Intelligent audio processing method, system and device

By automatically adjusting the sound field parameters of the conference room through voiceprint feature recognition and adjustment model, the problems of poor sound quality and complicated operation of traditional conference room sound reinforcement systems are solved, realizing personalized audio processing and convenient operation.

CN119031304APending Publication Date: 2024-11-26BEIJING DESHENG INTELLIGENT ELECTRONIC TECHNOLOGY CO LTD
View PDF 9 Cites 0 Cited by

Patent Information

Application Number
CN202411278072.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-12
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

Traditional conference room sound reinforcement systems are easily affected by environmental noise, room acoustics, and user operation during audio signal processing, resulting in poor sound quality, feedback howling, and uneven sound. They also require manual adjustment and are complicated to operate, making them unable to adapt to changes in the speaker.

Method used

By identifying speakers through voiceprint features, the system automatically adjusts the sound field parameters of the conference room, including transmission gain, sound field non-uniformity, and voice transmission index. The adjustment model is used to call up these parameters, enabling personalized audio system settings.

Benefits of technology

It improves the sound amplification effect in the conference room, meets the requirements of speakers, and enables convenient personalized use and operation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119031304A_ABST
    Figure CN119031304A_ABST
Patent Text Reader

Abstract

The invention provides an intelligent audio processing method, system and device, and belongs to the technical field of conference room sound amplification audio processing. Sound field parameters, including sound transmission gain, sound field unevenness and voice transmission index, of a conference room are automatically adjusted through an adjusting model, the requirements of the conference room for the sound transmission gain, sound field uniformity and sound quality are met, and the sound quality of the conference room is improved. The sound amplification effect of the conference room is improved, the use requirement of the speaker is met, meanwhile, the speaker is recognized through the voiceprint features, parameter calling of the matched audio system is carried out, and personalized use and operation convenience of the conference room are achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of conference room sound reinforcement audio processing, and in particular to an intelligent audio processing method, system and device. Background Art

[0002] In modern conference rooms, the audio quality of the sound reinforcement system has a direct impact on the efficiency of the meeting, as well as the communication effect and experience between participants. High-quality audio can ensure that participants can hear the sound clearly and promote the effective transmission of information. However, the traditional conference room sound reinforcement system is easily affected by many factors during the audio signal processing process. These factors include environmental noise, room acoustic characteristics, and user operations. These problems often lead to poor sound quality, feedback howling, and uneven sound, which significantly reduce the effectiveness of the meeting.

[0003] Existing technologies often require manual adjustment of system parameters when dealing with these audio issues. This process is not only time-consuming and labor-intensive, but also requires professional knowledge and experience, and non-professional users may not be able to achieve the desired effect when operating. In addition, existing automatic adjustment technologies mainly focus on suppressing feedback howling, but only dealing with this aspect is not enough to meet the comprehensive needs of conference room audio. This technology usually fails to fully consider whether the adjusted sound reinforcement system can meet the requirements of use. Finally, the static design of traditional systems cannot adapt to changes in speakers, which means that speakers still need to manually readjust in different situations, increasing the complexity and inconvenience of operation. Summary of the invention

[0004] In order to solve the above problems in the prior art, the present invention provides an intelligent audio processing method, system and device. The invention automatically adjusts the sound field parameters of the conference room, including the sound transmission gain, the sound field unevenness and the voice transmission index, by adjusting the model, so as to meet the requirements of the conference room for the sound transmission gain, the sound field uniformity and the sound quality, improve the sound reinforcement effect of the conference room, and meet the use requirements of the speaker. At the same time, the speaker is identified by the voiceprint feature, and the parameters of the matching audio system are called, so as to realize the personalized use of the conference room and the convenience of operation. To achieve the above purpose, the technical scheme is as follows:

[0005] In a first aspect, the present invention provides an intelligent audio processing method, the method comprising:

[0006] S1. Obtain the initial parameters of the audio system according to the spatial layout of the conference room and the audio system solution;

[0007] S2. At the beginning of the meeting, the speaker's voiceprint characteristics are obtained based on the speaker's audio data;

[0008] S3. Compare the voiceprint features of the speaker with the voiceprint features stored in the audio system, and extract the parameters of the audio system according to the comparison rules. The comparison rules are as follows:

[0009] Rule 1: If the voiceprint feature of the speaker matches the voiceprint feature stored in the audio system, then the system parameters of the speaker matching the voiceprint feature stored in the audio system are extracted as the parameters of the audio system;

[0010] Rule 2: If the voiceprint feature of the speaker does not match the voiceprint feature saved in the audio system, the initial parameters of the audio system are extracted as the parameters of the audio system and the voiceprint feature of the speaker is saved;

[0011] S4, inputting the parameters of the audio system into the audio system, and obtaining the sound field parameters of the conference room through the sound collection system of the conference room;

[0012] S5. If the sound field parameters meet the requirements, the process ends; if the sound field parameters do not meet the requirements, proceed to the next step;

[0013] S6, inputting the sound field parameters into the adjustment model to obtain adjusted parameters;

[0014] S7, replacing the parameters of the audio system with the adjusted parameters, repeating S4 to S7, and obtaining the final system parameters of the speaker;

[0015] S8. Match the voiceprint features of the speaker with the final system parameters of the speaker one by one and save them.

[0016] Optionally, in S2, the voiceprint features of the speaker are obtained according to the audio data of the speaker, including:

[0017] S21, preprocessing the audio data of the speaker according to the audio data of the speaker to obtain preprocessed audio data;

[0018] S22, performing frame processing on the preprocessed audio data to obtain N preprocessed audio data samples;

[0019] S23, according to the N preprocessed audio data samples, processing through a Mel filter to obtain Mel frequency cepstral coefficient features of the preprocessed audio data samples;

[0020] S24. According to the Mel-frequency cepstral coefficient features of the preprocessed audio data samples, representative features are selected to form a feature collection to obtain the voiceprint features of the speaker.

[0021] Optionally, the sound collection system of the conference room in S4 includes a plurality of standard omnidirectional microphones, which are distributed in the conference room.

[0022] Optionally, the standard omnidirectional microphone is distributed on four walls of the conference room, front, back, left, and right, and the height is 1m~1.5m.

[0023] Optionally, the sound field parameters of the conference room in S4 include sound transmission gain, sound field unevenness and speech transmission index.

[0024] Optionally, in S6, the sound field parameters are input into the adjustment model to obtain adjusted parameters, including:

[0025] S61, performing standardization processing according to the sound field parameters to obtain a standardized sound transmission gain, a standardized sound field unevenness, and a standardized speech transmission index;

[0026] S62, inputting the standardized sound transmission gain into the adjustment model to obtain the overall amplification factor of the adjusted audio system;

[0027] S63, inputting the standardized sound field unevenness into the adjustment model to obtain the directivity angle and speaker gain of the adjustable directional speaker in the audio system after adjustment;

[0028] S64, inputting the standardized voice transmission index into the adjustment model to obtain the delay parameter of the adjusted audio system;

[0029] S65, obtaining adjusted parameters after comprehensive processing according to the adjusted overall amplification factor of the audio system, the directivity angle and speaker gain of the adjustable directional speaker in the audio system, and the adjusted delay parameters of the audio system.

[0030] Optionally, the training process of the adjustment model includes:

[0031] S71, collecting parameters of the audio system and sound field parameters of the conference room to obtain a training data set;

[0032] S72, selecting an objective function according to the training data set to obtain a model prediction output;

[0033] S73, obtaining a loss function value of the model according to the model prediction output and the training data set;

[0034] S74, according to the loss function value of the model, using the gradient descent method to obtain the parameter value of the objective function;

[0035] S75, inputting the parameter value of the objective function into the objective function, and verifying the accuracy of the model based on the training data set;

[0036] S76. Determine the parameter value of the final objective function according to the accuracy of the model to obtain the adjustment model.

[0037] Optionally, it is characterized in that S4 to S7 are repeated in S7 to obtain the final system parameters of the speaker, including:

[0038] If the number of repetitions of S4 to S7 does not exceed 10, the parameters adjusted in the last repetition are selected as the final system parameters of the speaker;

[0039] If S4 to S7 are repeated more than 10 times, the process ends and the parameters closest to the requirements among the 10 times are selected as the final system parameters of the speaker.

[0040] In a second aspect, the present invention provides an intelligent audio processing system, which is applied to an intelligent audio processing method, and the system includes:

[0041] The initial module is used to obtain the initial parameters of the audio system according to the spatial layout of the conference room and the audio system solution;

[0042] A feature extraction module is used to obtain the voiceprint features of the speaker based on the audio data of the speaker;

[0043] The comparison module is used to compare the voiceprint features of the speaker with the voiceprint features stored in the audio system and extract the parameters of the audio system according to the comparison rules. The comparison rules are as follows:

[0044] Rule 1: If the voiceprint feature of the speaker matches the voiceprint feature stored in the audio system, then the system parameters of the speaker matching the voiceprint feature stored in the audio system are extracted as the parameters of the audio system;

[0045] Rule 2: If the voiceprint feature of the speaker does not match the voiceprint feature saved in the audio system, the initial parameters of the audio system are extracted as the parameters of the audio system and the voiceprint feature of the speaker is saved;

[0046] A first acquisition module is used to input the parameters of the audio system into the audio system, and obtain the sound field parameters of the conference room through the sound collection system of the conference room;

[0047] A judgment module, for terminating the process if the sound field parameters meet the requirements; and proceeding to the next step if the sound field parameters do not meet the requirements;

[0048] A second acquisition module is used to input the sound field parameters into the adjustment model to obtain adjusted parameters;

[0049] An adjustment module, used to replace the parameters of the audio system with the adjusted parameters to obtain the final system parameters of the speaker;

[0050] The saving module is used to make a one-to-one correspondence between the voiceprint features of the speaker and the final system parameters of the speaker and save them.

[0051] In a third aspect, the present invention provides an intelligent audio processing device, which is applied to an intelligent audio processing method and an intelligent audio processing system. The device includes: an audio processor, an adjustable directional speaker, a microphone, a sound collection system, a memory, and an intelligent audio processing adapter program stored on the audio processor and executable on the audio processor. When the intelligent audio processing adapter program is executed by the audio processor, the steps in the intelligent audio processing method are implemented.

[0052] Compared with the prior art, the technical solution of the present invention has at least the following beneficial effects:

[0053] On the one hand, the above scheme automatically adjusts the sound field parameters of the conference room, including transmission gain, sound field unevenness and voice transmission index, by adjusting the model, so as to meet the requirements of the conference room for transmission gain, sound field uniformity and sound quality, improve the sound reinforcement effect of the conference room, and meet the use requirements of the speakers. On the other hand, the speaker is identified through voiceprint features, and the parameters of the matching audio system are called, so as to realize the personalized use and convenient operation of the conference room. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0055] Figure 1 is a flow chart provided by an embodiment of the present invention;

[0056] Figure 2 is a flow chart of voiceprint feature extraction provided by an embodiment of the present invention;

[0057] Figure 3 is a flow chart of obtaining adjusted parameters provided by an embodiment of the present invention;

[0058] Figure 4 is a flow chart of the adjustment model training provided by an embodiment of the present invention;

[0059] Figure 5 is a schematic diagram of a conference room layout provided by an embodiment of the present invention;

[0060] Figure 6 is a system block diagram provided by an embodiment of the present invention;

[0061] Figure 7 It is a block diagram of a device provided by an embodiment of the present invention.

[0062] Explanation of the numbers in the figure: initial module 601, feature extraction module 602, comparison module 603, first acquisition module 604, judgment module 605, second acquisition module 606, adjustment module 607, storage module 608, microphone 701, sound collection system 702, audio processor 703, adjustable directional speaker 704 and memory 705. DETAILED DESCRIPTION

[0063] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0064] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0065] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0066] like Figure 1 The flowchart of the embodiment of the present invention shown in the figure provides an intelligent audio processing method, which is implemented by an intelligent audio processing system and an intelligent audio processing device. The method includes:

[0067] S1. Obtain the initial parameters of the audio system according to the spatial layout of the conference room and the audio system solution;

[0068] Specifically, Figure 5 The schematic diagram of the conference room layout provided by the embodiment of the present invention is shown, where four adjustable directional speakers 704 are placed at the four corners of the conference room, and 24 microphones 701 are located in the middle of the conference room. According to the conference room layout and the audio system solution, the initial parameters of the audio system are obtained through manual debugging, and the initial parameters include: the directivity angle and gain of each adjustable directional speaker 704, the gain of the 24 microphones 701, the gain and balance of the audio system, and the delay of the audio system.

[0069] S2. At the beginning of the meeting, the speaker's voiceprint characteristics are obtained based on the speaker's audio data;

[0070] Specifically, Figure 2 The flowchart of voiceprint feature extraction provided by the embodiment of the present invention shown in the figure first collects the audio data of the speaker, pre-processes the audio data of the speaker, removes noise using the spectral subtraction method of formula (1), and obtains the pre-processed audio data.

[0071] (1)

[0072] In the formula, is the spectrum of the speaker’s audio data after denoising, is the spectrum of the original audio data of the speaker, is the estimated noise spectrum.

[0073] Then, the preprocessed audio data is framed and N preprocessed audio data samples are obtained according to formula (2).

[0074] (2)

[0075] In the formula, is the preprocessed audio data sample of the f-th frame, is the sample of the original audio signal, is the index of the current frame, is the sample interval between each frame.

[0076] Then, the preprocessed audio data samples are subjected to short-time Fourier transform.

[0077] (3)

[0078] In the formula, is the short-time Fourier transform result at time t and frequency f, is the window function;

[0079] By performing Mel filter processing on the preprocessed audio data samples after short-time Fourier transform, as shown in formula (4) and formula (5), the Mel frequency cepstral coefficient characteristics of the preprocessed audio data samples are obtained.

[0080] (4)

[0081] (5)

[0082] In the formula, is the d-th Mel frequency feature, is the response of the Mel filter, is the kth Mel-frequency cepstral coefficient feature, and D is the number of Mel-frequency features.

[0083] Finally, based on the Mel-frequency cepstral coefficient features of the preprocessed audio data samples, representative features are selected to form a feature set, as shown in formula (6), to obtain the voiceprint features of the speaker.

[0084] (6)

[0085] In the formula, is the selected feature vector collection, q is the number of selected Mel frequency cepstral coefficient features, is the first selected Mel frequency cepstral coefficient feature, is the selected second Mel frequency cepstral coefficient feature, is the selected qth Mel-frequency cepstral coefficient feature.

[0086] S3. Compare the voiceprint features of the speaker with the voiceprint features stored in the audio system. If the voiceprint features of the speaker do not match the voiceprint features stored in the audio system, extract the initial parameters of the audio system as the parameters of the audio system and store the voiceprint features of the speaker.

[0087] S4, inputting the initial parameters of the frequency system into the audio system, and obtaining the sound field parameters of the conference room through the sound collection system 702 of the conference room;

[0088] Specifically, Figure 5 The schematic diagram of the layout of the conference room provided by the embodiment of the present invention is shown, and the four standard omnidirectional microphones of the sound collection system 702 are respectively arranged on the four walls of the conference room at a height of 1.5m, and respectively collect the sound field parameters of the conference room at four positions of the front, back, left and right of the conference room;

[0089] According to the sound field parameters of the conference room at the four locations, the sound transmission gain in the sound field parameters of the conference room at the four locations is averaged, and the sound transmission gain of the conference room is -8dB.

[0090] Compare the sound pressure levels in the sound field parameters of the conference room at the four locations, subtract the maximum sound pressure level from the minimum sound pressure level, and get the sound field unevenness of the conference room to be 10dB.

[0091] The speech transmission indexes in the conference room sound field parameters at the four locations are compared, and the minimum value is selected, and the speech transmission index of the conference room is obtained to be 0.6.

[0092] S5. Compare the sound field parameters with the requirements of the national standard GB 50371-2006.

[0093] Specifically, the requirements of national standard GB 50371-2006 are as follows:

[0094] Transmission gain ≥-10dB, sound field unevenness ≤8dB, speech transmission index ≥0.55;

[0095] By comparing the sound transmission gain of the conference room, the sound field unevenness of the conference room and the voice transmission index of the conference room, it is found that the sound transmission gain of the conference room and the voice transmission index of the conference room meet the requirements, while the sound field unevenness of the conference room does not meet the requirements.

[0096] S6, inputting the sound field parameters into the adjustment model to obtain adjusted parameters;

[0097] Specifically, Figure 4 The flowchart of the training of the adjustment model provided by the embodiment of the present invention shown in the figure firstly trains the adjustment model, sets the parameters of the audio system multiple times and obtains the sound field parameters of the conference room to form a training data set;

[0098] According to the training data set, the linear regression function is selected as the objective function, as shown in formula (7), and the model prediction output is obtained.

[0099] (7)

[0100] In the formula, is the model prediction output, is the intercept of the predicted output, is the weight of the sth feature, is the sth input feature.

[0101] Then, through the loss function, as shown in formula (8), the loss function value of the model is obtained:

[0102] (8)

[0103] In the formula, is the loss value, p is the number of samples, is the real output.

[0104] Using the gradient descent method, as shown in formula (9), the parameter value of the objective function is obtained

[0105] (9)

[0106] In the formula, is the learning rate.

[0107] Will Input into the linear regression function and verify the accuracy of the model based on the training data set;

[0108] The model is verified to be accurate and the final , and obtain the adjustment model.

[0109] Specifically, Figure 3The flowchart of obtaining the adjusted parameters provided by the embodiment of the present invention shown in the figure is to standardize the sound field unevenness of the conference room to obtain the standardized sound field unevenness, and input the standardized sound field unevenness into the above-mentioned trained adjustment model to obtain the directivity angle and speaker gain of the adjustable directional speaker 704 in the audio system after adjustment;

[0110] S7. The directivity angle and speaker gain of the adjustable directional speaker 704 in the audio system are adjusted to replace the initial directivity angle and initial speaker gain of the adjustable directional speaker 704 in the audio system. Other parameters remain unchanged. S4 to S7 are repeated. After the fifth cycle, the sound field unevenness of the conference room is 7.8 dB, which meets the requirements, and the final system parameters of the speaker are obtained.

[0111] S8. Match the voiceprint features of the speaker with the final system parameters of the speaker one by one and save them.

[0112] like Figure 6 The system block diagram of the embodiment of the present invention shown in the figure, the present invention provides an intelligent audio processing system, the system is applied to an intelligent audio processing method, the system includes an initial module 601, a feature extraction module 602, a comparison module 603, a first acquisition module 604, a judgment module 605, a second acquisition module 606, an adjustment module 607, and a storage module 608. Specifically,

[0113] Initial module 601, used to obtain initial parameters of the audio system according to the spatial layout of the conference room and the audio system solution;

[0114] A feature extraction module 602 is used to obtain the voiceprint features of the speaker based on the audio data of the speaker;

[0115] The comparison module 603 is used to compare the voiceprint features of the speaker with the voiceprint features stored in the audio system, and extract the parameters of the audio system according to the comparison rules. The comparison rules are as follows:

[0116] Rule 1: If the voiceprint feature of the speaker matches the voiceprint feature stored in the audio system, then the system parameters of the speaker matching the voiceprint feature stored in the audio system are extracted as the parameters of the audio system;

[0117] Rule 2: If the voiceprint feature of the speaker does not match the voiceprint feature saved in the audio system, the initial parameters of the audio system are extracted as the parameters of the audio system and the voiceprint feature of the speaker is saved;

[0118] A first acquisition module 604 is used to input the parameters of the audio system into the audio system, and obtain the sound field parameters of the conference room through the sound collection system 702 of the conference room;

[0119] The judging module 605 is used to terminate the process if the sound field parameters meet the requirements; if the sound field parameters do not meet the requirements, proceed to the next step;

[0120] A second acquisition module 606, used for inputting the sound field parameters into the adjustment model to obtain adjusted parameters;

[0121] An adjustment module 607, used to replace the parameters of the audio system with the adjusted parameters to obtain the final system parameters of the speaker;

[0122] The saving module 608 is used to make a one-to-one correspondence between the voiceprint feature of the speaker and the final system parameters of the speaker and save them.

[0123] like Figure 7 The device block diagram of the embodiment of the present invention shown in the figure, the present invention provides an intelligent audio processing device, the device is applied to an intelligent audio processing method and an intelligent audio processing system, the device includes: an audio processor 703, an adjustable directional speaker 704, a microphone 701, a sound collection system 702, a memory 705 and an intelligent audio processing adapter stored on the audio processor 703 and executable on the audio processor 703, when the intelligent audio processing adapter is executed by the audio processor 703, the steps in the intelligent audio processing method are implemented.

[0124] The present invention provides an intelligent audio processing method, system and device. The invention automatically adjusts the sound field parameters of the conference room, including transmission gain, sound field unevenness and voice transmission index, by adjusting the model, so as to meet the requirements of the conference room for transmission gain, sound field uniformity and sound quality, improve the sound reinforcement effect of the conference room, and meet the use requirements of the speakers. At the same time, the speaker is identified through voiceprint features, and the parameters of the matching audio system are called, so as to realize the personalized use of the conference room and the convenience of operation.

[0125] It is to be understood that the present invention is described by the above embodiments and should not be construed as limiting the embodiments of the present invention and the scope of the present invention. It is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of the present invention.

Claims

1. An intelligent audio processing method, characterized in that: The method comprises: S1. Obtain the initial parameters of the audio system according to the spatial layout of the conference room and the audio system solution; S2. At the beginning of the meeting, the speaker's voiceprint characteristics are obtained based on the speaker's audio data; S3. Compare the voiceprint features of the speaker with the voiceprint features stored in the audio system, and extract the parameters of the audio system according to the comparison rules. The comparison rules are as follows: Rule 1: If the voiceprint feature of the speaker matches the voiceprint feature stored in the audio system, then extract the system parameters of the speaker that matches the voiceprint feature stored in the audio system as the parameters of the audio system; Rule 2: If the voiceprint feature of the speaker does not match the voiceprint feature saved in the audio system, extract the initial parameters of the audio system as the parameters of the audio system and save the voiceprint feature of the speaker; S4, inputting the parameters of the audio system into the audio system, and obtaining the sound field parameters of the conference room through the sound collection system of the conference room; S5. If the sound field parameters meet the requirements, the process ends; if the sound field parameters do not meet the requirements, proceed to the next step; S6, inputting the sound field parameters into the adjustment model to obtain adjusted parameters; S7, replacing the parameters of the audio system with the adjusted parameters, repeating S4 to S7, to obtain the final system parameters of the speaker; S8. Match the voiceprint features of the speaker with the final system parameters of the speaker one by one and save them.

2. The intelligent audio processing method according to claim 1, characterized in that: In S2, the voiceprint features of the speaker are obtained according to the audio data of the speaker, including: S21, preprocessing the audio data of the speaker according to the audio data of the speaker to obtain preprocessed audio data; S22, performing frame processing on the preprocessed audio data to obtain N preprocessed audio data samples; S23, processing the N preprocessed audio data samples through a Mel filter to obtain Mel-frequency cepstral coefficient features of the preprocessed audio data samples; S24. According to the Mel-frequency cepstral coefficient features of the preprocessed audio data samples, representative features are selected to form a feature collection to obtain the voiceprint features of the speaker.

3. The intelligent audio processing method according to claim 1, characterized in that: The sound collection system of the conference room in S4 includes a plurality of standard omnidirectional microphones, and the standard omnidirectional microphones are distributed in the conference room.

4. The intelligent audio processing method according to claim 3, characterized in that: The standard omnidirectional microphones are distributed on the four walls of the conference room, front, back, left, and right, and the height is 1m~1.5m.

5. The intelligent audio processing method according to claim 1, characterized in that: The sound field parameters of the conference room in S4 include sound transmission gain, sound field unevenness and speech transmission index.

6. The intelligent audio processing method according to claim 5, characterized in that: In S6, the sound field parameters are input into the adjustment model to obtain adjusted parameters, including: S61, performing standardization processing according to the sound field parameters to obtain a standardized sound transmission gain, a standardized sound field unevenness, and a standardized speech transmission index; S62, inputting the standardized sound transmission gain into an adjustment model to obtain an adjusted overall amplification factor of the audio system; S63, inputting the standardized sound field unevenness into an adjustment model to obtain a directivity angle and a speaker gain of an adjustable directional speaker in an audio system after adjustment; S64, inputting the standardized voice transmission index into the adjustment model to obtain the delay parameter of the adjusted audio system; S65, obtaining adjusted parameters after comprehensive processing based on the adjusted overall amplification factor of the audio system, the directivity angle and speaker gain of the adjustable directional speaker in the audio system, and the adjusted delay parameters of the audio system.

7. The intelligent audio processing method according to claim 6, characterized in that: The training process of the regulation model includes: S71, collecting parameters of the audio system and sound field parameters of the conference room to obtain a training data set; S72, selecting an objective function according to the training data set to obtain a model prediction output; S73, obtaining a loss function value of a model according to the model prediction output and the training data set; S74, according to the loss function value of the model, using the gradient descent method to obtain the parameter value of the objective function; S75, inputting the parameter value of the objective function into the objective function, and verifying the accuracy of the model according to the training data set; S76. Determine the parameter value of the final objective function according to the accuracy of the model to obtain the adjustment model.

8. The intelligent audio processing method according to claim 1, characterized in that: In S7, S4 to S7 are repeated to obtain the final system parameters of the speaker, including: If the number of repetitions of S4 to S7 does not exceed 10, the last adjusted parameter is selected as the final system parameter of the speaker; If the number of repetitions of S4 to S7 exceeds 10, the process ends, and the parameters closest to the requirements among the 10 times are selected as the final system parameters of the speaker.

9. An intelligent audio processing system, characterized in that: The system comprises: The initial module is used to obtain the initial parameters of the audio system according to the spatial layout of the conference room and the audio system solution; A feature extraction module is used to obtain the voiceprint features of the speaker based on the audio data of the speaker; The comparison module is used to compare the voiceprint features of the speaker with the voiceprint features stored in the audio system, and extract the parameters of the audio system according to the comparison rules. The comparison rules are as follows: Rule 1: If the voiceprint feature of the speaker matches the voiceprint feature stored in the audio system, then extract the system parameters of the speaker that matches the voiceprint feature stored in the audio system as the parameters of the audio system; Rule 2: If the voiceprint feature of the speaker does not match the voiceprint feature saved in the audio system, extract the initial parameters of the audio system as the parameters of the audio system and save the voiceprint feature of the speaker; A first acquisition module, used for inputting the parameters of the audio system into the audio system, and obtaining the sound field parameters of the conference room through the sound collection system of the conference room; A judgment module, for terminating the process if the sound field parameters meet the requirements; and proceeding to the next step if the sound field parameters do not meet the requirements; A second acquisition module is used to input the sound field parameters into the adjustment model to obtain adjusted parameters; An adjustment module, used to replace the parameters of the audio system with the adjusted parameters to obtain the final system parameters of the speaker; The saving module is used to make a one-to-one correspondence between the voiceprint features of the speaker and the final system parameters of the speaker and save them.

10. An intelligent audio processing device, characterized in that: The device includes: an audio processor, an adjustable directional speaker, a microphone, a sound collection system, a memory, and an intelligent audio processing adapter program stored on the audio processor and executable on the audio processor. When the intelligent audio processing adapter program is executed by the audio processor, the steps in the intelligent audio processing method are implemented.

Citation Information

Patent Citations

  • Self-adaptive adjustment method for indoor sound reinforcement system

    CN103414991A

  • Pickup amplification method and system based on human body recognition and voiceprint matching

    CN115866499A

  • Remote conference communication intelligent noise reduction method and system

    CN117746880A

  • Audio equipment setting method and device

    CN117857972A

  • Automatic adjusting system for teaching sound field environment

    CN118042354A