Voice quality inspection method, system and device based on intelligent electronic work card and medium

Abstract

The invention discloses a voice quality inspection method, system and device based on an intelligent electronic work card and a medium, and relates to the technical field of voice quality inspection. According to the method, a remote control instruction containing a business work order identifier is issued through an Internet of Things platform, an intelligent electronic work card responds to and collects voice data slices and uploads the voice data slices, and the voice data slices are spliced to form a complete voice service file; and processing to generate a dialogue text with a speaker tag, generating cue words through the dialogue text and the structured quality inspection template, inputting the cue words into the large language model, and guiding the large language model to match a quality inspection rule according to a service process and a speaker role. According to the invention, accurate acquisition, standardized processing and intelligent quality inspection of service voice are realized, the black box of service process management and control is broken, the efficiency of service quality tracing and evaluation is improved, and the method is suitable for scene service quality management and control.

Description

Technical Field

[0001] This invention relates to the field of voice quality inspection technology, and in particular to a voice quality inspection method, system, device and medium based on smart electronic badges. Background Technology

[0002] In on-site service scenarios such as door-to-door service and frontline sales, quality control of the service process has long relied on manual supervision and post-event traceability, lacking real-time and effective technological support. While some existing voice recording devices exist for recording on-site interactions, they generally suffer from the following significant drawbacks:

[0003] Firstly, voice collection often relies on manual initiation, which can easily lead to missed or incorrect recordings. Furthermore, it cannot be accurately linked to specific business work orders, making it difficult to match the corresponding service scenarios during quality inspection.

[0004] Secondly, the collected audio files lack standardized processing procedures, the audio formats uploaded from different devices are inconsistent, and the identities of speakers are difficult to distinguish when multiple people are talking, which affects the efficiency of subsequent quality inspection.

[0005] Third, the quality inspection rules rely on fixed algorithm configurations, which are inflexible and cannot dynamically adapt to quality inspection needs based on service stages and speaker roles, resulting in insufficient accuracy and limited coverage of quality inspection results.

[0006] These problems turn the service process into a black box for quality control, making it impossible to achieve full traceability of the service process, timely and accurate quality assessment, and severely restricting the improvement of service quality and management efficiency. Summary of the Invention

[0007] This invention provides a voice quality inspection method, system, equipment, and medium based on smart electronic work badges to solve the problems of difficult traceability, low timeliness, and poor validity of on-site service quality inspection.

[0008] This invention is achieved through the following technical solution:

[0009] A first aspect of the present invention provides a voice quality inspection method based on a smart electronic employee badge, comprising:

[0010] Obtain the complete voice service file associated with the target business work order; the complete voice service file is collected and sliced ​​and uploaded by the smart electronic work badge in response to the remote control command containing the identifier of the target business work order issued by the Internet of Things platform;

[0011] The complete voice service file is processed to generate dialogue text with speaker tags;

[0012] The dialogue text and the structured quality inspection template are used to generate prompt words, which are then input into the large language model. The structured quality inspection template is used to instruct the large language model to select from the structured quality inspection template and apply the corresponding quality inspection rules for matching analysis based on the service process in which the dialogue occurs and the speaker's role, so as to generate structured quality inspection results.

[0013] The system receives the quality inspection results returned by the large language model. The quality inspection results include at least the triggered quality inspection rules, the corresponding dialogue text fragments, and their position identifiers in the dialogue text.

[0014] Furthermore, the smart electronic work badge, in response to a remote control command issued by the IoT platform containing the target business work order identifier, collects and uploads data in slices, including:

[0015] The smart electronic work badge receives a remote recording start command containing the target service work order identifier, and performs the following operations based on the current recording status:

[0016] If the current recording device is idle, start recording and associate the recording with the target service work order;

[0017] If the current recording device is recording for another service order, the instruction is rejected and a failure response containing the service order identifier associated with the current recording is returned to the IoT platform.

[0018] If the current recording device is recording for the target service work order, a success response is returned to the IoT platform, and the collected voice data is sliced ​​and uploaded to the IoT platform at set intervals.

[0019] Further, the complete voice service file is processed to generate dialogue text with speaker tags, including:

[0020] The complete voice service file is segmented using a voice endpoint detection model to obtain multiple voice segments;

[0021] Speaker segmentation model is used to label each speech segment with a speaker and distinguish speaker roles.

[0022] Merge consecutive speech segments belonging to the same speaker role to form a dialogue segment;

[0023] The dialogue fragments are converted into dialogue text using a speech recognition model, and the speaker roles are labeled in the dialogue text to obtain dialogue text with speaker tags.

[0024] Furthermore, the speech recognition model is a Chinese speech recognition model that has been trained twice using the local dialect of the service area.

[0025] Furthermore, the structured quality inspection template defines multiple quality inspection rules, each of which is bound to at least one service process identifier and at least one speaker role identifier, and has a unique binding code; wherein, the service process identifier is used to refer to different service processes in the service flow, and the speaker role identifier is used to distinguish between the service provider and the customer;

[0026] The binding encoding is used to constrain the large language model to perform matching analysis by enabling corresponding quality inspection rules at dialogue segments in the dialogue text that conform to specific service processes and speaker roles.

[0027] Furthermore, the service process includes an on-site visit phase, a service phase, and a termination phase;

[0028] The quality inspection rules include at least one of the following:

[0029] During the on-site visit, do the service personnel introduce themselves, remind customers that the conversation will be recorded, show their employee badge, and confirm the customer's needs?

[0030] During the service phase, we assess whether the service personnel have a good attitude, use standard language, make any unethical requests, use any uncivilized language, and actively seek business opportunities.

[0031] At the end of the process, did the service personnel provide a completion reminder, remind the staff of any precautions, and check for any arbitrary charges?

[0032] At the end of the process, does the customer give a satisfactory evaluation?

[0033] Furthermore, the IoT platform abstracts and manages the smart electronic badge through a thing model, which defines the thing attributes and thing behaviors of the smart electronic badge;

[0034] The object attributes include the device model and communication protocol of the smart electronic badge, and the object behaviors include event reporting and command control;

[0035] The IoT platform uses a configurable rule parser to convert standard instructions from the platform side into device instructions that are compatible with different smart electronic badge communication protocols, or to parse non-standard data reported by the device into standard data format.

[0036] A second aspect of the present invention provides a voice quality inspection system based on a smart electronic employee badge, comprising:

[0037] The Internet of Things (IoT) platform is used to issue remote control commands containing business work order identifiers to smart electronic work badges, and to receive response information uploaded by the smart electronic work badges.

[0038] The smart electronic badge is used to respond to the remote control command issued by the Internet of Things platform, collect voice data and upload it in slices to the voice quality inspection service platform;

[0039] A voice quality inspection service platform is used to respond to a voice quality inspection task, obtain a complete voice service file associated with a target business work order, and execute the voice quality inspection method based on a smart electronic work card as described in any one of the first aspects of the present invention.

[0040] A third aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the voice quality inspection method based on a smart electronic badge as described in any one of the first aspects of the present invention.

[0041] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the voice quality inspection method based on a smart electronic badge as described in any one of the first aspects of the present invention.

[0042] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0043] This invention, based on intelligent electronic work badges, uses an IoT platform to issue remote control commands containing business work order identifiers. This enables targeted collection and segmented uploading of voice data, ensuring accurate binding between voice files and business work orders, and solving the problems of missed recordings and confused associations in traditional collection methods. The complete voice service file is processed to generate dialogue text with speaker tags. Combined with structured quality inspection templates and large language model matching analysis, this guides the large language model to flexibly apply corresponding quality inspection rules based on the service process and speaker roles during the dialogue, achieving dynamic adaptation and accurate matching of quality inspection rules. This enables real-time traceability and accurate evaluation of service quality, significantly improving the efficiency and scientific nature of service management. Attached Figure Description

[0044] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0045] Figure 1 This is a flowchart of a voice quality inspection method based on a smart electronic work badge according to an embodiment of the present invention;

[0046] Figure 2This is a flowchart of a method for processing quality inspection voice data according to an embodiment of the present invention;

[0047] Figure 3 This is a schematic diagram illustrating the processing of quality inspection voice data according to an embodiment of the present invention;

[0048] Figure 4 This is an architecture diagram of a voice quality inspection system based on a smart electronic work badge, according to an embodiment of the present invention.

[0049] Figure 5 This is an interactive schematic diagram of a voice quality inspection system process according to an embodiment of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0051] It should be noted that the terms "comprising" and "having" and any variations thereof in the specification, claims, and accompanying drawings of this invention are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to other steps or units inherent in the device.

[0052] The terminology used in the various embodiments of the invention is for the purpose of describing particular embodiments only and is not intended to limit the various embodiments of the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the invention pertain. The terms (such as those defined in commonly used dictionaries) are to be interpreted as having the same meaning as in the context of the relevant technical field and are not to be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of the invention.

[0053] This invention's voice quality inspection method uses electronic work badge smart hardware as a carrier, relying on sound pickup hardware and AI large language model voice quality inspection technology. The front end uses electronic work badges to collect trajectory and voice data; the middle platform develops an IoT platform to issue commands and collect trajectory and voice files; and the back end uses an AI semantic large language model for voice quality inspection. This end-to-end approach enables manageable, controllable, and assessable front-line service operations. It is applicable to on-site service scenarios such as home services, broadband installation and repair, and home appliance repair, as well as service quality control for front-line sales personnel in car sales, real estate sales, and other similar fields.

[0054] This invention proposes a voice quality inspection method based on smart electronic badges, applied to a voice quality inspection server. The voice quality inspection server, in response to voice quality inspection tasks targeting a specific range, executes the following... Figure 1 The voice quality inspection process is shown.

[0055] S101, Obtain the complete voice service file associated with the target business work order. This file is collected and sliced ​​by the smart electronic work badge in response to a remote control command containing the target business work order identifier issued by the IoT platform.

[0056] The voice quality inspection server determines the business work order to be inspected based on the quality inspection task, obtains the voice slice associated with the target business work order, and splices all voice slices belonging to the same target business work order according to the collection time sequence to obtain a complete voice service file for voice quality inspection.

[0057] Voice segments are collected and uploaded by the smart electronic work badge in response to remote control commands issued by the IoT platform, which contain the target business work order identifier. Each voice segment carries a business work order identifier. The voice quality inspection server retrieves the relevant voice segments from storage based on the target business work order identifier and splices them to obtain a complete voice service file. The complete voice service file contains all recordings from the start to the end of the service.

[0058] The intelligent electronic work badge terminal device includes a GPS latitude and longitude positioning module and a voice pickup module, enabling real-time personnel location data collection and voice file acquisition. It supports real-time uploads via multiple wireless methods such as 4G / 5G / Wi-Fi. Voice files collected on-site are uploaded in audio segments, with the segmentation time adjustable.

[0059] The smart electronic badge features a display screen that allows service personnel's key information (photo, name, employee ID, company, etc.) to be displayed with a single click, enabling customers to easily identify the service personnel. It also allows for remote activation and deactivation of recording via an IoT platform, with voice prompts to indicate when recording is on or off.

[0060] S102, Process the complete voice service file to generate dialogue text with speaker tags.

[0061] This process includes multiple speech processing steps such as speech endpoint detection, speech separation, speech segment merging, and speech recognition. It can be performed by calling existing speech services, ultimately separating the speech segments of the conversation between the service provider and the customer from the complete speech service file and recognizing them as dialogue text. Specific processing steps are as follows: Figure 2 , Figure 3 As shown.

[0062] S201, the complete voice service file is sent to the endpoint detection model, and the voice endpoint detection model segments the complete voice service file to obtain multiple voice segments.

[0063] For example, this invention uses Alibaba's "FSMN Speech Endpoint Detection - Chinese - General - 16k" model to detect speech segments and effectively segment and punctuate the speech file. This model adopts the FSMN (Feedforward Sequential Memory Network) structure, which balances modeling accuracy and inference speed. It achieves low latency by adjusting the right-looking frame rate and subdivides the traditional speech unit (speech) into monophones to improve feature representation ability and model discriminability.

[0064] In addition, this invention uses ffmpeg as an audio format alignment tool, and regardless of the audio file format uploaded by the electronic work badge, it is first uniformly converted to 16K wav format.

[0065] S202, send all speech segments to the speaker separation model, and use the speaker separation model to label each speech segment with a speaker and distinguish the speaker role.

[0066] For example, the present invention uses Alibaba's "CAM++ Speaker Confirmation-Chinese-General-200k-Spkrs" to vectorize speech features and uses a binary method to distinguish between two people, A and B, to label the speaker of each speech.

[0067] S203, merge consecutive speech segments belonging to the same speaker role to form a dialogue segment.

[0068] like Figure 3 As shown, consecutive speech segments of the same speaker, A, are spliced ​​together to form the dialogue between A and B.

[0069] S204, the dialogue segment is sent to the speech recognition model, which converts the dialogue segment into dialogue text and marks the speaker roles in the dialogue text to obtain the dialogue text with speaker tags.

[0070] For example, the present invention uses the Alibaba Magic Platform "Paraformer Speech Recognition-Chinese-General-16k-Offline-large-pytorch Model" to split different labeled speech segments into different speech files, which are then fed into the speech recognition model to obtain the recognized text.

[0071] Finally, the identified voice dialogue text is sent to the voice quality inspection server in the form of a Kafka message.

[0072] Furthermore, the speech recognition model is based on the Chinese speech recognition model, and then trained on the local dialect of the service area to effectively recognize the speech containing the local dialect recorded on the electronic work badge during door-to-door service.

[0073] S103, input the dialogue text and the structured quality inspection template to generate prompt words into the large language model; wherein, the structured quality inspection template is used to instruct the large language model to select from the structured quality inspection template and apply the corresponding quality inspection rules for matching analysis based on the service process in which the dialogue occurs and the speaker's role, so as to generate structured quality inspection results.

[0074] In the speech quality inspection stage, a general-purpose large language model is introduced for speech quality inspection reasoning. The large language model is accessed through the Dify development platform, and different general-purpose large language models can be selected as needed. For example, this invention uses DeepSeek-R1:70b, which can meet the language reasoning requirements.

[0075] The structured quality inspection template defines multiple quality inspection rules, each bound to at least one service process identifier and at least one speaker role identifier, and possessing a unique binding code. The service process identifier refers to different service processes within the service flow, while the speaker role identifier distinguishes between the service provider and the customer. The binding code constrains the large language model to activate the corresponding quality inspection rule for matching analysis at dialogue fragments in the dialogue text that match specific service processes and speaker roles.

[0076] Based on this, through model training and optimization, users can customize multi-dimensional voice quality inspection parameters to perform quality inspection on dialogue text.

[0077] In one implementation, the service process is divided into an on-site visit phase, a service phase, and a termination phase. The quality inspection rules include at least one of the following:

[0078] (1) During the on-site visit, did the service personnel introduce themselves, remind the customer that they would record the conversation, show their employee ID, and confirm the customer's needs?

[0079] (2) During the service phase, whether the service personnel have a good attitude, use standard language, make any illegal requests, use any uncivilized language, and explore business opportunities;

[0080] (3) At the end stage, did the service personnel remind you of the completion, did they remind you of the precautions, and did they charge you arbitrarily?

[0081] (4) At the end stage, did the customer give a satisfactory evaluation?

[0082] An example of a prompt word is as follows:

[0083] "Your task is to analyze the conversation, identify the installation and maintenance personnel and the customer, and extract the voice quality inspection items from the conversation. The specific steps are as follows:"

[0084] "1. Analyze the dialogue content and identify the speaker type for each dialogue (installation and maintenance personnel are userType1, and customers are userType2).

[0085] 2. For each dialogue, check if it contains voice quality inspection items. The voice quality inspection items are as follows:

[0086] {

[0087] Installation and maintenance personnel voice quality inspection items: [

[0088] {type:'1_1',label:'Identity Introduction'},

[0089] {type:'1_2',label:'Warning: Recording will be recorded'},

[0090] {type:'1_3',label:'Show employee ID'},

[0091] {type:'1_4',label:'Confirm Customer Needs'},

[0092] {type:'2_1',label:'Good attitude'},

[0093] {type: '2_2',label:'Standard Script'},

[0094] {type: '2_3',label:'Violation Requirement'},

[0095] {type: '2_4',label:'Inappropriate language'},

[0096] {type: '2_5',label:'Conduct Business Mining'},

[0097] {type: '3_1',label:'Completion Reminder'},

[0098] {type: '3_2',label:'Important Notes'},

[0099] {type: '3_3', label: 'There are arbitrary charges'} ]

[0101] Customer service voice quality inspection items: [

[0102] {type: '3_4', label: 'Customer Satisfaction Rating'}, ]

[0104] }

[0105] 3. If a voice quality check is detected, extract each word or phrase that appears in the voice quality check item, categorize it by type, add the type value to the list of each data item, add the extracted words or phrases to the textList of the corresponding item, and then return the array data.

[0106] 4. Return the result data in JSON format.

[0107] S104 receives the quality inspection results returned by the large language model.

[0108] Based on the prompts, the quality inspection results should include at least the triggered quality inspection rule, the corresponding dialogue text fragment, and its position identifier in the dialogue text.

[0109] During the voice acquisition phase, this invention utilizes an Internet of Things (IoT) platform to define and control smart electronic badges as object models. The IoT platform, centered on object models, configurates devices by defining their attributes and behaviors, simplifying communication between devices and applications, reducing coupling between them, and lowering the cost of replacing electronic badges from different brands.

[0110] First, the smart electronic badge is defined in the IoT platform:

[0111] (1) Attributes of the object: Through the introduced Internet of Things platform, the basic attributes of the electronic work badge can be defined, including size, device model, communication protocol, etc.

[0112] (2) Item behavior: Define the event reporting (upward), instruction control (downward) and parsing rules of electronic work badges.

[0113] The IoT platform uses a configurable rule parser to convert standard platform-side commands into device commands adapted to different smart electronic badge communication protocols, or to parse non-standard data reported by devices into standard data formats. It automatically generates Lua scripts for use during uplink and downlink processes to match the protocol differences between different types of electronic badges.

[0114] (1) Incident reporting includes:

[0115] Heartbeat events: Regularly report heartbeat data, including real-time GPS information;

[0116] Recording start event: Initiated when the electronic work badge presses the start recording button, a recording start event is sent to the IoT platform, where the recording belongs to the business work order number;

[0117] Recording End Event: Initiated when the recording ends on the electronic work badge, a recording end event is sent to the IoT platform, and the recording belongs to the business work order number.

[0118] (2) Command control includes:

[0119] Business Number Binding: Receives business numbers and other information sent to the electronic work badge by the IoT platform, used to display the business personnel information on the electronic work badge;

[0120] Business work order list sending: Receives a list of business work orders sent to the electronic work badge by the IoT platform, which is used for selection during recording.

[0121] Specifically, the method for controlled collection of voice data by the smart electronic work badge is as follows: the smart electronic work badge receives a remote recording start command containing the target business work order identifier, and performs the following operations based on the current recording status:

[0122] If the recording device is currently idle, start recording and associate the recording with the target work order;

[0123] If the current recording device is recording another work order, the instruction will be rejected and a failure response containing the work order identifier associated with the current recording will be returned to the IoT platform.

[0124] If the current recording device is recording for the target business work order, a success response is returned to the IoT platform, and the collected voice data is sliced ​​and uploaded to the voice quality inspection service platform at set intervals.

[0125] After the electronic work badge starts recording voice messages, the voice file will be uploaded in segments at regular intervals (every 5 minutes). The voice quality inspection service platform will sort and merge the uploaded segments according to the work order number and sequence number to form a complete service voice file.

[0126] Based on the same inventive concept, this invention proposes a voice quality inspection system based on a smart electronic employee badge, such as... Figure 4 The system architecture shown includes an IoT platform, at least one smart electronic badge managed by the IoT platform, a voice quality inspection service that interacts with the smart electronic badge through the IoT platform to perform voice quality inspection, and optionally, an AI platform and a voice processing service. The voice quality inspection service can call the AI ​​platform and the voice processing service via API to complete the preprocessing and quality inspection tasks of voice data. Among them:

[0127] The Internet of Things (IoT) platform is used to issue remote control commands containing business work order identifiers to smart electronic work badges, and to receive response information uploaded by smart electronic work badges.

[0128] The smart electronic badge is used to respond to remote control commands issued by the Internet of Things platform, collect voice data, and upload it in segments to the voice quality inspection service platform.

[0129] A voice quality inspection service platform is used to respond to a voice quality inspection task, obtain a complete voice service file associated with a target business work order, and execute the voice quality inspection method of any of the foregoing embodiments.

[0130] By enabling device access and management through an IoT platform, it can proactively adapt to various hardware devices and support smart hardware products with multiple communication and data protocols, achieving unified data collection, operational status monitoring, remote device start / stop, remote command issuance, and remote device updates and upgrades.

[0131] In this embodiment, the communication protocol between the electronic work badge and the Internet of Things platform is MQTT.

[0132] Using this system, the remote recording activation command is initiated by the IoT platform sending a request to individual electronic work badges to activate the recording function. Upon receiving the request, the electronic work badge needs to perform the following operations in sequence:

[0133] 1. The electronic employee badge extracts the "business number" attribute value from the data packet sent by MQTT and saves it to the electronic employee badge's built-in storage space (while also saving the corresponding sending time).

[0134] The "business number" storage requirement allows multiple electronic work badges to be stored simultaneously, but the data cannot be duplicated. If duplicate data is received, only one must be retained, but the corresponding issuance time must be updated to the most recent value.

[0135] 2. If the "Business Number" attribute value fails to be saved, the subsequent operation will be terminated, and a command execution failure message will be sent to MQTT, specifying the reason for the failure in the message.

[0136] 3. After the "Business Number" attribute value is successfully saved, it is necessary to first check the current recording function status of the device, and then execute the following branch processing according to different statuses:

[0137] If the device's "Current Recording Status" is off (=0), then execute the device's "Enable Recording Function" and continue from step 4 to begin continuous acquisition of audio files;

[0138] If the device's "Current Recording Status" is in the recording enabled state (=1 or=2), and the "Service Number" in the currently sent data is inconsistent with the "Service Number" attached to the ongoing recording operation, then the function of sending an execution failure message in step 4 will be executed directly, and the reason for failure will be marked as: "The current device is in the recording state, and its corresponding work order number is: xxxxxxx", and the ongoing voice file upload process will continue.

[0139] If the device's "Current Recording Status" is in the recording enabled state (=1 or=2), but the "Service Number" in the currently sent data is consistent with the "Service Number" attached to the recording operation in progress, then the function of sending the execution success message in step 5 will be executed directly, and the ongoing voice file upload process will continue.

[0140] 4. If the execution of "enable recording function" on the device fails, terminate the subsequent operation, and then send a command execution failure message to MQTT, specifying the reason for the failure in the message.

[0141] 5. If the device successfully "enables recording function", the current "service number" will be displayed on the touch screen, and a message indicating successful execution will be sent to MQTT. Then, subsequent operations will continue.

[0142] 6. The electronic work badge segments the audio file according to the set "voice reporting interval duration" and continuously sends the segmented result file to the HTTP interface.

[0143] 7. The voice file reporting operation continues until a remote recording shutdown command is received, or the recording shutdown button is pressed directly on the electronic work badge.

[0144] The "Enable Recording Function" is triggered manually by pressing the recording start button on the electronic work badge or by clicking the recording start function on the touch screen. Manual triggering of the recording start function requires that the device's "Current Recording Status" is in the off state (=0). If it is not in the off state, an illegal operation prompt should be displayed on the touch screen.

[0145] After the electronic employee badge's manual recording is activated, the following operations need to be performed in sequence:

[0146] 1. The electronic work badge displays a "Business Number" selection box on the touchscreen (the selection box provides five functions: switch number, select number, refresh number, start recording, and cancel recording. The "Business Numbers" are displayed in descending order of issuance time). The construction worker selects the "Business Number" for which recording needs to be started:

[0147] If the "Business Number" is selected, clicking the "Enable Recording" function will activate the device's "Enable Recording" function.

[0148] If the device fails to enable the "recording function", the reason for the failure will be displayed on the touchscreen and the current operation will be terminated.

[0149] If the device's "recording activation function" is successfully executed, the current "service number" will be displayed on the touchscreen, and a manual recording activation message will be sent to MQTT before proceeding with subsequent operations.

[0150] 2. The electronic work badge segments the audio file according to the set "voice reporting interval duration" and continuously sends the segmented result file to the HTTP interface.

[0151] 3. The voice file reporting operation continues until a remote recording shutdown command is received, or the recording shutdown button is pressed directly on the electronic work badge.

[0152] If a new smart electronic work badge or a new business number needs to be added, the IoT platform will proactively send a request to each electronic work badge to save the new business number. After receiving the request, the electronic work badge needs to perform the following operations in sequence:

[0153] 1. After the electronic employee ID card receives the request from MQTT, it first extracts the "Business Number" attribute value from the data packet, then compares it with the business number stored locally on the device, and executes the following branch processing based on the comparison result:

[0154] If the currently issued "Business Number" is not in the locally stored "Business Number" list, the "Add Business Number" function will fail, and the subsequent processing will continue from step 2.

[0155] If the currently issued "business number" is in the locally stored list of "business numbers", then the function of sending the execution success message in step 3 will be executed directly, and then this operation will be completed.

[0156] 2. If the "Add Business Number" function fails, terminate the subsequent operation and send a command execution failure message to MQTT, specifying the reason for the failure in the message.

[0157] 3. If the "Add Business Number" function is executed successfully, the current operation is completed, and a message indicating successful execution of the command is sent to MQTT.

[0158] This invention upgrades and transforms electronic work badges by leveraging the capabilities of an IoT platform based on general voice acquisition, achieving the following effects:

[0159] (1) Recording can be remotely started in the background when the installation and maintenance personnel sign in;

[0160] (2) Each recording of the electronic work badge is required to be associated with the work order number so that quality inspectors can easily find the customer scenario involved in the conversation;

[0161] (3) Since the installation and maintenance personnel have several work orders at the same time, they are allowed to leave the site before work order A is completed, manually end the recording, and go to another customer's site for construction; when they return, they can manually select work order A and start recording again.

[0162] The platform decouples applications from hardware, supporting data collection from various forms of electronic employee badges, such as name tags and badges. Figure 5 The diagram illustrates the interactive process of voice quality inspection. Electronic work badge-shaped terminal devices collect and report on-site voice data. The IoT platform provides voice file upload, storage, and distribution services. Through a unified object model, it converts data from heterogeneous electronic work badges, transforming the collected data into a unified format for various application systems. Simultaneously, the platform utilizes voice capability services to perform voice file conversion, separation, and recognition. The voice quality inspection service interacts with the IoT platform and various voice capability services to obtain quality inspection data. The voice text is then submitted to an AI voice model for quality inspection, yielding results. Users can interact with the voice quality inspection service through terminal applications to obtain inspection results or store the original files locally.

[0163] Embodiments of the present invention also provide an electronic device, which includes a processor and a memory, wherein the number of processors may be one or more. The memory, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules. The processor executes various functional applications and data processing of the electronic device by running the software programs, instructions, and modules stored in the memory, thereby implementing the voice quality inspection method based on a smart electronic work badge according to any of the above embodiments of the present invention.

[0164] The memory may primarily comprise a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a given function; the data storage area may store data created based on terminal usage. Furthermore, the memory may include high-speed random access memory (RAM) and non-volatile memory, such as at least one disk storage device, flash memory, or other non-volatile solid-state storage device. In some instances, the memory may further include memory remotely located relative to the processor, which can be connected to the electronic device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks (LANs), mobile communication networks, and combinations thereof.

[0165] Embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the voice quality inspection method based on a smart electronic work badge according to any embodiment of the present invention.

[0166] The computer storage medium of this invention can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0167] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0168] Embodiments of the present invention also provide a computer program product that, when run on a computer, causes the computer to execute the voice quality inspection method based on a smart electronic work badge according to any of the above embodiments of the present invention.

[0169] The above embodiments are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the appended claims.

Claims

1. A voice quality inspection method based on a smart electronic employee badge, characterized in that, include: Obtain the complete voice service file associated with the target service order; The complete voice service file is collected and sliced ​​up by the smart electronic work badge in response to a remote control command issued by the Internet of Things platform containing the identifier of the target business work order; The complete voice service file is processed to generate dialogue text with speaker tags; The dialogue text and the structured quality inspection template are used to generate prompt words, which are then input into the large language model. The structured quality inspection template is used to instruct the large language model to select from the structured quality inspection template and apply the corresponding quality inspection rules for matching analysis based on the service process in which the dialogue occurs and the speaker's role, so as to generate structured quality inspection results. The system receives the quality inspection results returned by the large language model. The quality inspection results include at least the triggered quality inspection rules, the corresponding dialogue text fragments, and their position identifiers in the dialogue text.

2. The voice quality inspection method based on intelligent electronic work badges according to claim 1, characterized in that, The smart electronic work badge responds to a remote control command issued by the IoT platform, which includes the identifier of the target business work order, by collecting and uploading data in slices, including: The smart electronic work badge receives a remote recording start command containing the target service work order identifier, and performs the following operations based on the current recording status: If the current recording device is idle, start recording and associate the recording with the target service work order; If the current recording device is recording for another service order, the instruction is rejected and a failure response containing the service order identifier associated with the current recording is returned to the IoT platform. If the current recording device is recording for the target service work order, a success response is returned to the IoT platform, and the collected voice data is sliced ​​and uploaded to the IoT platform at set intervals.

3. The voice quality inspection method based on intelligent electronic work badges according to claim 1, characterized in that, The complete voice service file is processed to generate dialogue text with speaker tags, including: The complete voice service file is segmented using a voice endpoint detection model to obtain multiple voice segments; Speaker segmentation model is used to label each speech segment with a speaker and distinguish speaker roles. Merge consecutive speech segments belonging to the same speaker role to form a dialogue segment; The dialogue fragments are converted into dialogue text using a speech recognition model, and the speaker roles are labeled in the dialogue text to obtain dialogue text with speaker tags.

4. The voice quality inspection method based on intelligent electronic work badges according to claim 3, characterized in that, The speech recognition model is a Chinese speech recognition model that has been trained twice using the local dialect of the service area.

5. The voice quality inspection method based on intelligent electronic work badges according to claim 1, characterized in that, The structured quality inspection template defines multiple quality inspection rules. Each quality inspection rule is bound to at least one service process identifier and at least one speaker role identifier, and has a unique binding code. The service process identifier is used to refer to different service processes in the service flow, and the speaker role identifier is used to distinguish between the service provider and the customer. The binding encoding is used to constrain the large language model to perform matching analysis by enabling corresponding quality inspection rules at dialogue segments in the dialogue text that conform to specific service processes and speaker roles.

6. The voice quality inspection method based on intelligent electronic work badges according to claim 1, characterized in that, The service process includes an on-site visit phase, a service phase, and a termination phase. The quality inspection rules include at least one of the following: During the on-site visit, do the service personnel introduce themselves, remind customers that the conversation will be recorded, show their employee badge, and confirm the customer's needs? During the service phase, we assess whether the service personnel have a good attitude, use standard language, make any unethical requests, use any uncivilized language, and actively seek business opportunities. At the end of the process, did the service personnel provide a completion reminder, remind the staff of any precautions, and check for any arbitrary charges? At the end of the process, does the customer give a satisfactory evaluation? 7. The voice quality inspection method based on intelligent electronic work badges according to claim 1, characterized in that, The IoT platform abstracts and manages the smart electronic badge through a thing model, which defines the thing attributes and thing behaviors of the smart electronic badge. The object attributes include the device model and communication protocol of the smart electronic badge, and the object behaviors include event reporting and command control; The IoT platform uses a configurable rule parser to convert standard instructions from the platform side into device instructions that are compatible with different smart electronic badge communication protocols, or to parse non-standard data reported by the device into standard data format.

8. A voice quality inspection system based on a smart electronic employee badge, characterized in that, include: The Internet of Things (IoT) platform is used to issue remote control commands containing business work order identifiers to smart electronic work badges, and to receive response information uploaded by the smart electronic work badges. The smart electronic badge is used to respond to the remote control command issued by the Internet of Things platform, collect voice data and upload it in slices to the voice quality inspection service platform; A voice quality inspection service platform is used to respond to a voice quality inspection task, obtain a complete voice service file associated with the target business work order, and execute the voice quality inspection method based on a smart electronic work card as described in any one of claims 1-7.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the voice quality inspection method based on smart electronic badges as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the voice quality inspection method based on smart electronic badges as described in any one of claims 1-7.

Images