Periodic conference summary generation method and device, computer equipment and storage medium

Abstract

The embodiment of the invention belongs to the technical field of microphone systems, and relates to a periodic conference summary generation method and device, computer equipment and a storage medium, and the method comprises the steps: collecting a current audio signal of a current speaker through an annular microphone array or a multi-channel directional microphone when a conference starts; determining target participant information according to the current audio signal; inputting the current audio signal into an ASR model for voice conversion operation to obtain current audio text data; performing periodic summarization operation according to the current audio text data and the target participant information to obtain a periodic summarization framework; and outputting the periodic summary framework. The conference content can be accurately summarized in real time in the conference proceeding process, and powerful support and reference are provided for participants.

Description

Technical Field

[0001] The present application relates to the technical field of microphone systems, and in particular to a method, apparatus, computer equipment, and storage medium for generating periodic meeting summaries. Background Art

[0002] Current technologies for automatic speech recognition (ASR) and natural language processing (NLP) models are primarily based on deep learning. Core ASR technologies include end-to-end models (such as CTC and the Transformer architecture) and hybrid models (such as CTC / Attention Hybrid), which map speech signals into text sequences through acoustic feature extraction (such as Mel-spectrograms and MFCCs). NLP models rely on pre-trained language models (such as BERT and GPT) for semantic understanding, entity extraction, and intent recognition, and integrate knowledge graphs to enhance contextual relevance. In application scenarios, ASR and NLP technologies have been widely used in areas such as real-time conference transcription, customer service quality inspection, and live content review.

[0003] However, although existing technologies can complete speech transcription and basic analysis, they are limited by computing power allocation, model generalization capabilities, and the lack of real-time structuring technology in long-term, dynamic, and highly interactive scenarios, and cannot meet users' needs for real-time and accurate backtracking of the process. Summary of the Invention

[0004] The purpose of the embodiments of the present application is to propose a method, apparatus, computer equipment and storage medium for generating periodic meeting summaries, so as to solve the problem that the existing technology cannot meet the user's demand for real-time and accurate backtracking of the process.

[0005] In order to solve the above technical problems, the present application provides a method for generating periodic meeting summaries, which adopts the following technical solutions:

[0006] When the meeting starts, the current audio signal of the current speaker is collected through a ring microphone array or a multi-channel directional microphone;

[0007] confirming target participant information according to the current audio signal;

[0008] Inputting the current audio signal into the ASR model for voice conversion operation to obtain current audio text data;

[0009] Performing a periodic summary operation based on the current audio text data and the target participant information to obtain a periodic summary framework;

[0010] Output the periodic summary framework.

[0011] Furthermore, the step of confirming the target participant information according to the current audio signal specifically includes the following steps:

[0012] Obtaining current voiceprint feature information of the current audio signal;

[0013] Reading a system database, and obtaining pre-recorded participant voiceprint feature information in the system database;

[0014] The target participant information corresponding to the current voiceprint feature information is confirmed in the participant voiceprint feature information.

[0015] Furthermore, after the step of inputting the current audio signal into the ASR model for voice conversion to obtain the current audio text data, the following steps are also included:

[0016] Perform a text cleaning operation on the current audio text data.

[0017] Furthermore, the step of performing a periodic summary operation based on the current audio text data and the target participant information to obtain a periodic summary framework specifically includes the following steps:

[0018] Collect the current audio text data according to a preset duration;

[0019] Input the current audio text data into the BERT-CRF model to perform key information extraction operations to obtain named entity text;

[0020] Performing a logical relationship extraction operation on the named entity to obtain current hierarchical outline data;

[0021] The current hierarchical outline data is determined as the periodic summary framework.

[0022] Furthermore, the step of determining the current hierarchical outline data as the periodic summary framework specifically includes the following steps:

[0023] Determine whether there is historical hierarchical outline data;

[0024] If historical hierarchical outline data exists, performing an incremental update operation on the historical hierarchical outline data according to the current hierarchical outline data to obtain the periodic summary framework;

[0025] If no historical hierarchical outline data exists, the current hierarchical outline data is determined as the periodic summary framework.

[0026] Furthermore, if there is historical hierarchical outline data, the step of performing an incremental update operation on the historical hierarchical outline data according to the current hierarchical outline data to obtain the periodic summary framework specifically includes the following steps:

[0027] Comparing the current hierarchical outline data with the historical hierarchical outline data to determine newly added named entity text;

[0028] Calculating the semantic similarity between the newly added named entity text and the historical named entity text in the historical plan outline data respectively;

[0029] Mark the newly added named entity texts whose semantic similarity is greater than or equal to the preset threshold as sub-node content;

[0030] Mark the newly added named entity texts whose semantic similarity is less than a preset threshold as root node content or peer node content;

[0031] When all the newly added named entity texts are marked, the periodic summary framework is obtained.

[0032] In order to solve the above technical problems, the embodiment of the present application further provides a periodic meeting summary generation device, which adopts the following technical solution:

[0033] An audio signal acquisition module is used to collect the current audio signal of the current speaker through a ring microphone array or a multi-channel directional microphone when the meeting starts;

[0034] A participant confirmation module, configured to confirm target participant information based on the current audio signal;

[0035] A voice conversion module is used to input the current audio signal into the ASR model for voice conversion operation to obtain current audio text data;

[0036] A periodic summary module, configured to perform a periodic summary operation based on the current audio text data and the target participant information to obtain a periodic summary framework;

[0037] The framework output module is used to output the periodic summary framework.

[0038] Furthermore, the participant confirmation module includes:

[0039] A voiceprint feature acquisition submodule, configured to acquire current voiceprint feature information of the current audio signal;

[0040] A database reading submodule is used to read the system database and obtain the pre-recorded voiceprint feature information of the participants in the system database;

[0041] The participant confirmation submodule is used to confirm the target participant information corresponding to the current voiceprint feature information in the participant voiceprint feature information.

[0042] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0043] The method comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for generating periodic meeting summaries as described above are implemented.

[0044] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0045] The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the method for generating periodic meeting summaries as described above.

[0046] The present application provides a method for generating periodic meeting summaries, comprising: when a meeting begins, collecting the current audio signal of the current speaker through a circular microphone array or a multi-channel directional microphone; confirming the target participant information based on the current audio signal; inputting the current audio signal into an ASR model for speech conversion to obtain current audio text data; performing a periodic summary operation based on the current audio text data and the target participant information to obtain a periodic summary framework; and outputting the periodic summary framework. Compared with the existing technology, the present application can summarize the meeting content in real time and accurately during the meeting, providing strong support and reference for the participants. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the solutions in this application, a brief introduction will be given below to the drawings required for use in the description of the embodiments of this application. Obviously, the drawings described below are some embodiments of this application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0048] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0049] Figure 2 This is a flowchart of the implementation of the method for generating periodic meeting summaries provided in an embodiment of the present application;

[0050] Figure 3 This is a schematic diagram of an embodiment of the LLM structured processing return structure result provided by an embodiment of the present application;

[0051] Figure 4 This is a schematic diagram of an embodiment of an output periodic summary framework provided by an embodiment of the present application;

[0052] Figure 5 1 is a schematic diagram of an embodiment of generating an updated MindMap structure provided in an embodiment of the present application;

[0053] Figure 6 2 is a schematic diagram of another embodiment of the output periodic summary framework provided by an embodiment of the present application;

[0054] Figure 7 This is a schematic diagram of an embodiment of generating a modified MindMap and marking changes provided by an embodiment of the present application;

[0055] Figure 8 2 is a schematic diagram of another embodiment of the output periodic summary framework provided in an embodiment of the present application;

[0056] Figure 9 This is a schematic diagram of the structure of a periodic meeting summary generation device provided in an embodiment of the present application;

[0057] Figure 10 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0058] Unless otherwise defined, all technical and scientific terms used herein have the same meanings as commonly understood by those skilled in the art to which this application belongs. The terms used in the specification of the application are for the purpose of describing specific embodiments only and are not intended to limit this application. The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of this application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0059] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0060] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0061] like Figure 1 As shown, system architecture 100 may include a terminal device 101, a network 102, and a server 103. Terminal device 101 may be a laptop computer 1011, a tablet computer 1012, or a mobile phone 1013. Network 102 is a medium for providing a communication link between terminal device 101 and server 103. Network 102 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0062] The user can use the terminal device 101 to interact with the server 103 via the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0063] The terminal device 101 can be various electronic devices with a display screen and supporting web browsing. In addition to the laptop computer 1011, tablet computer 1012 or mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV) player, a laptop computer and a desktop computer, etc.

[0064] The server 103 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal device 101 .

[0065] It should be noted that the periodic meeting summary generation method provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the periodic meeting summary generation device is generally set in the server / terminal device.

[0066] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0067] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for generating periodic meeting summaries according to the present application. The method for generating periodic meeting summaries includes: step S201, step S202, step S203, step S204 and step S205.

[0068] In step S201 , when a meeting starts, a current audio signal of a current speaker is collected by a circular microphone array or a multi-channel directional microphone.

[0069] In the embodiments of the present application, the microphone array cited in the present application includes: a ring microphone array (such as the ReSpeaker series) or a multi-channel directional microphone.

[0070] In some optional implementations of the embodiments of the present application, when multiple people speak at the same time, a speech separation operation can be performed based on the Conv-TasNet technology to obtain audio signals corresponding to each speaker.

[0071] In step S202, target participant information is confirmed based on the current audio signal.

[0072] In an embodiment of the present application, confirming the target participant information based on the current audio signal can be obtaining the current voiceprint feature information of the current audio signal; reading the system database, obtaining the pre-recorded participant voiceprint feature information in the system database; and confirming the target participant information corresponding to the current voiceprint feature information in the participant voiceprint feature information.

[0073] In step S203, the current audio signal is input into the ASR model for speech conversion to obtain current audio text data.

[0074] In an embodiment of the present application, the ASR model can convert the acquired audio signal into text content, wherein the ASR model can be a streaming speech recognition model such as Google Cloud Speech-to-Text Streaming API / open source model (DeepSpeech 2.0). The ASR model can also be a Wav2Vec 2.0+CTC decoding model. It should be understood that the examples of the ASR model here are only for convenience of understanding and are not used to limit this application.

[0075] In step S204, a periodic summary operation is performed based on the current audio text data and the target participant information to obtain a periodic summary framework.

[0076] In an embodiment of the present application, the periodic summary operation can be to collect current audio text data according to a preset duration; input the current audio text data into the BERT-CRF model to perform a key information extraction operation to obtain a named entity text; perform a logical relationship extraction operation on the named entity to obtain the current hierarchical outline data; and determine the current hierarchical outline data as a periodic summary framework.

[0077] In step S205 , a periodic summary frame is output.

[0078] In an embodiment of the present application, the output periodic summary framework may be a dynamic rendering mind map and support interaction, specifically:

[0079] Front-end framework: Vue.js + D3.js (force-directed graph) / ECharts;

[0080] Projection protocol: WebSocket real-time push / HDM I direct connection;

[0081] Interactive functions: gesture recognition (Leap Motion) / stylus annotation / touch screen / keyboard and mouse input.

[0082] In an embodiment of the present application, a method for generating periodic meeting summaries is provided, comprising: when a meeting begins, collecting the current audio signal of the current speaker through a circular microphone array or a multi-channel directional microphone; confirming the target participant information based on the current audio signal; inputting the current audio signal into an ASR model for speech conversion to obtain current audio text data; performing a periodic summary operation based on the current audio text data and the target participant information to obtain a periodic summary framework; and outputting the periodic summary framework. Compared with the existing technology, the present application can summarize the meeting content in real time and accurately during the meeting, providing strong support and reference for participants.

[0083] In some optional implementations of the embodiments of the present application, the step of confirming the target participant information based on the current audio signal specifically includes the following steps:

[0084] Get the current voiceprint feature information of the current audio signal;

[0085] Read the system database and obtain the pre-recorded voiceprint feature information of the participants in the system database;

[0086] Confirm the target participant information corresponding to the current voiceprint feature information in the participant voiceprint feature information.

[0087] In an embodiment of the present application, voiceprint recognition operations can be performed through VGGVox, ResNet-ASV (based on PyTorch / TensorFlow), etc. to obtain the current voiceprint feature information of the current audio signal.

[0088] In the embodiment of the present application, the present application will perform voiceprint registration in advance. Specifically, the voiceprint of the participants will be pre-recorded. During the registration stage, the user's voiceprint feature information can be collected based on the user reading a fixed text.

[0089] In an embodiment of the present application, beamforming and voiceprint matching technology can be combined to confirm the target participant information corresponding to the current voiceprint feature information in the participant's voiceprint feature information.

[0090] Compared with the existing technology, the present application can identify different speakers through their voice features and mark audio clips.

[0091] In some optional implementations of the embodiments of the present application, after the step of inputting the current audio signal into the ASR model for speech conversion to obtain the current audio text data, the following steps are further included:

[0092] Perform text cleaning operations on the current audio text data.

[0093] In an embodiment of the present application, the text cleaning operation may be removing stop words, correcting ASR errors, etc.

[0094] In some optional implementations of the embodiments of the present application, the above-mentioned step of performing a periodic summary operation based on the current audio text data and the target participant information to obtain a periodic summary framework specifically includes the following steps:

[0095] Collect current audio text data according to the preset duration;

[0096] Input the current audio text data into the BERT-CRF model to extract key information and obtain named entity text;

[0097] Perform logical relationship extraction on named entities to obtain the current hierarchical outline data;

[0098] Identify the current hierarchical outline data as a periodic summary framework.

[0099] In the embodiment of the present application, the periodic summary operation includes:

[0100] 1. Receive structured semantic data:

[0101] Data source: The cleaned text output by the NLP module contains structured information such as entities (person, time, task), topic tags (such as "requirements discussion", "technical solution"), and intent classification (decision, question, suggestion).

[0102] Data format: JSON format, with timestamp and speaker ID.

[0103] Flow direction: Structured data is transmitted to the real-time processing channel of the summary framework engine through a message queue (such as Kafka).

[0104] 2. Dynamically adjust logical relationships

[0105] Dependency analysis: Automatically identify the causal relationship between nodes (e.g. "module development completion" depends on "interface document review").

[0106] Conflict detection: Compare task descriptions from different speakers and flag inconsistencies (e.g., two people making different suggestions for the same deadline).

[0107] Data flow: Relational data (dependency / conflict edge lists) are merged with the framework structure to form a complete logical graph.

[0108] In some optional implementations of the embodiments of the present application, the above step of determining the current hierarchical outline data as a periodic summary framework specifically includes the following steps:

[0109] Determine whether there is historical hierarchical outline data;

[0110] If there is historical hierarchical outline data, then incrementally update the historical hierarchical outline data according to the current hierarchical outline data to obtain a periodic summary framework;

[0111] If the historical hierarchical outline data does not exist, the current hierarchical outline data is determined as the periodic summary framework.

[0112] In the embodiment of the present application, the above-mentioned periodic summary operation further includes:

[0113] 3. Incremental update framework structure:

[0114] Add new nodes: Insert unmatched content as new nodes and generate a hierarchy in the order of discussion (such as "Requirement Discussion → Sub-Function A → Interface Definition").

[0115] Supplementary attributes: Add details to existing nodes (such as adding a responsible person and deadline to the "Task Assignment" node).

[0116] Status tagging: Update node status based on intent classification (e.g., “in dispute,” “confirmed”).

[0117] Data flow: The updated framework structure (JSON tree) is stored in an in-memory database (such as Redis) and pushed to the version control service.

[0118] In some optional implementations of the embodiments of the present application, if historical hierarchical outline data exists, the step of performing an incremental update operation on the historical hierarchical outline data according to the current hierarchical outline data to obtain a periodic summary framework specifically includes the following steps:

[0119] Compare the current hierarchical outline data with the historical hierarchical outline data to determine the newly added named entity text;

[0120] Calculate the semantic similarity between the newly added named entity text and the historical named entity text in the historical plan outline data respectively;

[0121] Mark the newly added named entity texts whose semantic similarity is greater than or equal to the preset threshold as child node content;

[0122] Mark the newly added named entity texts whose semantic similarity is less than a preset threshold as root node content or peer node content;

[0123] When all newly added named entity texts are marked, a periodic summary framework is obtained.

[0124] In an embodiment of the present application, the above-mentioned periodic summary operation also includes: comparing the new data with the node content in the existing framework, and determining whether they belong to the same topic branch through semantic similarity calculation (such as cosine similarity). If the match is successful (similarity threshold > 70%), it is marked as the sub-content of the existing node; if it is a new topic, a root node or a peer node is created. Data flow: The association result (matching node ID or new node instruction) is passed to the framework update module.

[0125] In actual application, if a meeting starts, A, B, and C will speak in turn. When A speaks and sets the meeting theme, the display device will display the meeting theme framework (mind map) and organize A's speech content in real time within the framework. When B and C speak later, the corresponding speech content will be organized and displayed within the framework. During the entire meeting, participants can review the corresponding content of the three speeches at any time. When content adjustments are required, the adjusted content can also be displayed in real time.

[0126] Specific system processing flow:

[0127] 1. Meeting start and A's speech (initial framework generation)

[0128] Step 1: Audio acquisition and preprocessing:

[0129] The audio stream of A's speech is captured by the microphone, buffered in 30-second slices, and converted to standard PCM format, ready for input into the STT service. Step 2: Speech to Text

[0130] Use a cloud-based STT engine (such as Volcano Engine) to convert the first 30 seconds of A's speech into text.

[0131] Example output:

[0132] [A] The theme of this meeting: 2024 product roadmap, focusing on feature prioritization and resource allocation.

[0133] Step 2: LLM Structuring Processing:

[0134] Attached Prompt1 (generate Markdown meeting minutes) and Prompt2 (generate MindMap framework).

[0135] The returned structure result is as follows Figure 3 shown.

[0136] Step 3: Visualization Rendering:

[0137] The front end receives the JSON data, renders it as a mind map root node ("2024 Product Roadmap"), and expands the two initial child nodes ("Function Priority" and "Resource Allocation"), such as Figure 4 shown.

[0138] 2. B's speech (following up A's speech to supplement and elaborate on the content):

[0139] Specific process:

[0140] Step 1: Audio Processing:

[0141] The audio of B's speech was sliced and converted into text: "In terms of function priority, it is recommended to give priority to the development of the AI assistant module, which requires 3 developers."

[0142] Step 2: LLM processing:

[0143] Merge the historical text of A with the new content of B to generate an updated MindMap structure such as Figure 5 shown.

[0144] Step 3: Incremental rendering:

[0145] The front end detects the addition of a new sub-node "AI Assistant Module" under the "Function Priority" node and expands the branch with animation, such as Figure 6 shown.

[0146] 3. C's speech (conflict marking and dynamic correction)

[0147] Processing Flow

[0148] Step 1: Audio Processing:

[0149] Text of C's speech: "Resource allocation needs to be adjusted. The AI assistant module should be assigned 5 people. The original plan of 3 people is insufficient."

[0150] Step 2: LLM processing:

[0151] Identify resource allocation conflicts, generate a revised MindMap and mark the changes as Figure 7 shown.

[0152] Step 3: Visualization Update:

[0153] The value of the "AI Assistant Module" node changes from 3 to 5, and the background color changes to yellow (conflict marker).

[0154] Add a note "2 more people needed" under the "Resource Allocation" node, such as Figure 8 shown.

[0155] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results.

[0156] Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0157] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing related hardware via computer-readable instructions. The computer-readable instructions can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes in the above-described method embodiments. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0158] It should be understood that although the steps in the flowcharts of the accompanying drawings are shown in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some of the steps in the flowcharts of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0159] Further references Figure 9 , as a response to the above Figure 2 The present application provides an embodiment of a device for generating periodic meeting summaries. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0160] like Figure 9 As shown, the periodic meeting summary generating apparatus 200 of the embodiment of the present application includes:

[0161] The audio signal collection module 210 is used to collect the current audio signal of the current speaker through a circular microphone array or a multi-channel directional microphone when the meeting starts;

[0162] Participant confirmation module 220, used to confirm target participant information based on the current audio signal;

[0163] The speech conversion module 230 is used to input the current audio signal into the ASR model for speech conversion operation to obtain the current audio text data;

[0164] A periodic summarizing module 240 is configured to perform a periodic summarizing operation based on the current audio text data and target participant information to obtain a periodic summarizing framework;

[0165] The framework output module 250 is used to output the periodic summary framework.

[0166] In an embodiment of the present application, a periodic meeting summary generation device 200 is provided, including: an audio signal acquisition module 210, which is used to collect the current audio signal of the current speaker through a circular microphone array or a multi-channel directional microphone when the meeting starts; a participant confirmation module 220, which is used to confirm the target participant information based on the current audio signal; a voice conversion module 230, which is used to input the current audio signal into the ASR model for voice conversion operation to obtain current audio text data; a periodic summary module 240, which is used to perform periodic summary operations based on the current audio text data and the target participant information to obtain a periodic summary framework; a framework output module 250, which is used to output the periodic summary framework. Compared with the existing technology, the present application can summarize the meeting content in real time and accurately during the meeting, providing strong support and reference for the participants.

[0167] In some optional implementations of the embodiments of the present application, the participant confirmation module includes:

[0168] The voiceprint feature acquisition submodule is used to obtain the current voiceprint feature information of the current audio signal;

[0169] The database reading submodule is used to read the system database and obtain the pre-recorded voiceprint feature information of the participants in the system database;

[0170] The participant confirmation submodule is used to confirm the target participant information corresponding to the current voiceprint feature information in the participant voiceprint feature information.

[0171] To solve the above technical problems, the present application also provides a computer device. Figure 10 , Figure 10 This is a basic structural block diagram of the computer device according to an embodiment of the present application.

[0172] The computer device 300 includes a memory 310, a processor 320, and a network interface 330 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 300 having components 310-330, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0173] The computer device may be a desktop computer, notebook computer, PDA, cloud server, etc. The computer device may interact with the user via a keyboard, mouse, remote control, touchpad, or voice control device.

[0174] The memory 310 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic storage, magnetic disk, optical disk, etc. In some embodiments, the memory 310 may be an internal storage unit of the computer device 300, such as a hard disk or memory of the computer device 300. In other embodiments, the memory 310 may also be an external storage device of the computer device 300, such as a plug-in hard disk equipped on the computer device 300, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Of course, the memory 310 may also include both the internal storage unit of the computer device 300 and its external storage device. In the embodiment of the present application, the memory 310 is generally used to store the operating system and various application software installed on the computer device 300, such as computer-readable instructions for the method for generating periodic meeting summaries. In addition, the memory 310 can also be used to temporarily store various data that has been output or is about to be output.

[0175] In some embodiments, the processor 320 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor 320 is generally used to control the overall operation of the computer device 300. In the embodiment of the present application, the processor 320 is used to execute computer-readable instructions or process data stored in the memory 310, such as computer-readable instructions for executing the periodic meeting summary generation method.

[0176] The network interface 330 may include a wireless network interface or a wired network interface. The network interface 330 is generally used to establish a communication connection between the computer device 300 and other electronic devices.

[0177] The computer equipment provided by this application can summarize the meeting content in real time and accurately during the meeting, providing strong support and reference for participants.

[0178] The present application also provides another embodiment, namely, providing a computer-readable storage medium, which stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the periodic meeting summary generation method as described above.

[0179] The computer-readable storage medium provided in this application can summarize the content of the meeting in real time and accurately during the meeting, providing strong support and reference for participants.

[0180] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0181] Obviously, the embodiments described above are only some of the embodiments of the present application, rather than all of the embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions described in the aforementioned specific embodiments, or to make equivalent replacements for some of the technical features therein. Any equivalent structure made using the contents of the present application specification and the accompanying drawings, directly or indirectly used in other related technical fields, is also within the scope of patent protection of the present application.

Claims

1. A method for generating periodic meeting summaries, characterized in that: The steps include: When the meeting starts, the current audio signal of the current speaker is collected through a ring microphone array or a multi-channel directional microphone; confirming target participant information according to the current audio signal; Inputting the current audio signal into the ASR model for voice conversion operation to obtain current audio text data; Performing a periodic summary operation based on the current audio text data and the target participant information to obtain a periodic summary framework; Output the periodic summary framework.

2. The method for generating periodic meeting summaries according to claim 1, characterized in that: The step of confirming the target participant information according to the current audio signal specifically includes the following steps: Obtaining current voiceprint feature information of the current audio signal; Reading a system database, and obtaining pre-recorded participant voiceprint feature information in the system database; The target participant information corresponding to the current voiceprint feature information is confirmed in the participant voiceprint feature information.

3. The method for generating periodic meeting summaries according to claim 1, characterized in that: After the step of inputting the current audio signal into the ASR model for voice conversion operation to obtain the current audio text data, the method further includes the following steps: Perform a text cleaning operation on the current audio text data.

4. The method for generating periodic meeting summaries according to claim 1, characterized in that: The step of performing a periodic summary operation based on the current audio text data and the target participant information to obtain a periodic summary framework specifically includes the following steps: Collect the current audio text data according to a preset duration; Input the current audio text data into the BERT-CRF model to perform key information extraction operations to obtain named entity text; Performing a logical relationship extraction operation on the named entity to obtain current hierarchical outline data; The current hierarchical outline data is determined as the periodic summary framework.

5. The method for generating periodic meeting summaries according to claim 4, characterized in that: The step of determining the current hierarchical outline data as the periodic summary framework specifically includes the following steps: Determine whether there is historical hierarchical outline data; If historical hierarchical outline data exists, performing an incremental update operation on the historical hierarchical outline data according to the current hierarchical outline data to obtain the periodic summary framework; If no historical hierarchical outline data exists, the current hierarchical outline data is determined as the periodic summary framework.

6. The method for generating periodic meeting summaries according to claim 5, characterized in that: If the historical hierarchical outline data exists, the step of performing an incremental update operation on the historical hierarchical outline data according to the current hierarchical outline data to obtain the periodic summary framework specifically includes the following steps: Comparing the current hierarchical outline data with the historical hierarchical outline data to determine newly added named entity text; Calculating the semantic similarity between the newly added named entity text and the historical named entity text in the historical plan outline data respectively; Mark the newly added named entity texts whose semantic similarity is greater than or equal to the preset threshold as child node content; Mark the newly added named entity texts whose semantic similarity is less than a preset threshold as root node content or peer node content; When all the newly added named entity texts are marked, the periodic summary framework is obtained.

7. A periodic meeting summary generation device, characterized in that: include: An audio signal acquisition module is used to collect the current audio signal of the current speaker through a ring microphone array or a multi-channel directional microphone when the meeting starts; A participant confirmation module, configured to confirm target participant information based on the current audio signal; A voice conversion module is used to input the current audio signal into the ASR model for voice conversion operation to obtain current audio text data; A periodic summary module, configured to perform a periodic summary operation based on the current audio text data and the target participant information to obtain a periodic summary framework; The framework output module is used to output the periodic summary framework.

8. The periodic meeting summary generating device according to claim 7, characterized in that: The participant confirmation module includes: A voiceprint feature acquisition submodule, configured to acquire current voiceprint feature information of the current audio signal; A database reading submodule is used to read the system database and obtain the pre-recorded voiceprint feature information of the participants in the system database; The participant confirmation submodule is used to confirm the target participant information corresponding to the current voiceprint feature information in the participant voiceprint feature information.

9. A computer device comprising a memory and a processor, characterized in that: The memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for generating a periodic meeting summary according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the method for generating a periodic meeting summary according to any one of claims 1 to 6.

Images