Conference quality evaluation method and device, storage medium and terminal

Abstract

The application discloses a conference quality evaluation method and device, a storage medium and a terminal. Audio and video data of a current conference are periodically acquired, and emotional representation information of at least one participant in the current conference is determined according to the audio and video data. Emotional states of the participants are recognized according to the emotional representation information, and an emotional state data set corresponding to the current conference is obtained. The emotional state data set is input into a conference evaluation large model, and a conference quality evaluation result output by the conference evaluation large model for the current conference is determined. In the embodiment of the application, the audio and video data of the conference are recognized, the emotional representation information of the participants in the conference is determined, and the emotional states of the participants are analyzed according to the information, so that the emotional states are input into the conference evaluation large model, the quality evaluation result corresponding to the conference is obtained, and the effect of accurately and quantitatively evaluating the quality of the conference is realized.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, storage medium, and terminal for evaluating meeting quality. Background Technology

[0002] Online conferencing has become a widely used communication method. It has no geographical limitations, allowing participants to join meetings from anywhere with internet access, increasing flexibility and saving costs. Furthermore, online meetings facilitate sharing of materials and real-time communication tools via the cloud, making it easy for participants to share and discuss relevant content. The participation status of each individual affects the efficiency of the meeting. Compared to offline meetings, online meetings may involve greater physical distance between participants. This means that when a user's participation is poor, it can significantly hinder the effective dissemination and sharing of meeting information. Summary of the Invention

[0003] This application provides a meeting quality assessment method, apparatus, storage medium, and terminal, which can solve the technical problem in related technologies that it is difficult to capture abnormal meeting participation status in a timely manner.

[0004] In a first aspect, embodiments of this application provide a method for evaluating meeting quality, the method comprising:

[0005] The audio and video data of the current meeting are periodically acquired, and the emotional representation information of at least one participant in the current meeting is determined based on the audio and video data.

[0006] Based on the emotional representation information, the emotional state of each participant is identified, and the emotional state dataset corresponding to the current meeting is obtained;

[0007] The emotional state dataset is input into the meeting evaluation model to determine the meeting quality evaluation result of the meeting evaluation model for the current meeting.

[0008] Secondly, embodiments of this application provide a meeting quality assessment device, the device comprising:

[0009] The information acquisition module is used to periodically acquire the audio and video data of the current meeting, and determine the emotional representation information of at least one participant in the current meeting based on the audio and video data.

[0010] The emotion recognition module is used to identify the emotional state of each participant based on the emotion representation information, and obtain the emotion state dataset corresponding to the current meeting.

[0011] The quality assessment module is used to input the emotional state dataset into the meeting assessment model and determine the meeting quality assessment result output by the meeting assessment model for the current meeting.

[0012] Thirdly, embodiments of this application provide a computer storage medium storing a plurality of instructions adapted for loading by a processor and executing the steps of the method described above.

[0013] Fourthly, embodiments of this application provide a terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being adapted to be loaded by the processor and to execute the steps of the above-described method.

[0014] The beneficial effects of the technical solutions provided in some embodiments of this application include at least the following: periodically acquiring audio and video data of the current meeting; determining the emotional representation information of at least one participant in the current meeting based on the audio and video data; identifying the emotional state of each participant based on the emotional representation information to obtain an emotional state dataset corresponding to the current meeting; and inputting the emotional state dataset into a large-scale meeting evaluation model to determine the meeting quality evaluation result output by the large-scale meeting evaluation model for the current meeting. Since the emotional state of each participant during the meeting reflects their level of reception of effective information, in this embodiment of the application, by identifying the audio and video data of the meeting, determining the emotional representation information of each participant in the meeting, and analyzing the emotional state of the participants based on this information, the emotional state is input into the large-scale meeting evaluation model to obtain the corresponding meeting quality evaluation result. This achieves the effect of accurately quantifying the evaluation of meeting quality. Effective and accurate meeting quality evaluation results help managers further evaluate and optimize meeting effectiveness and efficiency. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 An exemplary system architecture diagram of a meeting quality assessment method provided in this application embodiment;

[0017] Figure 2 A flowchart illustrating a meeting quality assessment method provided in this application embodiment;

[0018] Figure 3 A flowchart illustrating a meeting quality assessment method provided in this application embodiment;

[0019] Figure 4A schematic diagram illustrating the training process of a large-scale meeting evaluation model in a meeting quality assessment method provided in this application embodiment;

[0020] Figure 5 A structural block diagram of a meeting quality assessment device provided in this application embodiment;

[0021] Figure 6 This is a schematic diagram of the structure of a terminal provided in an embodiment of this application. Detailed Implementation

[0022] To make the features and advantages of this application more apparent and understandable, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims. Furthermore, in the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B; the word "and / or" in the text is merely a description of the relationship between related objects, indicating that three relationships can exist, for example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Additionally, in the description of the embodiments of this application, "multiple" refers to two or more.

[0024] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature.

[0025] In today's rapidly digitalizing world, online conferencing has become a widely adopted and deeply applied communication and exchange method across various industries. Its internet-based conferencing model breaks down geographical and spatial limitations of traditional meetings, allowing multiple users in different locations to participate anytime, anywhere with just one internet-connected device. This feature effectively improves meeting efficiency, making product demonstrations, application sharing, and project collaboration more convenient.

[0026] However, despite the numerous conveniences and advantages of online meetings, the participants' state of mind becomes a key factor affecting meeting efficiency and effectiveness in practical applications. Compared to offline meetings, online participants are often in different physical spaces, creating a greater distance between them. Therefore, it's difficult to intuitively perceive each other's emotions and state as in face-to-face communication. When a participant is not in a good state, such as experiencing severe network latency, being unresponsive, or lacking focus, it can easily hinder the transmission of information, even leading to misunderstandings and barriers, thus severely impeding the effective dissemination and sharing of meeting information. For example, in online classrooms, students and teachers are both in the meeting, but it's difficult for teachers to monitor students' classroom status in real time. Severe network latency, disconnections, or students becoming distracted or lacking focus significantly impact the completeness and accuracy of knowledge transmission, resulting in low overall meeting efficiency and quality.

[0027] Therefore, this application provides a meeting quality assessment method to solve the aforementioned technical problem of difficulty in timely capturing abnormal meeting participation status.

[0028] Please see Figure 1 , Figure 1 An exemplary system architecture diagram of a meeting quality assessment method provided in this application embodiment.

[0029] like Figure 1 As shown, the system architecture may include a terminal 101, a network 102, and a server 103. The network 102 serves as the medium for providing a communication link between the terminal 101 and the server 103. The network 102 may include various types of wired or wireless communication links, such as wired communication links including fiber optic cables, twisted-pair cables, or coaxial cables, and wireless communication links including Bluetooth communication links, Wireless-Fidelity (Wi-Fi) communication links, or microwave communication links, etc.

[0030] Terminal 101 can interact with server 103 via network 102 to receive messages from or send messages to server 103. Alternatively, terminal 101 can interact with server 103 via network 102 to receive messages or data sent to server 103 by other users. Terminal 101 can be hardware or software. When terminal 101 is hardware, it can be various electronic devices, including but not limited to smartwatches, smartphones, tablets, laptops, and desktop computers. When terminal 101 is software, it can be installed in the aforementioned electronic devices and can be implemented as multiple software programs or software modules (e.g., to provide distributed services) or as a single software program or software module; no specific limitation is made here.

[0031] In this embodiment, terminal 101 first periodically acquires the audio and video data of the current meeting, and determines the emotional representation information of at least one participant in the current meeting based on the audio and video data; further, terminal 101 identifies the emotional state of each participant based on the emotional representation information, and obtains the emotional state dataset corresponding to the current meeting; based on this, terminal 101 inputs the emotional state dataset into the meeting evaluation model, and determines the meeting quality evaluation result output by the meeting evaluation model for the current meeting.

[0032] Server 103 can be a business server providing various services. It should be noted that server 103 can be either hardware or software. When server 103 is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When server 103 is software, it can be implemented as multiple software programs or software modules (e.g., used to provide distributed services), or as a single software program or software module; no specific limitations are made here.

[0033] Alternatively, the system architecture may not include server 103. In other words, server 103 may be an optional device in the embodiments of this specification. That is, the method provided in the embodiments of this specification can be applied to a system structure that only includes terminal 101. The embodiments of this application do not limit this.

[0034] It should be understood that Figure 1 The number of terminals, networks, and servers shown is only illustrative; the number can be any number of terminals, networks, and servers depending on the implementation requirements.

[0035] Please see Figure 2 , Figure 2 This is a flowchart illustrating a meeting quality assessment method provided in an embodiment of this application. The executing entity in this embodiment can be a terminal performing the meeting quality assessment, a processor within the terminal performing the meeting quality assessment method, or a meeting quality assessment service within the terminal performing the meeting quality assessment method. For ease of description, the following example uses a processor within the terminal as the executing entity to illustrate the specific execution process of the meeting quality assessment method.

[0036] like Figure 2 As shown, meeting quality assessment methods may include at least:

[0037] S202. Periodically acquire the audio and video data of the current meeting, and determine the emotional representation information of at least one participant in the current meeting based on the audio and video data.

[0038] Optionally, current meeting quality assessments typically involve collecting and organizing objective data such as attendance, speaking volume, and meeting minutes. Related technologies mainly focus on the convenience of process data, such as automatic check-in, real-time voice recording, and automated information archiving. However, objective data recording alone cannot represent the quality of a meeting, because the essence of meeting quality lies in the effectiveness of information delivery, which is primarily reflected in participants' concentration, mental engagement, and positive emotions. To ensure the effectiveness of meeting information delivery, organizers generally need to proactively take a series of measures to ensure the smooth running of online meetings. For example, thorough testing and preparation before the meeting to ensure that each participant's equipment and network are in good working order; and maintaining participants' attention and participation during the meeting through regular roll call and interactive sessions. However, existing meeting quality assessment schemes fail to reflect the specific emotional state of participants in some objective data; for example, attendance figures do not reflect the actual participation status of each participant. Furthermore, the implementation of a series of intervention measures largely depends on the manager's experience-based judgment of the current meeting quality. The quality assessment results based on experience are not necessarily accurate, and each intervention measure initiated will interrupt the ongoing business in the current meeting. This means that the intervention measures do not effectively improve the meeting quality even when they interrupt the meeting.

[0039] Alternatively, to accurately assess meeting quality, considering that the emotional state of each participant reflects their level of reception of effective information, this means that the emotional state of participants can be used to evaluate meeting quality. Therefore, to obtain participants' emotional information, one can start with the meeting's audio and video data. Video data can be used to obtain facial expressions and body language information, thereby analyzing each participant's emotions and feelings; simultaneously, audio data can be used to analyze the speaker's emotional information.

[0040] Specifically, in conducting sentiment analysis, pre-set timers or event-triggered mechanisms can be used to periodically capture audio and video data from the current meeting. This allows for continuous quality assessment of each stage of the meeting as it progresses, ultimately yielding a more comprehensive and detailed evaluation of the entire meeting's quality. The audio and video data typically includes high-definition video streams and high-quality audio streams, ensuring accurate capture of every detail of the meeting. After acquiring the audio and video data, advanced image processing and speech recognition technologies are used to identify the emotional representations of the participants.

[0041] In one possible implementation, deep learning algorithms, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), can be used to identify facial expressions such as smiling, frowning, and nodding, as well as speech features such as speech rate, volume, and tone. These features serve as the emotional representation information of the participants, reflecting their positive, neutral, or negative emotional tendencies. Specifically, when a speaker is present, the audio data must contain at least the speaker's audio recording. Therefore, for the speaker among the participants, the emotional representation information in the audio and video data must include at least one of the following: tone of voice, keywords used in speech, and facial expressions. For other participants who are not speaking, their emotional representation information must include at least one of the following: body language or facial expressions.

[0042] Furthermore, the speaker's emotional representation includes tone of voice, encompassing speech speed, volume, and intonation variations. These factors directly reflect the speaker's emotional state, such as excitement, tension, or calmness. Key words in the speech reveal the focus and direction of the content; frequent use or emphasis of keywords indicates the speaker's attitude and level of importance attached to the current topic. Additionally, captured facial expressions, such as smiling, frowning, and nodding, further supplement the emotional expression beyond language, making the speaker's true attitude more three-dimensional and easier to understand. For other participants who did not speak, their emotional representation is also an important reference for assessing the meeting atmosphere and participation. Body language, such as leaning forward to indicate attentive listening, crossed hands suggesting questioning or reservation, and frequent nodding indicating agreement or consent, directly reflects inner emotions. Similarly, facial expressions are equally important; shaking the head, eye contact, or a slight smile can convey emotional inclinations such as interest, doubt, or satisfaction with the current topic. In summary, by carefully capturing and analyzing the emotional representations of participants, we can gain a deeper understanding of each member's level of engagement in the meeting and their stance and attitude towards the current discussion topics.

[0043] S204. Identify the emotional state of each participant based on the emotional representation information to obtain the emotional state dataset corresponding to the current meeting.

[0044] Optionally, after determining the emotional representation information of the participants, these emotional representations are further analyzed and processed in depth using emotion recognition algorithms. Machine learning models, such as Support Vector Machines (SVM) or Deep Neural Networks (DNN), are used to identify the current emotional state of each participant, such as happiness, sadness, anger, surprise, etc. These emotional states are integrated into an emotional state dataset corresponding to the current meeting, which contains the emotional state information of each participant identified during the meeting. It should be noted that the emotional states exemplified above are merely examples; in real-world scenarios, the output emotional states may have more refined and diverse distinctions, which are not limited to this embodiment.

[0045] S206. Input the emotional state dataset into the meeting evaluation model and determine the meeting quality evaluation result of the meeting evaluation model for the current meeting.

[0046] Optionally, after obtaining the sentiment state dataset, it can be further used for meeting quality assessment. To achieve accurate analysis of the sentiment state dataset, a large-scale meeting assessment model is trained using deep learning technology in this embodiment. This model pre-learns various meeting quality assessment metrics related to sentiment states during training. By inputting the sentiment state dataset into the pre-trained large-scale meeting assessment model, the model can comprehensively analyze the sentiment state dataset and other relevant information, ultimately outputting a comprehensive meeting quality assessment result. The meeting quality assessment result may be a score, a rating, or a detailed assessment report, used to help meeting administrators understand the overall effectiveness of the meeting for subsequent improvements and optimizations.

[0047] In a preferred embodiment, the model outputs a complete and detailed meeting quality assessment result, which includes meeting quality level, participant emotional category statistics, and meeting quality cause analysis report. By using data and reports from multiple dimensions to reflect meeting quality, a reasonable quantitative assessment of meeting quality is achieved. The effective and multi-dimensional meeting quality assessment result helps managers to further evaluate and optimize meetings.

[0048] This application provides a method for assessing meeting quality. It periodically acquires audio and video data of the current meeting, determines the emotional representation information of at least one participant based on the audio and video data, identifies the emotional state of each participant based on the emotional representation information, and obtains an emotional state dataset corresponding to the current meeting. The emotional state dataset is then input into a large-scale meeting assessment model to determine the meeting quality assessment result output by the model. Since the emotional state of each participant during the meeting reflects their reception of effective information, this application, by identifying the audio and video data of the meeting, determining the emotional representation information of each participant, and analyzing the emotional state based on this information, inputs the emotional state into a large-scale meeting assessment model to obtain the corresponding meeting quality assessment result. This achieves accurate quantitative assessment of meeting quality, and the effective and accurate meeting quality assessment result helps managers further evaluate and optimize meeting effectiveness and efficiency.

[0049] Please see Figure 3 , Figure 3 This is a flowchart illustrating a meeting quality assessment method provided in an embodiment of this application.

[0050] like Figure 3 As shown, meeting quality assessment methods may include at least:

[0051] S302. Periodically acquire the audio and video data of the current meeting, and determine the emotional representation information of at least one participant in the current meeting based on the audio and video data.

[0052] S304. Identify the emotional state of each participant based on the emotional representation information to obtain the emotional state dataset corresponding to the current meeting.

[0053] S306. Input the emotional state dataset into the meeting evaluation model and determine the meeting quality evaluation result of the meeting evaluation model for the current meeting.

[0054] For details regarding steps S302-S306, please refer to the detailed descriptions in steps S202-S206, which will not be repeated here.

[0055] S308. If the current meeting is determined to be of abnormal quality based on the meeting quality assessment results, a prompt message and intervention strategy corresponding to the meeting quality assessment results are generated, and the prompt message and intervention strategy are sent and displayed to the administrator terminal corresponding to the current meeting.

[0056] Optionally, if the meeting quality assessment determines that the current meeting is of abnormal quality—for example, many participants show no interest in the current topic—then, to optimize the meeting's quality, it is necessary to generate a notification message and intervention strategy corresponding to the meeting quality assessment results. This notification message and intervention strategy should then be sent and displayed to the administrator's terminal for the current meeting. The notification message clearly informs the administrator of the meeting's quality anomaly, while the intervention strategy is an automatically generated, optional approach to optimizing meeting quality based on the anomaly. The administrator can select and implement the intervention strategy as needed, thereby quickly and efficiently optimizing meeting quality.

[0057] Optionally, if the meeting quality assessment results indicate that the ongoing meeting has significant quality abnormalities, such as a lack of interest among most participants in the current discussion, manifested as inattentiveness, reduced interaction, and negative feedback, then it is particularly important to take a series of targeted measures to improve this unfavorable situation in a timely and effective manner and enhance the overall quality and efficiency of the meeting.

[0058] Specifically, when meeting quality is abnormal, a prompt message corresponding to the meeting quality assessment results can be generated. This message indicates the specific problems and their severity, such as "Over 60% of participants in this meeting have low interest in the topic, resulting in a significant decrease in participation." It also includes one or more carefully designed intervention strategies. These intervention strategies are automatically generated based on the analysis of historical meeting data, the identification of participant behavior patterns, and the understanding of meeting objectives, aiming to quickly respond to and resolve quality issues in the current meeting. For example, intervention strategies might include: adjusting the meeting agenda to introduce new topics or activities more relevant to participants' interests; increasing interactive elements, such as group discussions or Q&A sessions, to stimulate participant enthusiasm. Subsequently, both the prompt message and the intervention strategies are sent and displayed to the administrator's terminal for the current meeting. The administrator's terminal can be a dedicated meeting management software interface or an administrator's mobile device application, ensuring that the administrator can receive this important information anytime, anywhere. The prompt message will have a concise and clear presentation, ensuring that the administrator can immediately grasp the key issues in the meeting quality; while the intervention strategies should be presented in an easy-to-understand and operable format, facilitating the administrator's rapid evaluation and selection of the most appropriate strategy for implementation. Through this mechanism, administrators can not only be aware of any abnormalities in meeting quality immediately, but also flexibly select and implement appropriate optimization measures based on the provided intervention strategies, their professional knowledge, and understanding of the meeting context. This not only helps to quickly reverse the meeting atmosphere and improve participant engagement and satisfaction, but also provides valuable experience and reference for future meeting organization, promoting continuous improvement and enhancement of meeting quality.

[0059] S310. After the current meeting ends, generate the time sequence information of the meeting and each meeting quality assessment result based on all meeting quality assessment results of the current meeting, calculate the full-process quality statistics of the current meeting, generate the meeting report of the current meeting based on the full-process quality statistics, and store the meeting report to the storage server.

[0060] Optionally, after the current meeting concludes, to comprehensively and deeply analyze the overall effectiveness and quality of the meeting, a full meeting report can be generated based on all detailed quality assessment results. This process first involves summarizing and integrating all quality assessment data collected during the meeting according to the time-series information corresponding to each specific assessment result. The time-series information records in detail the changing trends of various quality indicators during the meeting process. Combining the time-series information with all quality assessment data clearly shows the changes in meeting quality as the meeting progresses, providing valuable time-dimensional data for subsequent in-depth analysis.

[0061] Furthermore, comprehensive quality statistics for the current meeting are calculated. These statistics collectively constitute a comprehensive and objective quantitative evaluation of the overall quality of the meeting. Supported by these statistics, a meeting report can be generated. The meeting report is a detailed document integrating a meeting overview, quality assessment results, time-series information analysis, and comprehensive quality statistics. In addition, the report may include comparative analysis, comparing the quality data of the current meeting with that of similar previous meetings to highlight progress and shortcomings. Once successfully generated, the meeting report is automatically uploaded and securely stored on a designated storage server. As an efficient and reliable data storage platform, the storage server ensures the long-term preservation and convenient access of the meeting report. Whether it's the meeting organizer, participants, or other relevant personnel, anyone with appropriate permissions can download and view the meeting report from the server, providing strong support for subsequent work improvements, experience sharing, or academic research. This process not only enhances the scientific and standardized nature of meeting management but also lays a solid foundation for the continuous optimization and improvement of meetings.

[0062] This application provides a meeting quality assessment method. If the meeting quality assessment results indicate an anomaly in the current meeting's quality, such as a significant number of participants showing disinterest in the current topic, then a corresponding prompt message and intervention strategy are generated and displayed on the administrator's terminal for the current meeting. The prompt message clearly informs the administrator of the meeting's quality anomaly, while the intervention strategy is an automatically generated, optional approach to optimizing meeting quality based on the anomaly. The administrator can select and implement the intervention strategy as needed, thereby quickly and efficiently optimizing meeting quality. After the meeting concludes, all meeting quality result data is added to the meeting report, allowing subsequent users to access the meeting's quality data when downloading the report. This helps participants review each stage of the meeting, laying the foundation for continuous meeting optimization and improvement.

[0063] Please see Figure 4 , Figure 4 This is a schematic diagram of the training process of a large-scale meeting evaluation model in a meeting quality evaluation method provided in this application embodiment.

[0064] like Figure 4 As shown, the training process for evaluating large models at the conference can include at least the following:

[0065] S402. Construct an initial large-scale meeting evaluation model for meeting quality assessment scenarios based on the basic prediction model.

[0066] Optionally, if the large-scale meeting evaluation model needs to perform quality assessment based on multiple features of the meeting's sentiment state dataset, a pre-trained basic prediction model can be used to construct an initial event classification model for the event classification scenario. The basic multimodal model can output prediction results for the predicted object based on multiple different types of features, thus the initial event classification model possesses the basic prediction logic of "outputting prediction results for the predicted object based on multiple different types of features." The initial meeting evaluation model includes at least a meeting scenario adaptation module and a natural language generation module. The meeting scenario adaptation module enables the initial meeting evaluation model to perform meeting quality assessment, and the natural language generation module enables the model to output natural language.

[0067] Optionally, when the basic prediction model is directly applied to a specific scenario, the unadjusted model is difficult to adapt to the new scenario. Therefore, after constructing the initial meeting evaluation model for the meeting evaluation scenario, it is also necessary to train the initial meeting evaluation model in a targeted manner.

[0068] S404. Obtain multiple sample sentiment state datasets, all of which have standard conference quality assessment labels.

[0069] Optionally, multiple sample sentiment state datasets are obtained for training the initial large-scale meeting evaluation model. Each sample sentiment state dataset carries a standard meeting quality assessment label. The standard meeting quality assessment label represents the correct meeting quality assessment result corresponding to each sample sentiment state dataset; therefore, using it as the standard label for sample meeting quality assessment is beneficial for model fitting.

[0070] S406. Input multiple sample sentiment state datasets into the initial meeting evaluation model and train the initial meeting evaluation model.

[0071] S408. During the training of the initial meeting evaluation model, the initial meeting evaluation model is controlled to output predicted meeting quality evaluation labels for multiple sample sentiment state datasets. The parameters of the initial meeting evaluation model are adjusted according to the predicted meeting quality evaluation labels and the standard meeting quality evaluation labels until the initial meeting evaluation model converges, thus obtaining the trained meeting evaluation model.

[0072] Optionally, during the training of the initial meeting evaluation model, the initial meeting evaluation model is controlled to output predicted meeting quality evaluation labels for multiple sample sentiment state datasets. These predicted meeting quality evaluation labels are the prediction results of the initial meeting evaluation model for multiple sample sentiment state datasets. The difference between the predicted meeting quality evaluation labels and the standard meeting quality evaluation labels represents the difference between the current state and the expected performance of the initial meeting evaluation model. Further, the training loss value of the model can be calculated based on the predicted and standard meeting quality evaluation labels. The parameters of the meeting scenario adaptation module in the initial meeting evaluation model are adjusted based on the training loss value, while the parameters of the natural language generation module remain unchanged, until the initial meeting evaluation model converges to obtain the trained meeting evaluation model.

[0073] Optionally, the model's training termination conditions may include, for example, the loss function value being less than or equal to a preset loss function threshold, or the number of iterations reaching a preset threshold. Specific training termination conditions can be determined based on actual circumstances and are not specifically limited here.

[0074] This application provides a method for assessing meeting quality, including a training method for a large-scale meeting assessment model. A large-scale meeting assessment model is constructed using a base prediction model, and then specifically trained for meeting scenarios. This enables the large-scale meeting assessment model to deeply analyze the emotional state dataset of the meeting, and through precise feature extraction and natural language processing, output accurate meeting quality assessment results.

[0075] Please see Figure 5 , Figure 5 This is a structural block diagram of a meeting quality assessment device provided in an embodiment of this application. Figure 5 As shown, the meeting quality assessment device 500 includes:

[0076] The information acquisition module 510 is used to periodically acquire the audio and video data of the current meeting and determine the emotional representation information of at least one participant in the current meeting based on the audio and video data.

[0077] The emotion recognition module 520 is used to identify the emotional state of each participant based on emotion representation information, and obtain the emotion state dataset corresponding to the current meeting.

[0078] The quality assessment module 530 is used to input the emotional state dataset into the meeting assessment model and determine the meeting quality assessment result of the meeting assessment model for the current meeting.

[0079] Optionally, when the participant is a speaker, the speaker's emotional representation information includes at least one of the following: tone of voice, key words, and facial expressions; when the participant is another person who is not speaking, the other person's emotional representation information includes at least one of the following: body language or facial expressions.

[0080] Optionally, the meeting quality assessment device 500 further includes a strategy prompt module, which generates prompt information and intervention strategies corresponding to the meeting quality assessment results if the current meeting is determined to be of abnormal quality based on the meeting quality assessment results, and sends and displays the prompt information and intervention strategies to the administrator terminal corresponding to the current meeting.

[0081] Optionally, the meeting quality assessment results may include at least one of the following: meeting quality level, participant emotional category statistics, or meeting quality cause analysis report.

[0082] Optionally, the meeting quality assessment device 500 further includes: a report generation module, used to generate meeting and time sequence information corresponding to each meeting quality assessment result based on all meeting quality assessment results of the current meeting after the current meeting ends, calculate the full-process quality statistics of the current meeting; generate a meeting report of the current meeting based on the full-process quality statistics, and store the meeting report to a storage server.

[0083] Optionally, the meeting quality assessment device 500 further includes: a model training module, used to construct an initial meeting assessment model for the meeting quality assessment scenario based on a basic prediction model; acquire multiple sample sentiment state datasets, each of which carries a standard meeting quality assessment label; input the multiple sample sentiment state datasets into the initial meeting assessment model to train the initial meeting assessment model; during the training process of the initial meeting assessment model, control the initial meeting assessment model to output predicted meeting quality assessment labels for the multiple sample sentiment state datasets, and adjust the parameters of the initial meeting assessment model according to the predicted meeting quality assessment labels and the standard meeting quality assessment labels until the initial meeting assessment model converges, thus obtaining the trained meeting assessment model.

[0084] Optionally, the model training module is also used to obtain a basic prediction model; based on the basic prediction model, an initial meeting evaluation model is created that includes at least a meeting scenario adaptation module and a natural language generation module; the training loss value is calculated based on the predicted meeting quality evaluation label and the standard meeting quality evaluation label; the parameters of the meeting scenario adaptation module in the initial meeting evaluation model are adjusted based on the training loss value, and the parameters of the natural language generation module are kept unchanged until the initial meeting evaluation model converges.

[0085] This application provides a meeting quality assessment device, comprising: an information acquisition module for periodically acquiring audio and video data of the current meeting and determining the emotional representation information of at least one participant based on the audio and video data; an emotion recognition module for identifying the emotional state of each participant based on the emotional representation information, obtaining an emotional state dataset corresponding to the current meeting; and a quality assessment module for inputting the emotional state dataset into a meeting assessment model to determine the meeting quality assessment result output by the meeting assessment model for the current meeting. Since the emotional state of each participant during the meeting reflects their reception of effective information, this application, by recognizing the audio and video data of the meeting, determining the emotional representation information of each participant, and analyzing the emotional state of the participants based on this information, inputs the emotional state into the meeting assessment model to obtain the corresponding meeting quality assessment result. This achieves accurate quantitative assessment of meeting quality, and the effective and accurate meeting quality assessment result helps managers further evaluate and optimize meeting effectiveness and efficiency.

[0086] This application also provides a computer storage medium that can store multiple instructions adapted for loading by a processor and executing the steps of any of the methods described in the above embodiments.

[0087] Please see Figure 6 , Figure 6This is a schematic diagram of the structure of a terminal provided in an embodiment of this application. Figure 6 As shown, terminal 600 may include: at least one terminal processor 601, at least one network interface 604, user interface 603, memory 605, and at least one communication bus 602.

[0088] The communication bus 602 is used to enable communication between these components.

[0089] The user interface 603 may include a display screen and a camera. Optionally, the user interface 603 may also include a standard wired interface and a wireless interface.

[0090] The network interface 604 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0091] The terminal processor 601 may include one or more processing cores. The terminal processor 601 connects to various parts within the terminal 600 using various interfaces and lines, and performs various functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 605, and by calling data stored in the memory 605. Optionally, the terminal processor 601 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The terminal processor 601 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content to be displayed on the screen; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the terminal processor 601 and may be implemented as a separate chip.

[0092] The memory 605 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory 605 may include a non-transitory computer-readable storage medium. The memory 605 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 605 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 605 may also be at least one storage device located remotely from the aforementioned terminal processor 601. Figure 6 As shown, the memory 605, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and a conference quality assessment program.

[0093] exist Figure 6 In the terminal 600 shown, the user interface 603 is mainly used to provide an input interface for the user and to obtain the user's input data; while the terminal processor 601 can be used to call the conference quality assessment program stored in the memory 605 and specifically perform the following operations:

[0094] Periodically acquire audio and video data of the current meeting, and determine the emotional representation information of at least one participant in the current meeting based on the audio and video data;

[0095] Based on the emotional representation information, the emotional state of each participant is identified, and the emotional state dataset corresponding to the current meeting is obtained.

[0096] Input the emotional state dataset into the large meeting evaluation model to determine the meeting quality evaluation result of the large meeting evaluation model for the current meeting.

[0097] In some embodiments, when the participant is a speaker, the speaker's emotional representation information includes at least one of the following: tone of voice, key words, and facial expressions; when the participant is another person who is not speaking, the other person's emotional representation information includes at least one of the following: body language and facial expressions.

[0098] In some embodiments, after executing the meeting quality assessment result of the large meeting assessment model for the current meeting, the terminal processor 601 further performs the following steps: if the meeting quality assessment result determines that the current meeting is of abnormal quality, then a prompt message and intervention strategy corresponding to the meeting quality assessment result are generated, and the prompt message and intervention strategy are sent and displayed to the administrator terminal corresponding to the current meeting.

[0099] In some embodiments, the meeting quality assessment results may include at least one of the following: meeting quality level, participant emotional category statistics, or meeting quality cause analysis report.

[0100] In some embodiments, the terminal processor 601 further performs the following steps: after the current meeting ends, it generates time-series information of the meeting and each meeting quality assessment result based on all meeting quality assessment results of the current meeting, calculates the full-process quality statistics of the current meeting, generates a meeting report of the current meeting based on the full-process quality statistics, and stores the meeting report to the storage server.

[0101] In some embodiments, the terminal processor 601 further performs the following steps: constructing an initial large-scale meeting evaluation model for the meeting quality assessment scenario based on the basic prediction large-scale model; acquiring multiple sample sentiment state datasets, each of which carries a standard meeting quality assessment label; inputting the multiple sample sentiment state datasets into the initial large-scale meeting evaluation model to train the initial large-scale meeting evaluation model; during the training process of the initial large-scale meeting evaluation model, controlling the initial large-scale meeting evaluation model to output predicted meeting quality assessment labels for the multiple sample sentiment state datasets, and adjusting the parameters of the initial large-scale meeting evaluation model according to the predicted meeting quality assessment labels and the standard meeting quality assessment labels until the initial large-scale meeting evaluation model converges, thereby obtaining the trained large-scale meeting evaluation model.

[0102] In some embodiments, when the terminal processor 601 constructs an initial large-scale meeting evaluation model for a meeting quality assessment scenario based on a basic prediction model, it specifically performs the following steps: obtaining the basic prediction model; creating an initial large-scale meeting evaluation model that includes at least a meeting scenario adaptation module and a natural language generation module based on the basic prediction model; when the terminal processor 601 adjusts the parameters of the initial large-scale meeting evaluation model according to the predicted meeting quality assessment labels and the standard meeting quality assessment labels until the initial large-scale meeting evaluation model converges, it specifically performs the following steps: calculating the training loss value according to the predicted meeting quality assessment labels and the standard meeting quality assessment labels; adjusting the parameters of the meeting scenario adaptation module in the initial large-scale meeting evaluation model based on the training loss value; and keeping the parameters of the natural language generation module unchanged until the initial large-scale meeting evaluation model converges.

[0103] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0104] The modules described as separate components may or may not be physically separate. Similarly, the components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0105] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this specification are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The aforementioned available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., Digital Versatile Discs (DVDs)), or semiconductor media (e.g., Solid State Disks (SSDs)).

[0106] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0107] In addition, it should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, data stored, data displayed, etc.) and signals involved in the embodiments of this application are all authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0108] The foregoing has described specific embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0109] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0110] The above is a description of a meeting quality assessment method, apparatus, storage medium, and terminal provided in this application. For those skilled in the art, based on the ideas of the embodiments of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for assessing the quality of a meeting, characterized in that, The method includes: The audio and video data of the current meeting are periodically acquired, and the emotional representation information of at least one participant in the current meeting is determined based on the audio and video data. Based on the emotional representation information, the emotional state of each participant is identified, and the emotional state dataset corresponding to the current meeting is obtained; The emotional state dataset is input into the meeting evaluation model to determine the meeting quality evaluation result of the meeting evaluation model for the current meeting.

2. The method according to claim 1, characterized in that, When the attendee is a speaker, the speaker's emotional representation information includes at least one of the following: tone of voice, key words used in the speech, and facial expression. When the attendee is another person who is not speaking, the other person's emotional representation information includes at least one of the following: body language or facial expression.

3. The method according to claim 1, characterized in that, After determining the meeting quality assessment result of the meeting evaluation model for the current meeting, the method further includes: If the current meeting is determined to be of abnormal quality based on the meeting quality assessment results, a prompt message and intervention strategy corresponding to the meeting quality assessment results are generated, and the prompt message and intervention strategy are sent and displayed to the administrator terminal corresponding to the current meeting.

4. The method according to claim 1, characterized in that, The meeting quality assessment results shall include at least one of the following: meeting quality level, statistical data on the emotional categories of participants, and analysis report on the causes of meeting quality issues.

5. The method according to claim 1, characterized in that, The method further includes: After the current meeting ends, based on all meeting quality assessment results of the current meeting, generate the time sequence information of the meeting and the corresponding meeting quality assessment results, and calculate the full-process quality statistics of the current meeting; A meeting report for the current meeting is generated based on the overall quality statistics, and the meeting report is stored on a storage server.

6. The method according to claim 1, characterized in that, The method further includes: An initial large-scale meeting evaluation model for meeting quality assessment scenarios is constructed based on the basic prediction model. Multiple sample sentiment state datasets were obtained, all of which were labeled with standard meeting quality assessment tags. The multiple sample sentiment state datasets are input into the initial meeting evaluation model to train the initial meeting evaluation model; During the training process of the initial meeting evaluation model, the initial meeting evaluation model is controlled to output predicted meeting quality evaluation labels for the multiple sample sentiment state datasets, and the parameters of the initial meeting evaluation model are adjusted according to the predicted meeting quality evaluation labels and the standard meeting quality evaluation labels until the initial meeting evaluation model converges, thus obtaining the trained meeting evaluation model.

7. The method according to claim 6, characterized in that, The initial meeting evaluation model, constructed based on the fundamental prediction model for meeting quality assessment scenarios, includes: Obtain the basic prediction model; Based on the aforementioned basic prediction model, an initial meeting evaluation model is created, which includes at least a meeting scenario adaptation module and a natural language generation module. The step of adjusting the parameters of the initial large-scale meeting evaluation model based on the predicted meeting quality assessment labels and the standard meeting quality assessment labels until the initial large-scale meeting evaluation model converges includes: The training loss value is calculated based on the predicted meeting quality assessment label and the standard meeting quality assessment label. The parameters of the meeting scenario adaptation module in the initial meeting assessment model are adjusted based on the training loss value, while the parameters of the natural language generation module are kept unchanged until the initial meeting assessment model converges.

8. A meeting quality assessment device, characterized in that, The device includes: The information acquisition module is used to periodically acquire the audio and video data of the current meeting, and determine the emotional representation information of at least one participant in the current meeting based on the audio and video data. The emotion recognition module is used to identify the emotional state of each participant based on the emotion representation information, and obtain the emotion state dataset corresponding to the current meeting. The quality assessment module is used to input the emotional state dataset into the meeting assessment model and determine the meeting quality assessment result output by the meeting assessment model for the current meeting.

9. A computer storage medium, characterized in that, The computer storage medium stores a plurality of instructions adapted for loading by a processor and executing the steps of the method as described in any one of claims 1 to 7.

10. A terminal, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method as described in any one of claims 1 to 7.

Images