Execution evaluation method, device and equipment of first-copy standard verbal skill and storage medium

Abstract

The invention relates to the technical field of voice detection and the field of financial science and technology, and discloses an execution evaluation method for a first-copy standard verbal skill, and the method comprises the steps: obtaining the audio data of a claim settlement task, and dividing the audio of the first-copy task into a plurality of independent voice segments through the detection of a silence interval and the duration of the silence interval. And constructing a fuzzy voice library based on preset standard verbal skill keywords, comparing syllable units in the voice segments with keyword syllables in the fuzzy voice library, sharing syllable similarity, and generating an evaluation result of standard verbal skill execution of the claim settlement personnel. According to the invention, automatic evaluation of the verbal skill execution condition of the claim settlement personnel is realized, manual intervention is reduced, the accuracy and efficiency of evaluation are improved, the automation level of the claim settlement process is effectively improved, and the quality of customer service is ensured.

Description

Technical Field

[0001] The present invention relates to the fields of speech detection technology and financial technology, and in particular to an execution evaluation method, device, equipment and storage medium for first-link standard speech. Background Art

[0002] With the digital transformation of the financial industry, financial service processes such as insurance claims are gradually moving from traditional manual processing to automation and intelligence. In particular, in insurance claims, the first contact between claims adjusters and the insured is crucial, which not only affects customer satisfaction, but also determines the efficiency and accuracy of claims processing.

[0003] In the existing claims process, the first contact communication of claims adjusters usually needs to follow standardized words to ensure accurate information transmission and avoid misunderstandings. However, with the increase in business volume and the demand for improved service quality, it is difficult to meet industry requirements by relying solely on manual supervision of the standard words of claims adjusters. Financial institutions need a more efficient and automated way to evaluate the performance of claims adjusters in the first contact task, especially to effectively evaluate the execution of words.

[0004] Advances in speech recognition and natural language processing technologies have provided new solutions for the financial industry. Through the automatic collection, analysis, and processing of call audio data, the system can automatically evaluate the claims adjuster's implementation of standard scripts without human intervention. In particular, the use of technical means such as fuzzy speech libraries and semantic similarity analysis can more accurately determine whether the claims adjuster's answers meet the requirements of standard scripts, and provide feedback and improvements as needed.

[0005] Currently, the financial industry urgently needs a technology that can automatically identify and evaluate the standard scripts executed by claims adjusters in the first-link task, so as to improve business efficiency, ensure service quality, and reduce the cost of manual intervention. This technology can help financial institutions realize the automation and intelligence of the claims process and improve customer satisfaction. Summary of the invention

[0006] The main purpose of the present invention is to provide a method, device, equipment and storage medium for evaluating the execution of first-link standard scripts, aiming to solve the technical problem that the existing technology cannot effectively and automatically evaluate the execution of standard scripts by claims adjusters in first-link tasks, resulting in low evaluation efficiency.

[0007] To achieve the above object, the present invention provides an execution evaluation method of the first-link standard speech, comprising:

[0008] Obtain the first-link task audio data of the claims adjuster;

[0009] Identify and detect the silence interval in the audio data of the first connection task, and divide the audio data of the first connection task into multiple independent voice segments according to the length of the silence interval;

[0010] Determine the standard speech keywords, and build a fuzzy speech library of the first standard speech based on the standard speech keywords;

[0011] The syllable units of the speech segment are compared with the keyword syllables in the fuzzy speech library, and the execution evaluation result of the claims adjuster on the first-link standard speech is generated based on the syllable similarity.

[0012] In one embodiment, obtaining the first task audio data of the claims adjuster includes:

[0013] Establish a call connection through a virtual number to connect the claims adjuster and the insured, and monitor the call connection throughout the process;

[0014] During the call connection process, audio data of the claims adjuster and the insured are collected separately;

[0015] Convert the collected audio data of the claims adjuster and the insured into preset audio formats respectively;

[0016] The converted audio data of the claims adjuster is associated with the identification information of the claims task to generate the first-link task audio data of the claims adjuster and store it in the audio database.

[0017] In one embodiment, identifying and detecting a silent interval in the first-link task audio data, and dividing the first-link task audio data into a plurality of independent voice segments according to the duration of the silent interval, includes:

[0018] Detecting the sound intensity of the audio data of the first connection task, and identifying the sound intensity value of each frame by analyzing the audio data frame by frame;

[0019] Setting a sound intensity threshold, and marking an interval in the audio data of the first connection task whose sound intensity value is lower than the sound intensity threshold as a silent interval;

[0020] Setting a silence duration threshold, and selecting valid silence intervals whose duration exceeds the silence duration threshold from the silence intervals;

[0021] The first-link task audio data is segmented at the detected effective silence interval to generate multiple independent voice segments.

[0022] In one embodiment, a fuzzy speech library of the first standard speech is constructed based on the standard speech keywords, including:

[0023] Decomposing the pronunciation structure of each standard speech keyword into independent syllable units, wherein the syllable units include initial consonants and finals, and forming a basic syllable library;

[0024] Constructing a syllable matching model, generating a plurality of similarly pronounced syllable variants based on each syllable unit in the basic syllable library by analyzing the pronunciation similarity of the syllable units;

[0025] The syllable units and the corresponding syllable variants are grouped according to pronunciation similarity to generate multiple groups of keyword syllables, and the fuzzy speech library is constructed.

[0026] In one embodiment, comparing the syllable units of the speech segment with the keyword syllables in the fuzzy speech library includes:

[0027] Acquire the audio data of the insured and extract the voice clip of the insured's question, convert the voice clip of the question into question text, and analyze the semantics of the question in the question text;

[0028] Performing semantic analysis on the divided multiple voice segments, analyzing the semantic similarity between each voice segment and the question in the question text;

[0029] Based on the analysis result of semantic similarity, the valid answering speech segments whose semantic similarity exceeds the semantic similarity threshold are screened out from the multiple speech segments;

[0030] The syllable units of the effective answer speech segment are compared with the keyword syllables in the fuzzy speech library.

[0031] In one embodiment, the syllable units of the speech segment are compared with the keyword syllables in the fuzzy speech library, and the evaluation result of the claims adjuster's execution of the first-link standard speech is generated based on the syllable similarity, including:

[0032] Comparing the pronunciation similarity between the syllable unit of the speech segment and the keyword syllable, analyzing the initial consonant, final vowel, tone and speech speed of the syllable unit and the keyword syllable, and identifying the similarity between the speech segment and the keyword syllable;

[0033] Setting a similarity matching threshold, when the similarity between the speech segment and the keyword syllable exceeds the similarity matching threshold, it is determined that the match is successful;

[0034] Based on the comparison results of the voice fragments and the keyword syllables, the execution evaluation results of the claims adjusters in the execution of Shoulian standard scripts are generated.

[0035] In one embodiment, after generating the execution evaluation result of the first standard speech of the claims adjuster based on the syllable similarity, the method further includes:

[0036] Receiving feedback from the claims adjuster on the execution evaluation result;

[0037] When the objection feedback from the claims adjuster is identified, the syllable units of the voice segment are re-compared with the keyword syllables in the fuzzy voice library to generate a review result;

[0038] Determine whether there is a misjudgment based on the review results;

[0039] If there is a misjudgment, the misjudged syllable in the speech segment is identified based on the review result and the execution evaluation result, and the misjudged syllable is added to the fuzzy speech library.

[0040] Furthermore, to achieve the above-mentioned purpose, the present invention also provides an execution evaluation device for Shoulian standard speech, the execution evaluation device for Shoulian standard speech includes a memory, a processor, and an execution evaluation program for Shoulian standard speech stored in the memory and executable on the processor, and when the execution evaluation program for Shoulian standard speech is executed by the processor, the steps of the execution evaluation method for Shoulian standard speech as described above are implemented.

[0041] Furthermore, to achieve the above-mentioned purpose, the present invention also provides a computer storage medium, on which is stored an execution evaluation program of the first-link standard speech, and when the execution evaluation program of the first-link standard speech is executed by a processor, the steps of the execution evaluation method of the first-link standard speech as described above are implemented.

[0042] Beneficial effects: The present invention relates to an evaluation method for the execution of first-line standard speech, which obtains the audio data of the claims task and divides the first-line task audio into multiple independent voice segments by detecting the silence interval and the duration of the silence interval. A fuzzy voice library is constructed based on predetermined standard speech keywords, and the syllable units in the voice segments are compared with the keyword syllables in the fuzzy voice library to share the syllable similarity and generate an evaluation result of the claims adjuster's execution of the standard speech. The present invention not only realizes the automated evaluation of the claims adjuster's speech execution, but also reduces manual intervention, improves the accuracy and efficiency of the evaluation, effectively improves the automation level of the claims process, and ensures the quality of customer service. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0044] Figure 1 It is a flowchart of an embodiment of a method for evaluating the execution of the first-link standard speech technique of the present invention;

[0045] Figure 2 It is a functional module diagram of a preferred embodiment of the execution evaluation device of the first-link standard speech technique of the present invention;

[0046] Figure 3 This is a structural diagram of the device hardware operating environment involved in the implementation scheme of the execution evaluation device of the first link standard speech technique of the present invention. DETAILED DESCRIPTION

[0047] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0048] With the digital transformation of the financial industry, financial service processes such as insurance claims are gradually moving from traditional manual processing to automation and intelligence. In particular, in insurance claims, the first contact between claims adjusters and the insured is crucial, which not only affects customer satisfaction, but also determines the efficiency and accuracy of claims processing.

[0049] In the existing claims process, the first contact communication of claims adjusters usually needs to follow standardized words to ensure accurate information transmission and avoid misunderstandings. However, with the increase in business volume and the demand for improved service quality, it is difficult to meet industry requirements by relying solely on manual supervision of the standard words of claims adjusters. Financial institutions need a more efficient and automated way to evaluate the performance of claims adjusters in the first contact task, especially to effectively evaluate the execution of words.

[0050] Advances in speech recognition and natural language processing technologies have provided new solutions for the financial industry. Through the automatic collection, analysis, and processing of call audio data, the system can automatically evaluate the claims adjuster's implementation of standard scripts without human intervention. In particular, the use of technical means such as fuzzy speech libraries and semantic similarity analysis can more accurately determine whether the claims adjuster's answers meet the requirements of standard scripts, and provide feedback and improvements as needed.

[0051] Currently, the financial industry urgently needs a technology that can automatically identify and evaluate the standard scripts executed by claims adjusters in the first-link task, so as to improve business efficiency, ensure service quality, and reduce the cost of manual intervention. This technology can help financial institutions realize the automation and intelligence of the claims process and improve customer satisfaction.

[0052] See also Figure 1 , Figure 1 This is a flowchart of an embodiment of the method for evaluating the execution of the first-link standard speech provided by the present invention. It should be noted that although the logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than here.

[0053] like Figure 1 As shown, the execution evaluation method of the first-link standard speech proposed by the present invention includes the following steps:

[0054] S10, obtaining the first task audio data of the claims adjuster;

[0055] In this embodiment, the first contact task refers to the first contact task in the insurance claim process. Specifically, during the claim process, when the insured submits a claim application, the claims adjuster needs to communicate with the insured for the first time by phone or other means. The main objectives of this task are:

[0056] Confirm claim information: The claims adjuster verifies basic claim-related information with the insured, such as accident details, policy information, etc.

[0057] Introduce the claims process: Explain the specific steps and time points of the subsequent claims process to the insured to ensure that the insured understands the subsequent cooperation required.

[0058] Preliminary assessment of claim feasibility: Claim adjusters make a preliminary assessment of the rationality and authenticity of the claim application based on the insured's statement.

[0059] Automatically collect audio data of calls between claims adjusters and insureds through the system background or third-party services. Claim adjusters use virtual numbers to talk to insureds, and the system monitors and records the call content in real time. Audio collection can use voice recording equipment or recording functions integrated into the customer service system. Audio files can be saved in multiple formats (such as WAV, MP3, etc.) to ensure data compatibility and processability.

[0060] The collected audio data is stored in the system and associated with the identification information related to the claims task (such as task ID, claims personnel information, etc.) to ensure that the audio data corresponds to the specific task. Data storage can be done in a local database or in the cloud to ensure data security and accessibility.

[0061] When audio data is stored, metadata such as timestamps and call duration are automatically generated to help manage and retrieve data in subsequent processing. During the process of audio data collection and transmission, the system uses data encryption or dedicated channels to ensure data integrity and security to prevent data leakage or tampering.

[0062] Through the above steps, the automatic collection and management of audio data for claims tasks is realized, ensuring the integrity and accuracy of the data and reducing the need for manual intervention. Through real-time recording and secure storage, the security and reliability of audio data are guaranteed, and high-quality basic data support is provided for subsequent voice processing and speech evaluation.

[0063] S20, identifying and detecting the silence interval in the first-link task audio data, and dividing the first-link task audio data into a plurality of independent voice segments according to the duration of the silence interval;

[0064] In this embodiment, in the audio data, the silence interval refers to a period of time when there is no sound or the audio signal is lower than a set threshold. The silence interval in the audio is identified and used as the basis for audio segmentation. By analyzing the intensity value of the audio signal frame by frame, a sound intensity threshold is set. When the intensity of the audio signal is lower than the threshold and lasts for a period of time, the system marks this period as a silence interval. The length of the silence interval can be set according to actual needs. Generally, a longer silence interval indicates a pause between conversations.

[0065] Once the silence intervals are identified, the system can use the start and end times of these silence intervals to divide the audio data and divide the long audio into multiple smaller, independent voice segments.

[0066] After identifying the silence interval, the system segments the audio data according to the location of the silence interval. Each segmented voice segment represents the complete answer or statement of the claims adjuster, ensuring that subsequent processing can accurately locate the start and end points of each answer. The segmented voice segments can be further used for subsequent steps such as voice analysis and semantic matching.

[0067] In one specific implementation, the system detects silence intervals by the intensity of the audio signal and analyzes the sound amplitude of the audio data frame by frame. In order to enhance the accuracy of detection, the system sets multiple silence thresholds. For example, for conversations with slower speech speeds, a longer silence time is set, while for conversations with faster speech speeds, a shorter silence threshold is used. The system dynamically adjusts the audio data multiple times according to the length of the silence interval to ensure the natural division of the voice segments. After the voice segments are generated, the system quickly compresses each segment for easy storage and subsequent transmission.

[0068] In another specific embodiment, the system not only relies on audio intensity to detect silence intervals, but also uses speech spectrum analysis to identify more subtle pauses. By detecting frequency changes, the system can identify speech pauses caused by changes in emotions or tone. Not only does it focus on volume changes, but it also detects the spectral structure of the speech signal. When the spectrum changes exceed a set threshold, the system identifies it as silence or the end of a speech. The system then segments the audio to ensure that the speech segments accurately reflect the emotions and attitudes of the claims adjuster.

[0069] By detecting the silence intervals in the audio, the voice data can be accurately segmented to ensure that each voice segment represents an independent answer or statement from the claims adjuster. This improves the processing efficiency of voice data, helps with subsequent voice analysis, keyword matching, and the evaluation of standard speech execution, while reducing irrelevant parts in the audio and optimizing data storage and processing.

[0070] S30, determining the standard speech keywords, and constructing a fuzzy speech library of the first standard speech based on the standard speech keywords;

[0071] In this embodiment, the keywords that the claims adjusters need to follow in the first contact task are extracted from the predefined standard script library. These keywords are the core content that must be included in the conversation to standardize the claims process. The system first obtains the standard script text preset by the insurance company or financial institution and parses these texts. During parsing, the system extracts core keywords (such as "policy", "compensation", "accident", etc.) according to predefined rules. These keywords will be used as the basis for subsequent comparison and analysis.

[0072] In order to cope with different speaking speeds, accents and pronunciation differences, the system needs to build a fuzzy voice library containing keyword pronunciation variants. The fuzzy voice library is used to match the actual pronunciation of the claims adjuster with the syllable units of the standard keywords to ensure that the system can accurately recognize even if there are slight changes in pronunciation. The system performs phonetic decomposition on each keyword and extracts the syllable units such as initial consonants, finals and tones of the pronunciation. The system further generates multiple pronunciation variants based on the similarity of pronunciation and stores them in the fuzzy voice library. The fuzzy voice library will include the standard pronunciation of each keyword and its common variant pronunciations for subsequent comparison.

[0073] In another specific implementation, the system dynamically updates the standard speech keyword library based on the historical call records between the claims adjuster and the insured. The system extracts common keywords from the text of each call and expands these keywords based on actual usage. Through the machine learning model, the system can learn the pronunciation habits of claims adjusters in different regions and automatically store these pronunciation differences in the fuzzy voice library, allowing it to respond to different pronunciation situations more flexibly. The constructed fuzzy voice library will also be optimized regularly to remove uncommon or incorrect pronunciation variants.

[0074] By determining the keywords of standard speech and building a fuzzy speech library, the system can flexibly deal with the pronunciation differences caused by different voice features and improve the recognition accuracy of the standard speech executed by the claims adjuster. Building a fuzzy speech library can accommodate pronunciation variants, reduce recognition errors caused by factors such as accent and speaking speed, and effectively improve the robustness and adaptability of the system.

[0075] S40, comparing the syllable units of the speech segment with the keyword syllables in the fuzzy speech library, and generating the evaluation result of the claims adjuster's execution of the first-link standard speech based on the syllable similarity.

[0076] In this embodiment, when processing a voice segment, the system first needs to decompose the voice segment into multiple syllable units in order to compare them with the keyword syllables in the fuzzy speech library. The system uses a speech recognition algorithm to divide the voice segment into the smallest pronunciation units (such as initial consonants, finals, and tone combinations). These syllable units are specific pronunciation segments uttered by the claims adjuster during the call, representing various parts of the speech. After the segmentation is completed, the system will subsequently compare these syllable units with the keywords in the fuzzy speech library.

[0077] The system compares the extracted syllable units with the keyword syllables in the fuzzy speech library and analyzes the similarity of pronunciation. The goal of the comparison is to determine whether the claims adjuster's answer meets the standard script through syllable similarity. By calculating the similarity of parameters such as the syllable's initial consonant, final consonant, tone, and speaking speed, the system can determine whether the syllables in the voice segment match the keywords of the standard script. The similarity is calculated using a specific algorithm (such as similarity calculation based on spectral features), and a similarity matching threshold is set to ensure that only segments with syllable similarity that meets the standard are considered to be matched successfully.

[0078] Based on the results of the syllable comparison, the system generates an evaluation result of the claims adjuster's execution of the standard script. Based on the comprehensive results of the similarity calculation of each comparison segment, it evaluates whether the claims adjuster has correctly executed the standard script. The system summarizes all syllable comparison results and generates a total evaluation score or status, which indicates the claims adjuster's execution of the standard script during the call. The evaluation result can be in the form of a percentage (such as the proportion of syllables that are successfully matched), or a qualified or unqualified judgment, depending on the configuration of the system.

[0079] In other specific embodiments, the system combines emotional speech recognition technology. When comparing syllables, it not only focuses on pronunciation similarity, but also analyzes the tone, emotional fluctuations and other characteristics of the claims adjuster during the answering process through emotional recognition. The system extracts emotional features in the voice fragments through the emotional analysis module, and compares these features with the speech emotion model of the standard speech. The fuzzy speech library not only contains syllable units with standard pronunciation, but also stores syllable variants under different emotional states. Through the dual comparison of syllables and emotions, the system generates a comprehensive evaluation result of the claims adjuster, which not only evaluates whether the pronunciation is standard, but also considers whether the tone and emotional expression meet the requirements of standard speech.

[0080] Through syllable matching technology, the system can efficiently and accurately evaluate whether the claims adjuster has followed the standard script during the call, avoiding the subjective bias and inefficiency caused by manual review. The use of fuzzy voice database enables the system to handle the recognition difficulties caused by different pronunciations, speaking speeds and accents, greatly improving the accuracy of the evaluation. At the same time, the generation of the evaluation report provides data support for the standardization of the claims process, which helps to further optimize the quality of claims services.

[0081] The present invention relates to an execution evaluation method for first-line standard speech, which obtains audio data of the claims task and divides the first-line task audio into multiple independent voice segments by detecting the silence interval and the duration of the silence interval. A fuzzy voice library is constructed based on predetermined standard speech keywords, and the syllable units in the voice segments are compared with the keyword syllables in the fuzzy voice library to share the syllable similarity and generate an evaluation result of the claims personnel's execution of the standard speech. The present invention not only realizes the automated evaluation of the claims personnel's speech execution, but also reduces manual intervention, improves the accuracy and efficiency of the evaluation, effectively improves the automation level of the claims process, and ensures the quality of customer service.

[0082] In one embodiment, the above S10 includes:

[0083] S101, establish a call connection through a virtual number, connect the claims adjuster and the insured, and monitor the call connection throughout the process;

[0084] S102, collecting audio data of the claims adjuster and the insured during the call connection process;

[0085] S103, converting the collected audio data of the claims adjuster and the insured into a preset audio format;

[0086] S104, associating the converted audio data of the claims adjuster with the identification information of the claims task, generating the first task audio data of the claims adjuster, and storing it in the audio database.

[0087] In this embodiment, the system establishes a call connection through a virtual number to ensure that the communication between the claims adjuster and the insured is monitored and recorded by the system throughout the process. This method ensures privacy protection, and the entire process of the call is automatically managed by the system. The system generates a virtual number to replace the actual number of the claims adjuster, ensuring the privacy of both parties while ensuring that the system can establish a call connection through the virtual number. The system automatically starts monitoring the call and maintains the continuity of the recording during the call.

[0088] During the call, the system will collect audio data from the claims adjuster and the insured separately to ensure that both audios are recorded and can be processed and analyzed separately. The system uses dual-channel recording technology to collect audio data from the claims adjuster and the insured separately. The audio data of each party is recorded through independent channels to ensure the clarity of sound quality and the integrity of two-way communication. Audio files may be recorded in high-quality audio formats (such as WAV or MP3).

[0089] After the audio data is collected, it needs to be converted into a unified standard format for subsequent processing and analysis. After the recording is completed, the system automatically converts the audio data of the claims adjuster and the insured into a preset audio format, such as WAV, MP3 or other compatible formats. During the audio conversion process, the system can perform format compression or sound quality optimization according to the set parameters for easy storage and processing.

[0090] In order to ensure that the audio data is associated with a specific claim task, the system needs to bind each audio file to the corresponding claim task ID. When processing audio data, the system will attach relevant metadata to each audio file, including information such as the claim task ID, claim adjuster ID, and timestamp. Through this association, the system can ensure that each audio file corresponds to a specific task, which is convenient for subsequent retrieval and analysis.

[0091] The audio data of the claims adjusters generated by the system needs to be stored in the database for subsequent call and analysis. The system stores the generated audio files in a dedicated audio database, which classifies and stores the audio files according to information such as claims tasks, time, and claims adjusters. The system also regularly backs up the audio data to ensure data security and accessibility.

[0092] This embodiment uses a virtual number to monitor calls and automatically collect audio data, ensuring complete audio records during the communication between the claims adjuster and the insured, greatly improving the security and privacy of call data. Two-way recording and audio format conversion ensure the clarity and processability of the audio, and the association of audio data with claims tasks improves the accuracy and efficiency of subsequent analysis, providing strong support for the automation and standardization of the claims process.

[0093] In one embodiment, the above S20 includes:

[0094] S201, detecting the sound intensity of the audio data of the first connection task, and identifying the sound intensity value of each frame by analyzing the audio data frame by frame;

[0095] S202, setting a sound intensity threshold, marking an interval in the audio data of the first connection task whose sound intensity value is lower than the sound intensity threshold as a silent interval;

[0096] S203, setting a silence duration threshold, and selecting valid silence intervals whose duration exceeds the silence duration threshold from the silence intervals;

[0097] S204, segmenting the first-link task audio data at the detected effective silence interval to generate multiple independent voice segments.

[0098] In this embodiment, the audio data is analyzed frame by frame, and the sound intensity of each frame of audio is identified as the basis for subsequent silent interval detection. Each frame of audio corresponds to an audio signal within a period of time, and its intensity determines whether it is a speech or silent state. The system analyzes the audio data frame by frame through an audio processing algorithm and calculates the average intensity value of each frame of audio. The sound intensity reflects the loudness or energy of the sound. A higher intensity indicates a speech signal, and a lower intensity may be silence. Frame-by-frame analysis ensures that the system can accurately capture subtle sound changes and pauses.

[0099] The system sets a sound intensity threshold as a standard for determining whether the audio is in a silent state. Intervals with intensity below the threshold are marked as silent intervals. The system sets the sound intensity threshold by default or by dynamic adjustment. When the sound intensity of an audio frame is below the threshold and lasts for a certain period of time, the system marks these frames as silent intervals. The threshold can be adjusted according to different audio environments or noise levels to improve detection accuracy.

[0100] According to the duration of the silence interval, a silence duration threshold is set, and those silence intervals exceeding this duration are selected as valid silence intervals. The system calculates the duration of each interval marked as silence and records the start and end time of the silence interval. When the duration of the silence interval exceeds the preset duration threshold, the system determines it as a valid silence interval. The duration threshold can be set according to the specific needs of the conversation to ensure that short pauses do not affect the segmentation accuracy.

[0101] The audio data is segmented at the detected effective silence interval, and the entire audio data is divided into multiple independent voice segments for subsequent processing. The system cuts the audio data according to the location of the effective silence interval. The voice content before and after each silence interval is divided into independent voice segments. The segmented segments can be used for tasks such as speech analysis, semantic processing or keyword matching. The segmentation points are recorded by timestamps to ensure the accuracy of subsequent processing.

[0102] In one specific implementation, the system detects the intensity of the audio signal frame by frame through a voice processing engine, and automatically adjusts the sound intensity threshold during the processing process to adapt to different audio environments. In a high-noise environment, the system dynamically increases the intensity threshold to reduce noise interference; in a low-noise environment, the threshold is appropriately lowered to capture subtle voice signals. When a silent interval is detected, the system selects valid silent intervals based on the silence duration threshold, and automatically segments the audio at these silent intervals to generate independent voice segments.

[0103] This embodiment can effectively divide long continuous conversations into multiple independent voice segments by identifying and detecting silence intervals in the audio data of the first connection task and segmenting the audio data according to the duration of the silence intervals. It can accurately capture the pauses between each voice segment to avoid affecting subsequent voice analysis due to irrelevant voice segments or silence periods. It improves the efficiency and accuracy of audio data processing and provides a clear and structured voice segment data foundation for subsequent speech recognition, semantic analysis and standard speech evaluation.

[0104] In one embodiment, in the above S30, constructing a fuzzy speech library of the first standard speech based on the standard speech keywords includes:

[0105] S301, decomposing the pronunciation structure of each standard speech keyword into independent syllable units, wherein the syllable units include initial consonants and finals, and forming a basic syllable library;

[0106] S302, constructing a syllable matching model, generating a plurality of similarly pronounced syllable variants based on each syllable unit in the basic syllable library by analyzing the pronunciation similarity of the syllable units;

[0107] S303, grouping the syllable units and the corresponding syllable variants according to pronunciation similarity, generating multiple groups of keyword syllables, and constructing the fuzzy speech library.

[0108] In this embodiment, the system decomposes the pronunciation structure of each standard speech keyword into basic speech units, namely syllable units, which mainly include initial consonants and finals. The system uses speech recognition and decomposition algorithms to decompose the pronunciation structure of keywords into independent syllable units. Each syllable unit represents the smallest pronunciation unit in the keyword, including a combination of initial consonants and finals. The system will store these syllable units and establish a basic syllable library to provide a basis for subsequent matching and comparison.

[0109] Based on the basic syllable library, the system needs to generate multiple syllable variants with similar pronunciations to cope with pronunciation differences caused by different pronunciation habits, speaking speeds or local accents. The syllable matching model is used to analyze and generate these pronunciation variants. The system builds a syllable matching model through a speech similarity analysis algorithm or a model based on spectral features. The model can analyze the changes in the initials and finals of a syllable unit and generate multiple similar syllable variants based on these changes. These syllable variants can simulate different speaking speeds, accents or pronunciation methods to ensure that the system can still recognize keywords in different situations.

[0110] In order to efficiently perform keyword matching, the system needs to group syllable units and their variants according to the pronunciation similarity between them, and generate multiple groups of keyword syllables for each keyword. This grouping helps optimize the matching process. The system calculates the similarity of syllable units and groups each syllable unit and its generated syllable variants according to similarity. Each group contains syllable units and their variants with relatively similar pronunciations, which facilitates subsequent matching processing. Through this grouping method, the system generates multiple groups of keyword syllables in the fuzzy speech library, ensuring flexibility and accuracy in the matching process.

[0111] The fuzzy speech library contains the syllable units of standard speech keywords and their variant groups for subsequent syllable comparison. The fuzzy speech library is constructed to ensure that the system can accurately identify keywords under different speech conditions. The system stores all generated keyword syllable groups in the fuzzy speech library. The library contains not only standard pronunciation, but also multiple pronunciation variants to ensure that the system can perform effective speech matching regardless of changes in speech speed, accent or voice quality.

[0112] In another specific embodiment, the system automatically learns common pronunciation change patterns based on historical voice data and builds a dynamic syllable library. The syllable unit of each keyword not only includes the standard pronunciation, but also generates new syllable variants through the voice learning algorithm. When building the fuzzy voice library, the system will regularly update the syllable variant group to ensure that the system is highly flexible when dealing with different pronunciation environments. When grouping syllable groups, the system automatically prioritizes the pronunciation frequency of keywords in different contexts to improve matching speed and accuracy.

[0113] This embodiment decomposes the pronunciation structure of the standard script keywords into independent syllable units and constructs a fuzzy voice library containing multiple groups of keyword syllables. The system can flexibly respond to different pronunciation situations, including changes in speech speed, local accents, and inaccurate pronunciation. It improves the robustness and accuracy of voice matching, ensures that the pronunciation of the claims adjuster can be accurately compared with the standard script keywords, and ultimately improves the accuracy of the standard script execution evaluation.

[0114] In one embodiment, in the above S40, comparing the syllable unit of the speech segment with the keyword syllable in the fuzzy speech library includes:

[0115] S401, obtaining audio data of the insured and extracting the insured's question voice segment, converting the question voice segment into question text, and analyzing the semantics of the question in the question text;

[0116] S402, performing semantic analysis on the divided multiple voice segments, analyzing the semantic similarity between each voice segment and the question in the question text;

[0117] S403, based on the analysis result of the semantic similarity, selecting valid answering voice segments whose semantic similarity exceeds a semantic similarity threshold from the multiple voice segments;

[0118] S404: Compare the syllable units of the effective answer speech segment with the keyword syllables in the fuzzy speech library.

[0119] In this embodiment, the voice data of the insured is obtained from the conversation between the insured and the claims adjuster, and the voice clips containing the questions are extracted from it. Subsequently, the system converts the voice data into text and performs semantic analysis on the text to understand the questions raised by the insured. The system uses speech recognition (ASR) technology to convert the insured's voice data into text. After the speech recognition is completed, the system uses natural language processing (NLP) technology to perform semantic analysis on the text, extract key semantic information, and understand the questions raised by the insured. These key semantics will serve as the basis for subsequent voice clip comparison.

[0120] The system needs to perform semantic analysis on each voice clip of the claims adjuster and calculate the semantic similarity between it and the insured's question text. The system first converts the claims adjuster's answer voice clips into text, and then uses a semantic analysis model (such as BERT or Word2Vec) to calculate the semantic similarity between these answer texts and the insured's question text. The system will determine whether the claims adjuster's answer is related to the insured's question based on the similarity score.

[0121] Based on the analysis results of semantic similarity, the system selects the answers of the claims adjuster that are related to the insured's questions and filters out irrelevant or invalid answers. The system sets a semantic similarity threshold. When the similarity score between the claims adjuster's answer and the insured's question exceeds the set threshold, the system marks the segment as a valid answer voice segment. The threshold setting can be adjusted according to the similarity standard in actual application to ensure that the selected answers are valid responses to the questions.

[0122] The system further decomposes the selected valid answer fragments into syllable units and compares them with the keyword syllables in the fuzzy voice library to confirm whether the claims adjuster has used the keywords in the standard speech. The system uses syllable decomposition technology to extract each syllable unit in the valid answer fragment, and then compares it with the keywords in the fuzzy voice library. During the comparison process, the system analyzes the similarity between the syllable unit and the keyword syllable through the pronunciation similarity algorithm to ensure that the standard speech keywords are included in the answer. Keywords with similarity exceeding a certain threshold are considered to be matched successfully.

[0123] In another specific implementation, the system uses a deep learning model to semantically understand the text of the insured's questions and the claims adjuster's answers, and dynamically adjusts the semantic similarity threshold to adapt to different dialogue scenarios. The system first extracts the insured's question fragments, and then uses an adaptive model to analyze the semantic similarity between the claims adjuster's answers and the insured's questions. After screening out the valid answer fragments, the system compares its syllable units with the keywords in the fuzzy speech library, and uses speech spectrum analysis technology to ensure the accuracy of syllable comparison, and then determines whether the claims adjuster has accurately executed the standard speech.

[0124] In this embodiment, through semantic similarity analysis and syllable comparison technology, the system can effectively screen out valid answer fragments of the claims adjuster and accurately determine whether they conform to the standard script. This improves the recognition rate of valid answers in the conversation, ensures that the claims adjuster's answers are not only targeted at the insured's questions, but also strictly follow the standard script, further improving the standardization of the claims process and service quality.

[0125] In one embodiment, the above S40 includes:

[0126] S405, comparing the pronunciation similarity between the syllable unit of the speech segment and the keyword syllable, analyzing the initial consonant, final vowel, tone and speech speed of the syllable unit and the keyword syllable, and identifying the similarity between the speech segment and the keyword syllable;

[0127] S406, setting a similarity matching threshold, when the similarity between the speech segment and the keyword syllable exceeds the similarity matching threshold, it is determined that the match is successful;

[0128] S407, generating an evaluation result of the claims adjuster's performance in the execution of the Shoulian standard script based on the comparison result of the voice segment and the keyword syllables.

[0129] In this embodiment, the syllable units of the claims adjuster's voice fragments are compared with the keyword syllables in the fuzzy speech library for similarity. The comparison dimensions include the pronunciation features of the syllables, such as the initial consonant, final consonant, tone and speaking speed. The system decomposes the claims adjuster's voice fragments into syllable units through speech recognition technology. The pronunciation features of each syllable unit are extracted, including initial consonants, final consonants, tone and speaking speed. Subsequently, these syllable units are compared with the standard keyword syllables in the fuzzy speech library for similarity. The system calculates the similarity between each syllable unit and the keyword syllable by analyzing the pronunciation features.

[0130] The system needs to set a threshold for the similarity of syllable comparison as a standard for judging whether the syllable unit matches the keyword syllable. When the comparison similarity exceeds the set threshold, the system determines that the match is successful. The system sets a dynamic or fixed similarity threshold based on the similarity calculation results of the pronunciation characteristics of the syllable (initials, finals, tone and speaking speed). This threshold can be adjusted according to the requirements of different application scenarios to ensure the accuracy of syllable comparison. When the similarity score of the comparison is higher than the set threshold, the system determines that the voice segment of the claims adjuster successfully matches the keyword syllable in the standard script.

[0131] Based on the similarity results of the syllable comparison, the system generates an evaluation of the claims adjuster's execution in the Shoulian standard script. This evaluation is used to determine whether the claims adjuster correctly used the keywords in the standard script during the call. The system summarizes the results of the syllable comparison and calculates the number or percentage of successful matches between the claims adjuster and the standard script keywords during the call. The evaluation results can generate a specific execution score or be presented in the form of pass / fail. The evaluation report includes the claims adjuster's performance in executing the standard script as a basis for subsequent management and training.

[0132] In this embodiment, by setting syllable comparison and similarity matching thresholds, the system can accurately identify the claims adjuster's execution of standard dialogue during the call. It effectively improves the accuracy of dialogue execution evaluation, ensures that the system can handle different pronunciations, speaking speeds and intonations, and avoids misjudgments caused by subtle pronunciation differences. The generated execution evaluation report provides strong data support for the standardized management of the claims process, which helps to improve the overall quality of claims services.

[0133] In one embodiment, in the above S40, after generating the execution evaluation result of the first standard speech of the claims adjuster based on the syllable similarity, the method further includes:

[0134] S501, receiving feedback information from the claims adjuster on the execution evaluation result;

[0135] S502, when the objection feedback from the claims adjuster is identified, re-comparing the syllable units of the voice segment with the keyword syllables in the fuzzy voice library to generate a review result;

[0136] S503, judging whether there is a misjudgment according to the review result;

[0137] S504: If there is a misjudgment, identify the misjudged syllables in the speech segment based on the review result and the execution evaluation result, and add the misjudged syllables to the fuzzy speech library.

[0138] In this embodiment, the system allows the claims adjuster to provide feedback on the evaluation results. In particular, when the claims adjuster disagrees with the evaluation results, the system will record and process the feedback information. After generating the evaluation report, the system provides a feedback channel to the claims adjuster, allowing the claims adjuster to view the evaluation results and submit feedback information. The feedback content may include the part that the claims adjuster believes the system comparison is inaccurate or there is a misjudgment. The system stores this feedback information and marks it as pending, starting the subsequent review process.

[0139] When the system detects objection feedback from the claims adjuster on the assessment results, the system will re-compare the syllable units with the keyword syllables and generate a review result. After receiving the objection feedback, the system automatically re-analyzes the original voice segment, re-extracts the syllable units, and compares them again with the keyword syllables in the fuzzy voice library. When re-matching, the system may adjust the similarity threshold or enable additional comparison models to more accurately analyze the match between the voice segment and the standard speech. After the comparison is completed, the system generates a review result, indicating whether it is consistent with the original assessment result.

[0140] The system determines whether there is a misjudgment in the original assessment based on the re-checked review results. A misjudgment refers to the system's failure to correctly identify the claims adjuster's standard speech execution during the initial comparison. The system compares the review results with the initial assessment results. If there is a significant difference between the review results and the initial assessment results, the system will mark it as a misjudgment. The reasons for the misjudgment may include the system's inaccurate recognition of certain syllables, fuzzy speech, etc. The system records the misjudgment information and uses it to improve the comparison process.

[0141] When the system determines that there is a misjudgment, the system will identify the syllable units that caused the misjudgment based on the review results and evaluation results, and update these misjudged syllables to the fuzzy speech library to avoid misjudgments in similar situations in the future. The system automatically analyzes the syllable units that caused the misjudgment, extracts the initial consonant, final vowel, tone and other features of the misjudged syllable, and adds it to the fuzzy speech library as a new pronunciation variant. Through this update, the fuzzy speech library can recognize more variant pronunciations, further improving the accuracy of subsequent comparisons.

[0142] In one specific implementation, after the system detects objection feedback, the system starts an automated review process and submits the review results to manual assisted review. At this time, the syllable unit is automatically compared with the keyword syllable in the fuzzy speech library to generate a preliminary review result. The system first automatically reanalyzes the comparison between the syllable unit and the keyword syllable and generates a review result. At the same time, the review results and related audio data are submitted to the manual reviewer. During the manual assisted review, the reviewer confirms the accuracy of the system comparison through manual judgment and makes a final judgment. The system combines the results of automatic review and manual assisted review to determine whether there is a misjudgment. The system compares the opinions of the manual review with the results of the automatic review. If the manual assisted review and the system results are inconsistent, the system will determine whether there is a misjudgment based on the results of the manual review. The manual review results can supplement the system's possible misjudgments in complex pronunciation situations.

[0143] This embodiment ensures the accuracy and flexibility of the system in processing standard speech evaluation by receiving feedback from claims adjusters, re-matching syllables, identifying misjudgments, and updating the fuzzy speech library. By continuously updating the fuzzy speech library, the system can adapt to more pronunciation variants, reduce the misjudgment rate, and optimize the subsequent evaluation process. It effectively improves the system's adaptive ability and recognition accuracy, and provides claims adjusters with more fair and accurate evaluation results.

[0144] The present invention also provides an execution evaluation device for the first-link standard speech technique, referring to Figure 2 , Figure 2 This is a functional module diagram of a preferred embodiment of the execution evaluation device of the first-link standard speech of the present invention. The execution evaluation device of the first-link standard speech includes:

[0145] The first-call task audio acquisition module acquires the first-call task audio data of the claims adjuster;

[0146] An audio processing and division module, which identifies and detects the silence interval in the audio data of the first connection task, and divides the audio data of the first connection task into multiple independent voice segments according to the duration of the silence interval;

[0147] A fuzzy speech library construction module determines the standard speech keywords and constructs a fuzzy speech library of the first standard speech based on the standard speech keywords;

[0148] The syllable comparison and execution evaluation module compares the syllable units of the speech segment with the keyword syllables in the fuzzy speech library, and generates the execution evaluation result of the claims adjuster on the first-link standard speech based on the syllable similarity.

[0149] The specific implementation of the execution evaluation device of the first-link standard speech of the present invention is basically the same as the various embodiments of the execution evaluation method of the first-link standard speech mentioned above, and will not be repeated here.

[0150] The present invention also provides an execution evaluation device for the first-link standard speech, such as Figure 3 As shown, the execution evaluation device of the first-link standard speech may include: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the user interface 1003 may optionally include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or it may be a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may optionally be a storage device independent of the aforementioned processor 1001.

[0151] Those skilled in the art will understand that Figure 3 The hardware structure of the execution evaluation device of the Shoulian standard speech shown in the figure does not constitute a limitation of the execution evaluation device of the Shoulian standard speech, and may include more or less components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0152] like Figure 3 As shown, the memory 1005 as a storage medium may include an operating system, a network communication module, a user interface module, and an execution evaluation program of the first-link standard speech. Among them, the operating system is a program for managing and controlling the execution evaluation device and software resources of the first-link standard speech, supporting the operation of the network communication module, the user interface module, the execution evaluation program of the first-link standard speech, and other programs or software; the network communication module is used to manage and control the network interface 1004; the user interface module is used to manage and control the user interface 1003.

[0153] exist Figure 3 In the hardware structure of the execution evaluation device of the Shoulian standard speech shown, the network interface 1004 is mainly used to connect to the background server and communicate data with the background server; the user interface 1003 is mainly used to connect to the client and communicate data with the client; the processor 1001 can call the execution evaluation program of the Shoulian standard speech stored in the memory 1005, and perform the same operation as the execution evaluation method of the Shoulian standard speech.

[0154] The specific implementation of the execution evaluation device of the first-link standard speech of the present invention is basically the same as the various embodiments of the execution evaluation method of the first-link standard speech mentioned above, and will not be repeated here.

[0155] In addition, an embodiment of the present invention also proposes a computer storage medium, on which an execution evaluation program of the first-link standard speech is stored. When the execution evaluation program of the first-link standard speech is executed by a processor, the steps of the execution evaluation method of the first-link standard speech as described above are implemented.

[0156] The specific implementation of the computer storage medium of the present invention is basically the same as the various embodiments of the execution evaluation method of the above-mentioned first-link standard speech, and will not be repeated here.

[0157] It should be noted that if software tools or components other than those of our company appear in the embodiments of the present application, they are only used for illustrative purposes and do not represent actual use.

Claims

1. A method for evaluating the execution of the first-link standard speech, characterized in that: The following steps are involved: Obtain the first-link task audio data of the claims adjuster; Identify and detect the silence interval in the audio data of the first connection task, and divide the audio data of the first connection task into multiple independent voice segments according to the length of the silence interval; Determine the standard speech keywords, and build a fuzzy speech library of the first standard speech based on the standard speech keywords; The syllable units of the speech segment are compared with the keyword syllables in the fuzzy speech library, and the execution evaluation result of the claims adjuster on the first-link standard speech is generated based on the syllable similarity.

2. The method for evaluating the execution of the first-link standard speech technique as claimed in claim 1, characterized in that: Obtain the first-link task audio data of the claims adjuster, including: Establish a call connection through a virtual number to connect the claims adjuster and the insured, and monitor the call connection throughout the process; During the call connection process, audio data of the claims adjuster and the insured are collected separately; Convert the collected audio data of the claims adjuster and the insured into preset audio formats respectively; The converted audio data of the claims adjuster is associated with the identification information of the claims task to generate the first-link task audio data of the claims adjuster and store it in the audio database.

3. The method for evaluating the execution of the first-link standard speech technique as claimed in claim 1, characterized in that: Identifying and detecting a silent interval in the audio data of the first-link task, and dividing the audio data of the first-link task into a plurality of independent voice segments according to the duration of the silent interval, including: Detecting the sound intensity of the audio data of the first connection task, and identifying the sound intensity value of each frame by analyzing the audio data frame by frame; Setting a sound intensity threshold, and marking an interval in the audio data of the first connection task whose sound intensity value is lower than the sound intensity threshold as a silent interval; Setting a silence duration threshold, and selecting valid silence intervals whose duration exceeds the silence duration threshold from the silence intervals; The first-link task audio data is segmented at the detected effective silence interval to generate multiple independent voice segments.

4. The method for evaluating the execution of the first-link standard speech technique as claimed in claim 1, characterized in that: Based on the standard speech keywords, a fuzzy speech library of the first link standard speech is constructed, including: Decomposing the pronunciation structure of each standard speech keyword into independent syllable units, wherein the syllable units include initial consonants and finals, and forming a basic syllable library; Constructing a syllable matching model, generating a plurality of similarly pronounced syllable variants based on each syllable unit in the basic syllable library by analyzing the pronunciation similarity of the syllable units; The syllable units and the corresponding syllable variants are grouped according to pronunciation similarity to generate multiple groups of keyword syllables, and the fuzzy speech library is constructed.

5. The method for evaluating the execution of the first-link standard speech technique as claimed in claim 1, characterized in that: Comparing the syllable units of the speech segment with the keyword syllables in the fuzzy speech library includes: Acquire the audio data of the insured and extract the voice clip of the insured's question, convert the voice clip of the question into question text, and analyze the semantics of the question in the question text; Performing semantic analysis on the divided multiple voice segments, analyzing the semantic similarity between each voice segment and the question in the question text; Based on the analysis result of semantic similarity, the valid answering speech segments whose semantic similarity exceeds the semantic similarity threshold are screened out from the multiple speech segments; The syllable units of the effective answer speech segment are compared with the keyword syllables in the fuzzy speech library.

6. The method for evaluating the execution of the first-link standard speech as described in claim 1, characterized in that: The syllable units of the speech segment are compared with the keyword syllables in the fuzzy speech library, and the evaluation results of the claims adjuster's execution of the first-link standard speech are generated based on the syllable similarity, including: Comparing the pronunciation similarity between the syllable unit of the speech segment and the keyword syllable, analyzing the initial consonant, final vowel, tone and speech speed of the syllable unit and the keyword syllable, and identifying the similarity between the speech segment and the keyword syllable; Setting a similarity matching threshold, when the similarity between the speech segment and the keyword syllable exceeds the similarity matching threshold, it is determined that the match is successful; Based on the comparison results of the voice fragments and the keyword syllables, the execution evaluation results of the claims adjusters in the execution of Shoulian standard scripts are generated.

7. The method for evaluating the execution of the first-link standard speech technique as claimed in claim 1, characterized in that: After generating the execution evaluation result of the first standard speech of the claims adjuster based on the syllable similarity, it also includes: Receiving feedback from the claims adjuster on the execution evaluation result; When the objection feedback from the claims adjuster is identified, the syllable units of the voice segment are re-compared with the keyword syllables in the fuzzy voice library to generate a review result; Determine whether there is a misjudgment based on the review results; If there is a misjudgment, the misjudged syllable in the speech segment is identified based on the review result and the execution evaluation result, and the misjudged syllable is added to the fuzzy speech library.

8. A device for evaluating the execution of first-link standard speech, characterized in that: The execution evaluation device of the first-link standard speech technique comprises: The first-call task audio acquisition module is used to obtain the first-call task audio data of the claims adjuster; An audio processing and division module, used to identify and detect the silence interval in the audio data of the first connection task, and divide the audio data of the first connection task into multiple independent voice segments according to the length of the silence interval; A fuzzy speech library construction module is used to determine the standard speech keywords and construct a fuzzy speech library of the first standard speech based on the standard speech keywords; The syllable comparison and execution evaluation module is used to compare the syllable units of the speech segment with the keyword syllables in the fuzzy speech library, and generate the execution evaluation result of the claims adjuster on the first-link standard speech based on syllable similarity.

9. A device for evaluating the execution of the first-link standard speech, characterized in that: The execution evaluation device of the Shoulian standard speech includes a memory, a processor, and an execution evaluation program of the Shoulian standard speech stored in the memory and executable on the processor. When the execution evaluation program of the Shoulian standard speech is executed by the processor, the steps of the execution evaluation method of the Shoulian standard speech as described in any one of claims 1 to 7 are implemented.

10. A computer storage medium, characterized in that: The storage medium stores an execution evaluation program of the first-link standard speech, and when the execution evaluation program of the first-link standard speech is executed by the processor, the steps of the execution evaluation method of the first-link standard speech as described in any one of claims 1-7 are implemented.

Images