quality inspection cases.

[0044] According to a method for quality inspection of call recordings based on a big model provided by the present invention, historical quality inspection cases are obtained, and case prompt words are determined according to the historical quality inspection cases, which specifically includes: according to the text data to be quality inspected, historical quality inspection cases related to it are screened from the case library; the historical quality inspection cases are analyzed, and a big model prompt word framework is constructed according to the analysis results; the big model prompt word framework is filled according to the text data to be quality inspected, and the big model prompt words are determined.

[0045] Furthermore, the present invention can manage and optimize prompts. Prompts are the interface language for humans to communicate with large models. To facilitate the management of prompts, a clear directory structure or database can be created for each type of prompt to facilitate retrieval and update. In addition to managing prompts, the present invention will also optimize and precompile the prompts entered by users to better ensure the execution effect.

[0046] Due to the limitation of professionalism, the prompts directly entered by users often cannot be well executed in the big model. However, enterprises often already have a large number of manual quality inspection cases and quality inspection scoring rules. If these quality inspection cases and rule data can be used as background context to supplement the prompts of the big model, it can greatly improve the accuracy and consistency of the quality inspection of the big model and improve the quality of quality inspection. The present invention provides the operation manager with the ability of case management, and can select recommended cases (historical quality inspection cases) from the existing quality inspection library, which will be provided as context to the prompts of the big model.

[0047] Specifically, according to the subject, content or keywords of the text data to be inspected, relevant historical quality inspection cases are retrieved from the case library. The case library can be classified by industry, problem type, service scenario, etc., for more accurate matching and screening.

[0048] Conduct in-depth analysis of the selected historical quality inspection cases to identify key issues, violations, risk points and solutions. Based on the analysis results, extract the key elements and judgment logic in the quality inspection process. Based on the analysis results of historical quality inspection cases, build a general big model prompt word framework. The framework contains key prompt words used to guide the big model to focus on specific quality inspection points, judgment logic and potential problems.

[0049] Fill relevant keywords, phrases or sentences into the prompt word frame according to the specific content of the text data to be inspected. Ensure that the filled prompt words can accurately reflect the core information and potential risk points of the text data to be inspected.

[0050] After filling in the prompt word framework, the prompt words are reviewed and optimized to ensure their accuracy and effectiveness. The finalized large model prompt words will be used to guide the large model to conduct in-depth analysis and judgment of the quality inspection text data.

[0051] By using historical quality inspection cases and big model prompt words, key issues and risk points in the text data to be inspected can be more accurately identified, thereby improving the accuracy of quality inspection. With the rapid processing capabilities of the big model and combined with targeted prompt words, the text data to be inspected can be quickly scanned and analyzed to improve the efficiency of quality inspection. By learning and analyzing historical quality inspection cases, the constructed big model prompt words can better identify potential risks and provide strong support for timely warning and response. By building a general big model prompt word framework, the standardization and normalization of the quality inspection process can be promoted, and the consistency and comparability of quality inspection work can be improved.

[0052] S130. Perform quality inspection on the text data to be inspected based on the case prompt words through the big model, and generate quality inspection results according to the preset quality inspection standards.

[0053] According to a method for quality inspection of call recordings based on a big model provided by the present invention, quality inspection is performed on text data to be inspected based on case prompt words through the big model, and quality inspection results are generated according to preset quality inspection standards. The method specifically includes: integrating the case prompt words, the text data to be inspected and the preset quality inspection standards into a prompt word as a quality inspection request; sending the quality inspection request as input to the big model for processing, and obtaining the output result of the big model after processing the quality inspection request; parsing the output result to extract key information; and generating quality inspection results in JSON format according to the preset quality inspection standards and key information.

[0054] Furthermore, the present invention implements the encapsulation and management of multiple large models based on LangChain, shields the differences between large models, and provides a unified API for upper-level business modules. It is more flexible in the face of the flourishing large model market and can better meet domestic business needs (many foreign large models are not open to domestic users).

[0055] Specifically, the previously determined case prompt words, the text data to be inspected, and the preset quality inspection standards are combined to form a complete prompt word. This prompt word will serve as the input of the big model to guide the big model to perform quality inspection analysis. The prompt word contains all the necessary information so that the big model can fully understand the background and requirements of the quality inspection.

[0056] Through the corresponding interface, the integrated quality inspection request is sent to the big model for processing. After receiving the request, the big model will use its powerful natural language processing and pattern recognition capabilities to conduct in-depth analysis and evaluation of the text data. After the big model completes the processing, its output results are received.

[0057] The output results are parsed to extract key information related to quality inspection, such as identified problem points, risk assessment, and recommended improvement measures.

[0058] According to the preset quality inspection standards and the extracted key information, the quality inspection results are organized and presented in a standardized JSON format. This format has good readability and cross-platform compatibility, and can ensure the structure, readability and convenience of subsequent processing of the quality inspection results. The quality inspection results in JSON format can contain multiple fields, such as quality inspection status (qualified / unqualified), specific problem description, improvement suggestions, etc.

[0059] For example, by calling the large model module, the recording data that has been converted into text data to be quality checked is quality checked. Recording content: Customer: Hello, why is your washing machine not responding when I turn it on?

[0060] Customer Service: Hello, you can try to confirm whether the power cord is plugged in properly.

[0061] Customer: Oh, really. I forgot to plug it in. Thank you.

[0062] Customer Service: You’re welcome. Is there anything else I can help you with?

[0063] Customer: No more, bye.

[0064] In the above conversation, the customer service solved the customer's problem, and the customer satisfaction was high. And it complied with the SOP. The corresponding quality inspection score was high.

[0065] The large model prompts are as follows: Please conduct quality inspection on the above documents based on historical quality inspection cases and give them scores. The quality inspection content includes: 1: Whether to solve customer problems

[0066] 2: Customer satisfaction, high / medium / low

[0067] 3: Areas for improvement

[0068] 4: Overall rating (1 to 5, 5 is the highest)

[0069] The above content needs to generate corresponding JSON data.

[0070] After the above prompts are executed, the corresponding JSON data will be generated.

[0071] By using big models to perform quality inspection on recordings, text data can be automatically analyzed, reducing manual intervention, significantly improving quality inspection efficiency, and saving corporate operating costs. Big models can process complex text data, more comprehensively identify potential problems, and improve the accuracy of quality inspection. Automated quality inspection processes can significantly shorten the quality inspection cycle and improve work efficiency. Structured quality inspection results facilitate subsequent tracking of problems, analysis of causes, and adoption of improvement measures, which helps to continuously improve service quality.

[0072] According to a method for quality inspection of call recordings based on a large model provided by the present invention, after generating the quality inspection results, the application programming interface of the business system is called to convert the quality inspection results in JSON format into business data in the business system.

[0073] Furthermore, when the call recording is quality-checked based on the big model and the quality-check results are generated, the API of the business extension module is called to convert the JSON data into business data in the system. The business system call interface is encapsulated based on RestAPI to achieve the connection between the big model and the business system. The big model can call the business system capabilities more flexibly and easily to complete various business operations.

[0074] Specifically, by calling the API of the business system, the quality inspection results in JSON format are pushed to the business system. In the business system, these JSON data will be further parsed and converted into a business data format that matches the business system. This process ensures that the quality inspection results can be effectively integrated into the existing business process, providing business personnel with instant quality inspection feedback, thereby supporting faster and more accurate business decisions.

[0075] Through automated quality inspection result conversion and integration, the time and cost of manual processing are reduced, and the business processing speed is accelerated. The use of standardized JSON format and strict data verification mechanism ensures the accuracy and completeness of quality inspection results during the conversion process.

[0076] In one embodiment of the present invention, in order to further improve the consistency and effect of quality inspection, it is necessary to provide a manual re-inspection module to re-inspect or spot-check the results of some large model quality inspections, and make timely corrections and fine-tuning. In this way, the quality of the data can be further ensured, and the number of false positives and missed positives can be reduced, thereby providing users with more reliable and accurate data services.

[0077] Through the above method, the present invention improves the efficiency of quality inspection compared with traditional manual quality inspection, shields the human differences of quality inspectors, and has better consistency in quality inspection results. By introducing case prompt words and manual re-inspection modules, double protection is achieved, further ensuring the accuracy and consistency of quality inspection.

[0078] The following is a description of the device for quality inspection of call recordings based on a large model provided by the present invention. The device for quality inspection of call recordings based on a large model described below and the method for quality inspection of call recordings based on a large model described above can be referenced to each other.

[0079] A device for quality inspection of call recordings based on a large model, such as Figure 2 As shown, it includes the following modules: The module 210 for acquiring text data to be inspected is used to acquire call recording files and convert the call recording files into text data to be inspected; A case prompt word determination module 220 is used to obtain historical quality inspection cases and determine case prompt words according to the historical quality inspection cases; The large model quality inspection module 230 is used to perform quality inspection on the text data to be inspected based on the case prompt words, and generate quality inspection results according to preset quality inspection standards.

[0080] Figure 3 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 3 As shown, the electronic device may include: a processor 310, a communications interface 320, a memory 330 and a communication bus 340, wherein the processor 310, the communications interface 320 and the memory 330 communicate with each other through the communication bus 340. The processor 310 may call the logic instructions in the memory 330 to execute a method for quality inspection of call recordings based on a large model, the method comprising: obtaining a call recording file, converting the call recording file into text data to be quality inspected; obtaining historical quality inspection cases, determining case prompt words according to the historical quality inspection cases; performing quality inspection on the text data to be quality inspected based on the case prompt words through the large model, and generating quality inspection results according to preset quality inspection standards.

[0081] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0082] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the method for quality inspection of call recordings based on a large model provided by the above methods. The method includes: obtaining a call recording file, and converting the call recording file into text data to be quality inspected; obtaining historical quality inspection cases, and determining case prompt words based on the historical quality inspection cases; performing quality inspection on the text data to be quality inspected based on the case prompt words through a large model, and generating quality inspection results according to preset quality inspection standards.

[0083] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, it is implemented to execute the method for quality inspection of call recordings based on a big model provided by the above-mentioned methods. The method includes: obtaining a call recording file, and converting the call recording file into text data to be quality inspected; obtaining historical quality inspection cases, and determining case prompt words based on the historical quality inspection cases; performing quality inspection on the text data to be quality inspected based on the case prompt words through a big model, and generating quality inspection results according to preset quality inspection standards.

[0084] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, i.e., they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative effort.

[0085] Through the description of the above implementation modes, those skilled in the art can clearly understand that each implementation mode can be implemented by means of software plus a necessary general hardware platform, or of course by hardware. Based on such an understanding, the above technical solution can essentially or in other words be embodied in the form of a software product that contributes to the prior art. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiment.

[0086] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for quality inspection of call recordings based on a large model, characterized in that: include: Obtaining a call recording file, and converting the call recording file into text data to be inspected; Obtain historical quality inspection cases, and determine case prompt words according to the historical quality inspection cases; The text data to be inspected is quality inspected based on the case prompt words through the large model, and a quality inspection result is generated according to a preset quality inspection standard.

2. The method for quality inspection of call recordings based on a large model according to claim 1, characterized in that: The acquiring of the call recording file and converting the call recording file into text data to be inspected specifically includes: Obtain real-time call recordings or obtain existing call recordings from historical records to form a call recording file; Determine an automatic speech recognition tool, and configure parameters of the automatic speech recognition tool; Uploading the call recording file to a configured server or application programming interface of an automatic speech recognition tool and initiating a conversion request; The converted original text data is received, and the original text data is preprocessed to obtain text data to be inspected.

3. The method for quality inspection of call recordings based on a large model according to claim 1, characterized in that: The obtaining of historical quality inspection cases and determining case prompt words according to the historical quality inspection cases specifically includes: According to the text data to be inspected, relevant historical inspection cases are selected from the case library; Analyze the historical quality inspection cases and build a large model prompt word framework based on the analysis results; The large model prompt word framework is filled according to the text data to be quality inspected to determine the large model prompt words.

4. The method for quality inspection of call recordings based on a large model according to claim 1, characterized in that: The method of performing quality inspection on the text data to be inspected based on the case prompt words by using the big model and generating quality inspection results according to the preset quality inspection standards specifically includes: Integrate the case prompt word, the text data to be inspected, and the preset quality inspection standard into a prompt word as a quality inspection request; Sending the quality inspection request as input to the big model for processing, and obtaining the output result of the big model after processing the quality inspection request; Parsing the output results to extract key information; Generate quality inspection results in JSON format based on the preset quality inspection standards and the key information.

5. The method for quality inspection of call recordings based on a large model according to claim 2, characterized in that: The preprocessing of the original text data to obtain the text data to be inspected specifically includes: Performing text cleaning on the original text data to remove noise, irrelevant characters or special characters in the original text data; After text cleaning, the original text data is segmented to determine the dialogue turns; Perform text tagging on the segmented data to determine the identity of the interlocutor; The data after text cleaning, text segmentation and text labeling are used as text data for quality inspection.

6. The method for quality inspection of call recordings based on a large model according to claim 4, characterized in that: After generating the quality inspection result, the method further includes: The application programming interface of the business system is called to convert the quality inspection results in JSON format into business data in the business system.

7. A device for quality inspection of call recordings based on a large model, characterized in that: include: A module for acquiring text data to be inspected, used to acquire call recording files and convert the call recording files into text data to be inspected; A case prompt word determination module is used to obtain historical quality inspection cases and determine case prompt words according to the historical quality inspection cases; The large model quality inspection module is used to perform quality inspection on the text data to be inspected based on the case prompt words, and generate quality inspection results according to preset quality inspection standards.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for quality inspection of call recordings based on a large model as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for quality inspection of call recordings based on a large model as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for quality inspection of call recordings based on a large model as described in any one of claims 1 to 6 is implemented.

Citation Information

Cited By

  • Intelligent voice quality inspection method, system and device, storage medium and product

    CN120412549A

  • Intelligent agent-based home broadband installation and maintenance service work order quality inspection method and device

    CN121997926A