Conference recording method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a conference recording method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring audio data of a target conference; determining a target conference template corresponding to the audio data according to the audio data and a target conference arrangement model; the target conference template comprises a plurality of structured conference record items; and according to the audio data and the target conference arrangement model, respectively determining first text data corresponding to the plurality of structured conference record items, and respectively filling the plurality of structured conference record items with the first text data. Therefore, the conference content can be organized and recorded through the target conference organizing model, and the conference content does not need to be summarized and organized by spending time and energy.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a conference recording method, a conference recording device, an electronic device and a computer-readable storage medium. Background Art

[0002] In daily office work or communication, it is often necessary to communicate through meetings. During the meeting, it is often necessary to record the meeting for subsequent summary or review of the meeting. When related technologies record meetings, they are generally recorded manually, which is cumbersome and difficult, or recorded through voice recognition, which is only used in a single scenario. After the meeting, it takes time and effort to summarize and organize the content of the meeting. Summary of the invention

[0003] In view of the above problems, embodiments of the present invention are proposed to provide a meeting recording method, a meeting recording device, an electronic device and a computer-readable storage medium that overcome the above problems or at least partially solve the above problems.

[0004] In order to solve the above problem, an embodiment of the present invention discloses a conference recording method, which includes:

[0005] Get the audio data of the target conference;

[0006] Determining a target conference template corresponding to the audio data according to the audio data and the target conference arrangement model; the target conference template includes a plurality of structured conference record items;

[0007] According to the audio data and the target conference organizing model, first text data corresponding to the multiple structured conference record items are respectively determined, and the first text data are respectively filled into the multiple structured conference record items.

[0008] Optionally, the target conference arrangement model is trained in the following manner:

[0009] Get sample data of meeting minutes;

[0010] The preset conference arrangement model is trained according to the conference record sample data to obtain a trained conference arrangement model; the trained conference arrangement model is used to output a corresponding target conference template according to the audio data;

[0011] Obtaining conference history text sample data and corresponding record template sample data;

[0012] The trained meeting arrangement model is trained according to the meeting history text sample data and the corresponding record template sample data to obtain a target meeting arrangement model.

[0013] Optionally, after filling the first text data into the plurality of structured meeting record items respectively, the method further includes:

[0014] Acquire a user's modification operation on the target conference arrangement model to obtain a modified target conference arrangement model;

[0015] Based on the audio data and the revised target conference organizing model, the target conference template corresponding to the audio data and the first text data corresponding to the multiple structured meeting record items are determined again, and the first text data are filled into the multiple structured meeting record items respectively.

[0016] Optionally, after filling the first text data into the plurality of structured meeting record items respectively, the method further includes:

[0017] Performing vectorization processing on the first text data respectively filled in the plurality of structured meeting record items to obtain meeting vector data, and saving the meeting vector data to a vector database;

[0018] Get the question voice input by the user;

[0019] Determining a question text corresponding to the question speech according to a preset understanding model;

[0020] Performing vectorization processing on the question text to obtain question vector data;

[0021] matching target meeting vector data corresponding to the question vector data from the vector database;

[0022] Determine the target text data corresponding to the target conference vector data, and output the target text data.

[0023] Optionally, after determining the target text data corresponding to the target conference vector data, the method further includes:

[0024] According to a preset conversion model, the target text data is converted into corresponding voice data, and the voice data is output.

[0025] Optionally, after obtaining the question voice input by the user, the method further includes:

[0026] Verifying the identity of the user according to the question voice;

[0027] If the user's identity verification is passed, the question text corresponding to the question speech is determined according to a preset understanding model.

[0028] Optionally, verifying the identity of the user according to the question voice includes:

[0029] Acquire voiceprint data of the problematic speech;

[0030] Determining whether the similarity between the voiceprint data and a preset voiceprint is greater than a preset similarity;

[0031] If the similarity between the voiceprint data and the preset voiceprint is greater than the preset similarity, it is determined that the identity verification of the user has passed.

[0032] Optionally, after acquiring the audio data of the target conference, the method further includes:

[0033] Determining second text data corresponding to the audio data;

[0034] The second text data is marked according to the time of the target meeting, so that the user can trace back the corresponding second text data according to the mark.

[0035] Optionally, after filling the first text data into the plurality of structured meeting record items respectively, the method further includes:

[0036] Obtaining user satisfaction with the target conference organization model;

[0037] When the satisfaction level is less than a preset satisfaction level, the target meeting organization model is optimized.

[0038] The embodiment of the present invention discloses a conference recording device, the device comprising:

[0039] A first acquisition module, used to acquire audio data of a target conference;

[0040] A first determination module is used to determine a target conference template corresponding to the audio data according to the audio data and the target conference arrangement model; the target conference template includes a plurality of structured conference record items;

[0041] The second determination module is used to determine the first text data corresponding to the multiple structured meeting record items according to the audio data and the target meeting arrangement model, and fill the first text data into the multiple structured meeting record items respectively.

[0042] Optionally, the target conference arrangement model is trained in the following manner:

[0043] The second acquisition module is used to acquire sample data of meeting records;

[0044] A first training module is used to train a preset conference arrangement model according to the conference record sample data to obtain a trained conference arrangement model; the trained conference arrangement model is used to output a corresponding target conference template according to the audio data;

[0045] The third acquisition module is used to acquire the conference history text sample data and the corresponding record template sample data;

[0046] The second training module is used to train the trained meeting arrangement model according to the meeting history text sample data and the corresponding record template sample data to obtain a target meeting arrangement model.

[0047] Optionally, the device further comprises:

[0048] A correction module, used to obtain a user's correction operation on the target conference arrangement model, and obtain a corrected target conference arrangement model;

[0049] The third determination module is used to re-determine the target conference template corresponding to the audio data and the first text data corresponding to the multiple structured meeting record items based on the audio data and the revised target conference organization model, and fill the first text data into the multiple structured meeting record items respectively.

[0050] Optionally, the device further comprises:

[0051] A first processing module, configured to perform vectorization processing on the first text data respectively filled in the plurality of structured meeting record items to obtain meeting vector data, and save the meeting vector data to a vector database;

[0052] The fourth acquisition module is used to acquire the question voice input by the user;

[0053] A fourth determination module, used to determine the question text corresponding to the question speech according to a preset understanding model;

[0054] A second processing module is used to perform vectorization processing on the question text to obtain question vector data;

[0055] A matching module, used for matching target conference vector data corresponding to the question vector data from the vector database;

[0056] The fifth determination module is used to determine the target text data corresponding to the target conference vector data and output the target text data.

[0057] Optionally, the device further comprises:

[0058] The conversion module is used to convert the target text data into corresponding voice data according to a preset conversion model, and output the voice data.

[0059] Optionally, the device further comprises:

[0060] A verification module, used to verify the identity of the user according to the question voice;

[0061] The sixth determination module is used to determine the question text corresponding to the question voice according to a preset understanding model if the identity verification of the user passes.

[0062] Optionally, the verification module includes:

[0063] An acquisition submodule, used to acquire the voiceprint data of the problematic speech;

[0064] A judgment submodule, used to judge whether the similarity between the voiceprint data and a preset voiceprint is greater than a preset similarity;

[0065] The determination submodule is used to determine that the identity verification of the user has passed if the similarity between the voiceprint data and the preset voiceprint is greater than the preset similarity.

[0066] Optionally, the device further comprises:

[0067] A seventh determination module, configured to determine second text data corresponding to the audio data;

[0068] The marking module is used to mark the second text data according to the time of the target meeting, so that the user can trace the corresponding second text data according to the mark.

[0069] Optionally, the device further comprises:

[0070] A fifth acquisition module, used to acquire user satisfaction with the target conference arrangement model;

[0071] The optimization module is used to optimize the target meeting arrangement model when the satisfaction level is less than a preset satisfaction level.

[0072] The present invention also discloses an air conditioner, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned meeting recording method when executing the computer program.

[0073] The present invention also discloses a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the conference recording method as described above are implemented.

[0074] The embodiments of the present invention include the following advantages:

[0075] In an embodiment of the present invention, audio data of the target meeting is first obtained, and then a target meeting template corresponding to the audio data is determined based on the audio data and the target meeting organization model, wherein the target meeting template includes multiple structured meeting record items, and finally, first text data corresponding to the multiple structured meeting record items are respectively determined based on the audio data and the target meeting organization model, and the first text data is respectively filled into the multiple structured meeting record items, so that the content of the meeting can be organized and recorded through the target meeting organization model, without spending time and energy to summarize and organize the content of the meeting. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 is a flowchart of the steps of a conference recording method provided by an embodiment of the present invention;

[0077] Figure 2 is a flowchart of another method for recording a meeting provided by an embodiment of the present invention;

[0078] Figure 3 It is a flowchart of the steps of an interactive method for conference records provided by an embodiment of the present invention;

[0079] Figure 4 It is a structural block diagram of a conference recording device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0080] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0081] In the related art, when recording a meeting, it is generally recorded manually or through voice recognition. Both methods require time and effort to summarize and organize the content of the meeting. In order to solve the above technical problems, the present invention discloses a meeting recording method, which first obtains the audio data of the target meeting, and then determines the target meeting template corresponding to the audio data based on the audio data and the target meeting arrangement model, wherein the target meeting template includes multiple structured meeting record items, and finally determines the first text data corresponding to the multiple structured meeting record items based on the audio data and the target meeting arrangement model, and fills the first text data into the multiple structured meeting record items respectively, and then the content of the meeting can be organized and recorded through the target meeting arrangement model, without spending time and effort to summarize and organize the content of the meeting.

[0082] Reference Figure 1, shows a flowchart of a method for recording a meeting provided by an embodiment of the present invention, and the method may specifically include the following steps:

[0083] Step 101, obtaining audio data of the target conference.

[0084] In an embodiment of the present invention, when arranging the target meeting, the audio data of the target meeting can be obtained first. The target meeting in this application can be a meeting in a public scene such as a business negotiation, a multi-person meeting, a project report, or a personal record, or a meeting in a scene built by the user himself. The specific scene of the target meeting can be set according to actual needs, and the present invention does not make specific limitations here.

[0085] In an embodiment of the present invention, after obtaining the audio data of the target meeting, the method may also include: determining second text data corresponding to the audio data; marking the second text data according to the time of the target meeting so that the user can trace the corresponding second text data according to the mark.

[0086] Specifically, after obtaining the audio data of the target meeting, the audio data of the target meeting can be converted into the corresponding second text data through a preset language model, wherein the second text data is the text data corresponding to the entire target meeting. After obtaining the second text data corresponding to the audio data, the second text data can be marked according to the time of the target meeting, for example, the time of each sentence corresponding to the text data can be marked according to the time of the target meeting. In this way, when the user needs to view the entire content of the target meeting, or needs to trace the related second text data, he can trace it according to the marked time.

[0087] Step 102: Determine a target conference template corresponding to the audio data according to the audio data and the target conference arrangement model; the target conference template includes a plurality of structured conference record items.

[0088] In an embodiment of the present invention, after the audio data of the target meeting is acquired, the audio data can be input into a trained target meeting organization model, and the target meeting organization model can output a target meeting module corresponding to the audio data. In the present invention, the target meeting organization model can output different target meeting modules according to different audio data, and different audio data represent different meeting scenes of the target meeting. The target meeting organization model can output target meeting templates corresponding to different meeting scenes. Among them, the target meeting model can include multiple structured meeting record items, and the multiple structured meeting record items can include: meeting topic, meeting time, participants, meeting content and meeting summary, etc.

[0089] In an embodiment of the present invention, the target meeting organization model is trained in the following manner: obtaining meeting record sample data; training a preset meeting organization model based on the meeting record sample data to obtain a trained meeting organization model; the trained meeting organization model is used to output a corresponding target meeting template based on audio data; obtaining meeting history text sample data and corresponding record template sample data; training the trained meeting organization model based on the meeting history text sample data and the corresponding record template sample data to obtain a target meeting organization model.

[0090] Specifically, first obtain meeting record sample data, which can be sample data of meeting minutes based on industries and meeting types, or sample data of meeting records related to specific industries obtained through interviews, network downloads, or purchases. Then, train the preset meeting organization model based on the meeting record sample data to obtain a trained meeting organization model. In the present invention, the obtained trained meeting organization model is used to output the corresponding target meeting template according to the audio data, that is, the target meeting template corresponding to different audio data can be determined according to the trained meeting organization model. Because the content of the audio data obtained for different meetings is not the same, the determined target meeting template is also different. After obtaining the trained meeting organization model, the meeting history text sample data and the corresponding record template sample data can be obtained again, and the trained meeting organization model can be trained again based on the meeting history text sample data and the corresponding record template sample data to obtain the target meeting organization model. The present invention trains the trained meeting organization model again, and its purpose is to automatically fill the content in the target meeting template through the target meeting organization model. When the trained meeting organization model is trained again, the keywords and high-frequency words for related tasks, plans, and key points recognition in the target meeting template can be optimized. For example, keywords such as project name, project cycle, project leader, project team, project goals, etc. that appear in the project report meeting can be marked, and the output of tasks and plans for keywords such as completion time can be set.

[0091] Step 103: Determine first text data corresponding to the multiple structured meeting record items respectively according to the audio data and the target meeting arrangement model, and fill the first text data into the multiple structured meeting record items respectively.

[0092] In an embodiment of the present invention, after the target conference arrangement model determines the target conference template corresponding to the audio data, the target conference arrangement model can further determine the first text data corresponding to multiple structured meeting record items based on the audio data, and then automatically fill the first text data into the multiple structured meeting record items, so that a complete meeting report can be output. In the present invention, after the first text data is filled into multiple structured meeting record items, a complete arranged meeting report can be obtained, and then the complete meeting report and the corresponding audio data are stored in a database, so that the user can view the relevant meeting report, or iterate or optimize the target meeting arrangement model based on the meeting report and audio data in the database.

[0093] In an embodiment of the present invention, after the first text data is filled into multiple structured meeting record items respectively, it can also include: obtaining the user's modification operation on the target meeting organization model to obtain the modified target meeting organization model; based on the audio data and the modified target meeting organization model, re-determining the target meeting template corresponding to the audio data and the first text data corresponding to the multiple structured meeting record items, and filling the first text data into the multiple structured meeting record items respectively.

[0094] Specifically, after the first text data is filled into the multiple structured meeting record items respectively, the user can judge whether it meets the requirements or whether it meets the meeting content according to the filled content. If the user feels that the filled content does not meet the requirements, or deviates from the content of the meeting, or there is content that is not filled, the user can manually modify the content in the meeting record item, or perform correction operations on the target meeting arrangement model to make the target meeting arrangement model more perfect. After obtaining the corrected target meeting arrangement model, the target meeting template corresponding to the audio data can be determined again, and the first text data corresponding to the multiple structured meeting record items can be determined again based on the audio data, and the first text data corresponding to the multiple structured meeting record items can be filled into the multiple structured meeting record items respectively. Make the obtained meeting report more complete and accurate.

[0095] In an embodiment of the present invention, after the first text data is respectively filled into the multiple structured meeting record items, it also includes: vectorizing the first text data respectively filled in the multiple structured meeting record items to obtain meeting vector data, and saving the meeting vector data to a vector database; obtaining the question voice input by the user; determining the question text corresponding to the question voice according to a preset understanding model; vectorizing the question text to obtain question vector data; matching the target meeting vector data corresponding to the question vector data from the vector database; determining the target text data corresponding to the target meeting vector data, and outputting the target text data.

[0096] Specifically, after the first text data is filled into the multiple structured meeting record items respectively, the first text data filled into the multiple structured meeting record items respectively can be vectorized, and the vectorized first text data can be stored in the vector database. When the user asks questions about the relevant content of the target meeting, the user can input a question voice, preset the understanding model, convert the question voice into question text, and the question text can be vectorized. Then, the similarity between the vectorized question text and the vectorized first text data is calculated to obtain the target text data corresponding to the vectorized question text and output it. That is, when the user asks questions about the relevant content of the target meeting, the answer corresponding to the user's question can be determined. For example: if the user asks what the meeting summary of the project meeting is, it can be determined that the content of the project meeting summary is output.

[0097] In the present invention, vectorization is performed on the question text to obtain question vector data, that is, converting the text into a numerical vector, which is an important step in natural language processing (NLP). The purpose of doing this is to enable computers to process and analyze text data. The following are several common text vectorization methods: Bag of Words (BoW): In this model, the text is represented as a vector, in which each dimension corresponds to a word in the vocabulary. Each element in the vector represents the number of times the word appears in the text. This method ignores the word order, but is simple and easy to implement. TF-IDF (TermFrequency-Inverse Document Frequency): TF-IDF is a weighted technique that takes into account term frequency (TF) and inverse document frequency (IDF). Term frequency measures the frequency of occurrence of a word in a document, while inverse document frequency measures the prevalence of a word in all documents. Each element in the TF-IDF vector is the product of term frequency and inverse document frequency, which is used to represent the importance of the word in the document. Word Embeddings: Word Embeddings is a technique for mapping words to real number vectors that can capture the semantic and grammatical relationships between words. Common word embedding models include Word2Vec, GloVe, and FastText. Word embeddings are usually pre-trained on a larger corpus and can then be used for various NLP tasks. Sentence Embeddings: Sentence embeddings are an extension of word embeddings that map entire sentences or paragraphs to a fixed-length vector. This approach can capture more complex semantic information and takes word order into account. Common sentence embedding models include Doc2Vec, BERT, GPT, etc. Topic Models: Topic models such as LDA (Latent Dirichlet Allocation) can represent documents as a mixture of multiple topics. Each topic is represented by a probability distribution of a set of words, and the document vector represents the distribution of these topics. When performing text vectorization, which method to choose depends on the specific application scenario and task requirements. For example, for simple text classification tasks, BoW or TF-IDF may be sufficient; for tasks that require understanding the deep semantics of the text, word embedding or sentence embedding may be more appropriate.

[0098] In the present invention, the question speech input by the user is obtained, and the question text corresponding to the question speech is determined according to the preset understanding model, which is usually called automatic speech recognition (Automatic Speech Recognition, ASR). The following are the basic steps for the ASR system to convert speech into text: Voice input: The user inputs the voice through a microphone or other recording device. The system needs to ensure good recording quality and reduce background noise and other interference. Preprocessing: The voice signal first needs to be preprocessed, including noise reduction, endpoint detection (determining the beginning and end of the voice), framing (segmenting the continuous voice signal into short frames), etc. Feature extraction: Extract features from the preprocessed voice signal, and these features are usually numerical vectors that can represent the voice signal. Common features include Mel frequency cepstral coefficients (MFCC), linear predictive coding (LPC), etc. Speech recognition: Use a preset understanding model (ASR model) to identify the text corresponding to the voice feature. Traditional ASR systems usually use hidden Markov models (HMM) combined with Gaussian mixture models (GMM) or deep neural networks (DNN) for recognition. Modern ASR systems increasingly use end-to-end deep learning models, such as recurrent neural networks (RNNs), long short-term memory networks (LSTMs), convolutional neural networks (CNNs), and Transformers. Decoding: The decoder is responsible for converting the probability distribution output by the ASR model into a final text sequence. The decoding process may involve a language model (LM), which can help improve recognition accuracy, especially when dealing with ambiguity. Post-processing: The recognized text may require post-processing to correct errors, standardize text formats, etc. Post-processing may include spelling checking, grammar correction, punctuation addition, etc. Output text: Finally, the system outputs the recognized and post-processed text to the user or subsequent applications. The performance of the ASR system is affected by many factors, including speech quality, the speaker's accent, speaking speed, background noise, as well as the complexity of the ASR model and the quality of the training data.

[0099] In an embodiment of the present invention, after determining the target text data corresponding to the target conference vector data, it may also include: converting the target text data into corresponding voice data according to a preset conversion model, and outputting the voice data.

[0100] In the present invention, the determined target text data can be converted into corresponding voice data through a preset conversion model, and then the voice data is output, thereby performing voice interaction with the user.

[0101] In the present invention, the process of converting the target text data into the corresponding voice data according to the preset conversion model generally includes the following steps: Text preprocessing: First, the target text data needs to be preprocessed, including word segmentation, punctuation removal, case conversion, spelling check, etc. This step ensures that the text format is consistent and facilitates subsequent processing. Text analysis: The text analysis module is responsible for converting the preprocessed text into an intermediate representation that can be understood by the speech synthesis system, such as phoneme sequence, prosodic features (such as pitch, volume, speaking speed, etc.). For TTS systems based on deep learning, this step may involve neural network models, such as Tacotron or FastSpeech, which can generate voice features directly from text. Speech synthesis: Based on the intermediate representation obtained by text analysis, the speech synthesis module generates the corresponding speech signal. In parametric TTS, this usually involves a vocoder (such as Griffin-Lim algorithm, WaveNet, LPCNet, etc.), which converts speech features into waveforms. In an end-to-end TTS system, such as Tacotron combined with WaveNet, this step can be completed in one step to generate speech waveforms directly from text. Post-processing: The generated speech data may need to be post-processed to improve the naturalness and quality of the speech. Post-processing may include removing noise, adjusting volume, adding reverberation, etc. Output speech data: Finally, the generated speech data is output in the form of audio files. Common formats include WAV, MP3, etc. These audio files can be further used in various applications. In practical applications, the performance of the TTS system depends on multiple factors, including the accuracy of text pre-processing, the depth of text analysis, the naturalness of speech synthesis, and the sophistication of post-processing.

[0102] In the embodiment of the present invention, after obtaining the question voice input by the user, the method may further include: verifying the identity of the user according to the question voice; if the user's identity verification passes, determining the question text corresponding to the question voice according to a preset understanding model.

[0103] Specifically, when a user asks a question, that is, after obtaining the question voice input by the user, the user's identity will be verified first. Only when the user's identity verification is passed can the subsequent steps be performed to prevent information leakage caused by questions from others.

[0104] In one embodiment, verifying the identity of the user based on the question voice may include: obtaining voiceprint data of the question voice; determining whether the similarity between the voiceprint data and the preset voiceprint is greater than the preset similarity; if the similarity between the voiceprint data and the preset voiceprint is greater than the preset similarity, determining that the user's identity verification has passed.

[0105] Specifically, for example, if the preset similarity is 90%, and it is determined that the similarity between the voiceprint data of the problematic voice and the preset voiceprint is 91%, it can be determined that the user's identity verification has passed; if it is determined that the similarity between the voiceprint data of the problematic voice and the preset voiceprint is 70%, it can be determined that the user's identity verification has failed.

[0106] In the present invention, when verifying the identity of the user based on the voiceprint data of the problem voice, the user's problem voice can be preprocessed first to remove noise and other interference. The preprocessing steps may include filtering, noise reduction, framing and endpoint detection. Extract the voiceprint data from the preprocessed problem voice. The voiceprint data may include fundamental frequency, resonance peak, Mel frequency cepstral coefficient (MFCC), etc. The extraction of voiceprint data is a key step in voiceprint recognition because these features will be used to distinguish different speakers. When judging the similarity between the voiceprint data and the preset voiceprint, it can be judged according to the preset recognition model. The preset recognition model may include a Gaussian mixture model-general background model, or a deep learning model (such as a convolutional neural network).

[0107] In an embodiment of the present invention, after filling the first text data into a plurality of structured meeting record items respectively, it also includes: obtaining the user's satisfaction with the target meeting organization model; when the satisfaction is less than the preset satisfaction, optimizing the target meeting organization model.

[0108] Specifically, the user can evaluate the target meeting organization model, that is, the user can input the satisfaction level for the target meeting organization model. If the satisfaction level is greater than the preset satisfaction level, it proves that the target meeting organization model can accurately record and organize the target meeting. If the satisfaction level is less than the preset satisfaction level, the target meeting organization model can be continuously optimized.

[0109] In an embodiment of the present invention, the audio data of the target meeting is obtained; based on the audio data and the target meeting arrangement model, the target meeting template corresponding to the audio data is determined; the target meeting template includes multiple structured meeting record items; based on the audio data and the target meeting arrangement model, the first text data corresponding to the multiple structured meeting record items are respectively determined, and the first text data is respectively filled into the multiple structured meeting record items. Thus, the content of the meeting can be arranged and recorded through the target meeting arrangement model, without spending time and energy to summarize and arrange the content of the meeting.

[0110] In order to better illustrate the embodiments of the present invention, Figure 2 As shown, a flow chart of the steps of another conference recording method provided by an embodiment of the present invention is shown.

[0111] The audio data of the target conference is obtained. The target conference organization model determines the conference module corresponding to the audio data according to the audio data. The target conference organization model fills the meeting record items in the target conference template according to the audio data. The user determines whether the meeting record items in the target conference template are completely filled. If the user determines that the meeting record items in the target conference template are completely filled, a complete meeting record is obtained, and the complete meeting record is stored in the database. If the user determines that the meeting record items in the target conference template are incomplete, the target conference organization model is corrected to obtain a corrected target conference organization model, and a complete meeting record is obtained based on the corrected target conference organization model, and then the complete meeting record is stored in the database. Finally, the filled text in the target conference template is vectorized to obtain conference vector data, and the conference vector data is saved to the vector database.

[0112] like Figure 3 As shown, a flowchart of the steps of an interactive method for conference records provided by an embodiment of the present invention is shown.

[0113] Get the question voice input by the user, verify the user's identity based on the question voice, if the user's identity verification passes, determine the question text corresponding to the question voice according to the preset understanding model, vectorize the question text to obtain question vector data, match the target conference vector data corresponding to the question vector data from the vector database, determine the target text data corresponding to the target conference vector data, convert the target text data into the corresponding voice data according to the preset conversion model, and output the voice data.

[0114] In an embodiment of the present invention, the audio data of the target meeting is obtained; based on the audio data and the target meeting arrangement model, the target meeting template corresponding to the audio data is determined; the target meeting template includes multiple structured meeting record items; based on the audio data and the target meeting arrangement model, the first text data corresponding to the multiple structured meeting record items are respectively determined, and the first text data is respectively filled into the multiple structured meeting record items. Thus, the content of the meeting can be arranged and recorded through the target meeting arrangement model, without spending time and energy to summarize and arrange the content of the meeting.

[0115] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present invention are not limited by the order of the actions described, because according to the embodiments of the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.

[0116] Reference Figure 4 , shows a structural block diagram of a conference recording device provided by an embodiment of the present invention, which may specifically include the following modules:

[0117] A first acquisition module 401 is used to acquire audio data of a target conference;

[0118] A first determination module 402 is used to determine a target conference template corresponding to the audio data according to the audio data and the target conference arrangement model; the target conference template includes a plurality of structured conference record items;

[0119] The second determination module 403 is used to determine the first text data corresponding to the multiple structured meeting record items according to the audio data and the target meeting arrangement model, and fill the first text data into the multiple structured meeting record items respectively.

[0120] In the embodiment of the present invention, the target conference arrangement model is trained in the following manner:

[0121] The second acquisition module is used to acquire sample data of meeting records;

[0122] A first training module is used to train a preset conference arrangement model according to the conference record sample data to obtain a trained conference arrangement model; the trained conference arrangement model is used to output a corresponding target conference template according to the audio data;

[0123] The third acquisition module is used to acquire the conference history text sample data and the corresponding record template sample data;

[0124] The second training module is used to train the trained meeting arrangement model according to the meeting history text sample data and the corresponding record template sample data to obtain a target meeting arrangement model.

[0125] In an embodiment of the present invention, the device further includes:

[0126] A correction module, used to obtain a user's correction operation on the target conference arrangement model, and obtain a corrected target conference arrangement model;

[0127] The third determination module is used to re-determine the target conference template corresponding to the audio data and the first text data corresponding to the multiple structured meeting record items based on the audio data and the revised target conference organization model, and fill the first text data into the multiple structured meeting record items respectively.

[0128] In an embodiment of the present invention, the device further includes:

[0129] A first processing module, configured to perform vectorization processing on the first text data respectively filled in the plurality of structured meeting record items to obtain meeting vector data, and save the meeting vector data to a vector database;

[0130] The fourth acquisition module is used to acquire the question voice input by the user;

[0131] A fourth determination module, used to determine the question text corresponding to the question speech according to a preset understanding model;

[0132] A second processing module is used to perform vectorization processing on the question text to obtain question vector data;

[0133] A matching module, used for matching target conference vector data corresponding to the question vector data from the vector database;

[0134] The fifth determination module is used to determine the target text data corresponding to the target conference vector data and output the target text data.

[0135] In an embodiment of the present invention, the device further includes:

[0136] The conversion module is used to convert the target text data into corresponding voice data according to a preset conversion model, and output the voice data.

[0137] In an embodiment of the present invention, the device further includes:

[0138] A verification module, used to verify the identity of the user according to the question voice;

[0139] The sixth determination module is used to determine the question text corresponding to the question voice according to a preset understanding model if the identity verification of the user passes.

[0140] In an embodiment of the present invention, the verification module includes:

[0141] An acquisition submodule, used to acquire the voiceprint data of the problematic speech;

[0142] A judgment submodule, used to judge whether the similarity between the voiceprint data and a preset voiceprint is greater than a preset similarity;

[0143] The determination submodule is used to determine that the identity verification of the user has passed if the similarity between the voiceprint data and the preset voiceprint is greater than the preset similarity.

[0144] In an embodiment of the present invention, the device further includes:

[0145] A seventh determination module, configured to determine second text data corresponding to the audio data;

[0146] The marking module is used to mark the second text data according to the time of the target meeting, so that the user can trace the corresponding second text data according to the mark.

[0147] In an embodiment of the present invention, the device further includes:

[0148] A fifth acquisition module, used to acquire user satisfaction with the target conference arrangement model;

[0149] The optimization module is used to optimize the target meeting arrangement model when the satisfaction level is less than a preset satisfaction level.

[0150] In an embodiment of the present invention, the first acquisition module is used to acquire the audio data of the target meeting; the first determination module is used to determine the target meeting template corresponding to the audio data based on the audio data and the target meeting arrangement model; the target meeting template includes multiple structured meeting record items; the second determination module is used to determine the first text data corresponding to the multiple structured meeting record items respectively based on the audio data and the target meeting arrangement model, and fill the first text data into the multiple structured meeting record items respectively. Therefore, the content of the meeting can be sorted and recorded through the target meeting arrangement model, without spending time and energy to summarize and organize the content of the meeting.

[0151] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0152] An embodiment of the present invention further provides an electronic device, including:

[0153] It includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When the computer program is executed by the processor, the various processes of the above-mentioned meeting recording method embodiment are implemented and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0154] An embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the above-mentioned meeting recording method embodiment are implemented and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0155] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0156] It will be appreciated by those skilled in the art that the embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the embodiments of the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0157] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0158] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0159] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable terminal device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0160] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0161] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0162] The meeting recording method, the meeting recording device, the electronic device and the computer-readable storage medium provided by the present invention are introduced in detail above. The principles and implementation modes of the present invention are explained in this article by using specific examples. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation mode and the scope of application. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A conference recording method, characterized in that: The method comprises: Get the audio data of the target conference; Determining a target conference template corresponding to the audio data according to the audio data and the target conference arrangement model; the target conference template includes a plurality of structured conference record items; According to the audio data and the target conference organizing model, first text data corresponding to the multiple structured conference record items are respectively determined, and the first text data are respectively filled into the multiple structured conference record items.

2. The conference recording method according to claim 1, characterized in that: The target conference arrangement model is trained in the following way: Get sample data of meeting minutes; The preset conference arrangement model is trained according to the conference record sample data to obtain a trained conference arrangement model; the trained conference arrangement model is used to output a corresponding target conference template according to the audio data; Obtaining conference history text sample data and corresponding record template sample data; The trained meeting arrangement model is trained according to the meeting history text sample data and the corresponding record template sample data to obtain a target meeting arrangement model.

3. The conference recording method according to claim 1, characterized in that: After the first text data is filled into the plurality of structured meeting record items respectively, the method further includes: Acquire a user's modification operation on the target conference arrangement model to obtain a modified target conference arrangement model; Based on the audio data and the revised target conference organizing model, the target conference template corresponding to the audio data and the first text data corresponding to the multiple structured meeting record items are determined again, and the first text data are filled into the multiple structured meeting record items respectively.

4. The conference recording method according to claim 1, characterized in that: After the first text data is filled into the plurality of structured meeting record items respectively, the method further includes: Performing vectorization processing on the first text data respectively filled in the plurality of structured meeting record items to obtain meeting vector data, and saving the meeting vector data to a vector database; Get the question voice input by the user; Determining a question text corresponding to the question speech according to a preset understanding model; Performing vectorization processing on the question text to obtain question vector data; matching target meeting vector data corresponding to the question vector data from the vector database; Determine the target text data corresponding to the target conference vector data, and output the target text data.

5. The conference recording method according to claim 4, characterized in that: After determining the target text data corresponding to the target conference vector data, the method further includes: According to a preset conversion model, the target text data is converted into corresponding voice data, and the voice data is output.

6. The conference recording method according to claim 4, characterized in that: After obtaining the question voice input by the user, the method further includes: Verifying the identity of the user according to the question voice; If the user's identity verification is passed, the question text corresponding to the question speech is determined according to a preset understanding model.

7. The conference recording method according to claim 6, characterized in that: The verifying the identity of the user according to the question voice includes: Acquire voiceprint data of the problematic speech; Determining whether the similarity between the voiceprint data and a preset voiceprint is greater than a preset similarity; If the similarity between the voiceprint data and the preset voiceprint is greater than the preset similarity, it is determined that the identity verification of the user has passed.

8. The conference recording method according to claim 1, characterized in that: After obtaining the audio data of the target conference, the method further includes: Determining second text data corresponding to the audio data; The second text data is marked according to the time of the target meeting, so that the user can trace back the corresponding second text data according to the mark.

9. The method according to claim 1, characterized in that: After the first text data is filled into the plurality of structured meeting record items respectively, the method further includes: Obtaining user satisfaction with the target conference organization model; When the satisfaction level is less than a preset satisfaction level, the target meeting organization model is optimized.

10. A conference recording device, characterized in that: The device comprises: A first acquisition module, used to acquire audio data of a target conference; A first determination module is used to determine a target conference template corresponding to the audio data according to the audio data and the target conference arrangement model; the target conference template includes a plurality of structured conference record items; The second determination module is used to determine the first text data corresponding to the multiple structured meeting record items according to the audio data and the target meeting arrangement model, and fill the first text data into the multiple structured meeting record items respectively.

11. An electronic device, characterized in that: include: A processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein when the computer program is executed by the processor, the steps of the conference recording method according to any one of claims 1 to 9 are implemented.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the meeting recording method according to any one of claims 1 to 9 are implemented.

Images