A low-delay audio signal processing method for vehicle-mounted audio DSP

Abstract

The present invention discloses a low-delay processing method for audio signals of an in-vehicle audio DSP, comprising: the in-vehicle audio DSP receives an audio signal, decomposes and processes the audio signal to obtain a first separated signal and a second separated signal; performs noise reduction processing on the first separated signal to obtain a first noise reduction signal; performs noise reduction processing on the second separated signal to obtain a second noise reduction signal; performs signal fusion processing on the first noise reduction signal and the second noise reduction signal to obtain a target signal; performs signal classification on the target signal to obtain a plurality of classification sets; generates a transmission task for each classification set, and performs a plurality of transmission tasks based on a preset low-delay channel to perform signal transmission. The signal processing delay and transmission delay are reduced, and the rapidity and accuracy of signal processing are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of audio processing, and in particular to a low-delay processing method for audio signals of a vehicle-mounted audio DSP. Background Art

[0002] In-car audio DSP refers to an in-car audio digital signal processor, which is a digital audio signal processor used in in-car audio systems. In the prior art, in-car audio DSP processes audio signals with high processing delay and poor processing effect, resulting in poor output audio effect. At the same time, when transmitting the output audio signal, it is transmitted based on a conventional transmission channel, resulting in high transmission delay, which is not conducive to signal output. Summary of the invention

[0003] The present invention aims to solve at least one of the technical problems in the above-mentioned technology to a certain extent. To this end, the purpose of the present invention is to provide a low-delay audio signal processing method of a vehicle-mounted audio DSP, reduce signal processing delay and transmission delay, and improve the speed and accuracy of signal processing.

[0004] To achieve the above object, an embodiment of the present invention proposes a low-delay audio signal processing method of a vehicle-mounted audio DSP, comprising:

[0005] The vehicle audio DSP receives the audio signal, decomposes and processes the audio signal, and obtains a first separation signal and a second separation signal;

[0006] Performing noise reduction processing on the first separated signal to obtain a first noise reduction signal;

[0007] Performing noise reduction processing on the second separated signal to obtain a second noise reduction signal;

[0008] Performing signal fusion processing on the first noise reduction signal and the second noise reduction signal to obtain a target signal;

[0009] The target signal is classified to obtain several classification sets; each classification set generates a transmission task, and several transmission tasks are executed based on a preset low-latency channel to transmit the signal.

[0010] According to some embodiments of the present invention, decomposing an audio signal to obtain a first separated signal and a second separated signal includes:

[0011] Performing audio encoding on the audio signal, and determining a time domain feature of the audio signal based on the encoding result;

[0012] Performing multi-scale mapping processing on the time domain features of the audio signal to determine the multi-scale features of the audio signal;

[0013] Based on the multi-scale features, the audio signal is decomposed and processed by a binary tree decomposition method composed of a high-pass filter and a low-pass filter to obtain a plurality of sub-audio signals;

[0014] The ratio of the average energy of the frequency line of each sub-audio signal to the average energy of the amplitude spectrum is calculated and compared with the preset separation threshold, and the sub-audio signal whose ratio is greater than the preset separation threshold is used as the first separation signal; the sub-audio signal whose ratio is less than or equal to the preset separation threshold is used as the second separation signal.

[0015] According to some embodiments of the present invention, performing noise reduction processing on the first separated signal to obtain a first noise-reduced signal includes:

[0016] Extracting signal energy distribution information of frequency points in each audio signal frame in the first separation signal;

[0017] Performing noise estimation on each audio signal frame according to signal energy distribution information to obtain estimated noise of each audio signal;

[0018] Inputting the estimated noise into a pre-trained noise recognition model, and outputting the noise characteristics of each audio signal and the corresponding noise reduction gain;

[0019] The corresponding audio signal is subjected to noise reduction processing according to the noise reduction gain to obtain a first noise reduction signal.

[0020] According to some embodiments of the present invention, performing noise reduction processing on the corresponding audio signal according to the noise reduction gain to obtain a first noise reduction signal includes:

[0021] The audio signal is segmented to obtain P sub-signals, and each sub-signal is extracted to determine the maximum audio value and the minimum audio value in each sub-signal to obtain a feature matrix of the audio signal;

[0022] Performing denoising processing on the feature matrix according to the denoising gain to obtain a first denoised signal;

[0023]

[0024] Among them, Bi , t is the value of the ith row and t column of the feature matrix corresponding to the first denoised signal; Ai , t is the value of the i-th row and t-th column of the characteristic matrix of the audio signal; e is a natural constant, i = 1, 2, 3 ... P, t = 1, 2; A s,t is the value of the sth row and tth column of the characteristic matrix of the audio signal; A j,t is the value of the j-th row and t-th column of the feature matrix of the audio signal; H is the noise reduction gain corresponding to the audio signal.

[0025] According to some embodiments of the present invention, performing noise reduction processing on the second separated signal to obtain a second noise-reduced signal includes:

[0026] Analyze the second separated signal to determine frequency distribution information and amplitude change information of the second separated signal;

[0027] determining a target filter based on the frequency distribution information;

[0028] determining a suppression parameter of a noise suppressor according to the amplitude variation information;

[0029] The second separated signal is subjected to noise reduction processing according to the target filter and the noise suppressor to obtain a second noise reduction signal.

[0030] According to some embodiments of the present invention, before classifying the target signal, the method further includes: evaluating the sound quality of the target signal, determining an evaluation value, and comparing it with a preset evaluation value, and judging whether the target signal is qualified based on the comparison result.

[0031] According to some embodiments of the present invention, performing a plurality of transmission tasks based on a preset low-latency channel to perform signal transmission includes:

[0032] A matching relationship between the transmission task corresponding to each classification set and the corresponding preset low-latency channel is established, and signal transmission is performed based on the matching relationship as a matching result.

[0033] According to some embodiments of the present invention, the further comprising:

[0034] Get the first time information of the car audio DSP receiving the audio signal;

[0035] Acquire second time information when a plurality of transmission tasks are executed based on a preset low-latency channel to transmit a signal;

[0036] Determine a processing delay of the vehicle audio DSP according to the first time information and the second time information, and compare it with a preset processing delay threshold;

[0037] When it is determined that the processing delay is greater than a preset processing delay threshold, an alarm prompt is issued.

[0038] According to some embodiments of the present invention, performing a plurality of transmission tasks based on a preset low-latency channel to perform signal transmission includes:

[0039] Determine the control relationship and transmission cost between the preset low-latency channels according to the preset low-latency channel control relationship diagram;

[0040] The transfer task G i Assigned to the preset low latency channel P j Perform the calculation, that is, assign the matrix elements x i，j=1, then preset low latency channel P j Complete the transmission task G i The transmission cost is

[0041]

[0042] in, For preset low latency channel P j The transfer fee for executing a transfer instruction; For the transmission task G i The number of transfer instructions included;

[0043] Preset low latency channel P j Complete the transmission task G i The transmission benefit obtained is

[0044]

[0045] in, The transmission fee paid for the car audio DSP, For preset low latency channel P j Due to the transmission task G i Delay compensation costs resulting from delayed completion;

[0046] Several preset low-latency channels complete the transmission task set G = {G1, G2, ... G m The total transfer income Q obtained from all tasks in} is:

[0047]

[0048] The total transmission cost M paid by several preset low-latency channels:

[0049]

[0050] Where n is the number of preset low-latency channels; m is the number of transmission tasks;

[0051] Based on the fixed transmission fee of each preset low-latency channel, the management goal of resource management for several transmission tasks according to the total transmission revenue and total transmission fee is A:

[0052]

[0053] Based on the genetic algorithm, the management target is approximately solved within the polynomial time complexity to obtain the solution result, which is used as the allocation relationship between several preset low-latency channels and several transmission tasks, and signal transmission is performed based on the allocation relationship.

[0054] According to some embodiments of the present invention, an adaptive fusion method is used to perform signal fusion processing on the first noise reduction signal and the second noise reduction signal to obtain a target signal.

[0055] The present invention proposes a low-latency audio signal processing method for an in-vehicle audio DSP, which improves the efficiency and accuracy of signal noise reduction and facilitates reducing signal processing delay. While improving the noise reduction effect on the signal, the amount of signal noise is also greatly reduced, thereby reducing the amount of signal transmission, which facilitates reducing transmission delay. When transmitting signals, different types of signals are transmitted based on preset low-latency channels, further reducing transmission delay, improving the rapidity and accuracy of signal processing, and achieving low signal delay.

[0056] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0057] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0059] Figure 1 is a flow chart of a method for low-delay processing of audio signals by a vehicle-mounted audio DSP according to an embodiment of the present invention;

[0060] Figure 2 is a flow chart of obtaining a first separated signal and a second separated signal according to an embodiment of the present invention;

[0061] Figure 3 is a flow chart of obtaining a first noise reduction signal according to an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0063] like Figure 1 As shown, the embodiment of the present invention proposes a low-delay processing method for audio signals of a vehicle-mounted audio DSP, comprising steps S1-S5:

[0064] S1, the vehicle audio DSP receives an audio signal and decomposes the audio signal to obtain a first separated signal and a second separated signal;

[0065] S2. Perform noise reduction processing on the first separated signal to obtain a first noise reduction signal;

[0066] S3, performing noise reduction processing on the second separated signal to obtain a second noise reduction signal;

[0067] S4, performing signal fusion processing on the first noise reduction signal and the second noise reduction signal to obtain a target signal;

[0068] S5. Classify the target signal to obtain several classification sets; generate a transmission task for each classification set, and execute several transmission tasks based on a preset low-latency channel to transmit the signal.

[0069] Working principle of the above technical solution: In this embodiment, the first separated signal belongs to a high-frequency signal. The second separated signal belongs to a low-frequency signal. The first separated signal and the second separated signal are processed by different noise reduction methods to obtain a first noise reduction signal and a second noise reduction signal. The first noise reduction signal and the second noise reduction signal are subjected to signal fusion processing to obtain a target signal; the target signal is classified, including: classification based on a pre-trained signal classification model. The signal classification model is obtained, including: data preprocessing: preparing a signal data set and performing preprocessing, including noise removal, normalization, feature extraction and other operations. Feature extraction: extracting features from the signal, which can be time domain features or frequency domain features. Feature extraction methods include Fourier transform, wavelet transform, time domain statistical features, etc. Data labeling: labeling the corresponding category label for each signal sample, that is, data labeling. Data partitioning: dividing the data set into a training set and a test set, using cross-validation or holdout method. Model selection: selecting a suitable classification model, such as support vector machine (SVM), random forest, deep neural network, etc. Model training: Use the training set to train the selected model and adjust the model parameters to improve performance. Model evaluation: Use the test set to evaluate the trained model and calculate indicators such as accuracy, precision, and recall. Model application: Use the trained model to classify and predict new signal data. Signal transmission for different types of signals is based on the preset low-latency channel. The low-latency channel is a transmission channel determined by designing and optimizing the transmission channel to reduce the delay time introduced during data transmission.

[0070] The beneficial effects of the above technical solution are: improving the efficiency and accuracy of signal noise reduction, which is convenient for reducing signal processing delay. When improving the noise reduction effect on the signal, the amount of signal noise is also greatly reduced, thereby reducing the amount of signal transmission, which is convenient for reducing transmission delay. When transmitting signals, different types of signals are transmitted based on preset low-latency channels, which further reduces transmission delays, improves the speed and accuracy of signal processing, and achieves low signal delay.

[0071] like Figure 2 As shown, according to some embodiments of the present invention, the audio signal is decomposed to obtain a first separated signal and a second separated signal, including steps S11-S14:

[0072] S11, performing audio encoding on the audio signal, and determining a time domain feature of the audio signal based on the encoding result;

[0073] S12, performing multi-scale mapping processing on the time domain features of the audio signal to determine the multi-scale features of the audio signal;

[0074] S13, decomposing the audio signal by a binary tree decomposition method consisting of a high-pass filter and a low-pass filter based on the multi-scale feature to obtain a plurality of sub-audio signals;

[0075] S14. Calculate the ratio of the average energy of the frequency line of each sub-audio signal to the average energy of the amplitude spectrum, and compare it with the preset separation threshold, and take the sub-audio signal whose ratio is greater than the preset separation threshold as the first separation signal; take the sub-audio signal whose ratio is less than or equal to the preset separation threshold as the second separation signal.

[0076] Working principle of the above technical solution: In this embodiment, the time domain characteristics of the audio signal include: waveform shape: represents the change of the amplitude of the audio signal over time; envelope: the envelope describes the overall trend of the amplitude of the audio signal over time. The envelope can be used to understand the overall change characteristics of the audio signal, such as the duration, fluctuations and other information of the audio signal. Energy distribution: The distribution of energy can help analyze the energy distribution of the audio signal in different time periods, so as to understand the dynamic range of the audio signal and other information. Impulse response: The impulse response describes the response of the audio signal to a specific input pulse.

[0077] In this embodiment, multi-scale mapping processing is performed on the time domain features of the audio signal to determine the multi-scale features of the audio signal, including: using time-frequency analysis methods, such as short-time Fourier transform (STFT), time-frequency transform, etc., to perform multi-scale mapping processing on the time domain features of the audio signal, analyze the audio signal in time and power, and capture signal characteristics at different scales.

[0078] In this embodiment, the audio signal is decomposed and processed by a binary tree decomposition method composed of a high-pass filter and a low-pass filter based on multi-scale features to obtain a plurality of sub-audio signals, including: filtering the audio signal through a high-pass filter and a low-pass filter respectively to obtain a high-frequency signal and a low-frequency signal. The low-frequency signal is filtered again through a high-pass filter and a low-pass filter to obtain a second-level sub-signal, wherein the low-frequency signal continues to be filtered twice to obtain a third-level sub-signal, and so on, forming a binary tree decomposition structure. Finally, the high-frequency signals of each level and the low-frequency signals of the last level constitute a plurality of sub-audio signals, and these sub-audio signals contain information of the original audio signal at different scales. Multi-scale decomposition in this way can help us better understand the characteristics of the audio signal in different frequency ranges.

[0079] In this embodiment, the average frequency line energy is the average value of the energy of each frequency component in the spectrum of the sub-audio signal; the average amplitude spectrum energy is the average value of the sum of the amplitude spectrum energies of multiple sub-audio signals. The amplitude spectrum energy itself refers to the sum of the energy of all frequency components of an audio signal in the frequency domain, which can be calculated by summing the squares of the amplitudes of each frequency component in the spectrum of the signal (usually obtained through Fourier transform). The sub-audio signal whose ratio is greater than the preset separation threshold is taken as the first separation signal, and the signal is identified again, indicating that the first separation signal belongs to a high-frequency signal; the sub-audio signal whose ratio is less than or equal to the preset separation threshold is taken as the second separation signal, and the signal is identified again, indicating that the second separation signal belongs to a low-frequency signal, so as to facilitate accurate determination of the first separation signal and the second separation signal.

[0080] The beneficial effects of the above technical solution are as follows: determining the time domain features of the audio signal, performing multi-scale mapping processing, and determining the multi-scale features of the audio signal; decomposing the audio signal based on the multi-scale features, determining the information of the original audio signal at different scales, and obtaining a number of sub-audio signals. The ratio of the average energy value of the frequency line to the average energy value of the amplitude spectrum of each sub-audio signal is calculated, and compared with the preset separation threshold, and the first separation signal and the second separation signal are accurately determined according to the comparison result.

[0081] like Figure 3 As shown, according to some embodiments of the present invention, performing noise reduction processing on the first separated signal to obtain a first noise reduction signal includes steps S21-S24:

[0082] S21, extracting signal energy distribution information of frequency points in each audio signal frame in the first separation signal;

[0083] S22, performing noise estimation on each audio signal frame according to the signal energy distribution information to obtain an estimated noise of each audio signal;

[0084] S23, inputting the estimated noise into a pre-trained noise recognition model, and outputting the noise characteristics of each audio signal and the corresponding noise reduction gain;

[0085] S24. Perform noise reduction processing on the corresponding audio signal according to the noise reduction gain to obtain a first noise reduction signal.

[0086] Working principle of the above technical solution: In this embodiment, extracting the signal energy distribution information of the frequency points in each audio signal frame in the first separated signal includes: performing audio signal framing processing on the first separated signal. Usually, a fixed-length window function is used to frame the signal to generate multiple audio signal frames. Fourier transform is performed on each audio signal frame to convert the audio signal from the time domain to the frequency domain to obtain the spectrum information of each frame of the audio signal. According to the spectrum information of each audio signal frame, the signal energy of each frequency point is calculated. According to the calculated spectrum energy information, the signal energy distribution information of the frequency point in each audio signal frame can be extracted.

[0087] In this embodiment, a signal segment whose spectrum energy is lower than a preset threshold is determined as a noise frequency segment, and the range of the noise frequency segment is determined. For the determined noise frequency segment, the energy mean of each audio signal frame in the frequency segment is calculated as the estimated value of the noise energy, so that the estimated noise energy of each frame of the audio signal can be obtained and the estimated noise can be determined.

[0088] In this embodiment, the estimated noise is input into a pre-trained noise recognition model, and the noise characteristics and corresponding noise reduction gains of each audio signal are output, including: the noise recognition model is a deep learning model, such as a neural network. The estimated noise is input into a pre-trained noise recognition model, and the model will perform noise feature recognition on each audio signal frame and output corresponding noise feature information. Based on the noise feature information output by the model, the noise reduction gain of each audio signal frame can be further calculated. The noise reduction gain generally indicates what kind of gain processing needs to be performed on the original signal to reduce or remove noise. The corresponding audio signal is subjected to noise reduction processing according to the noise reduction gain to obtain a first noise reduction signal.

[0089] The beneficial effects of the above technical solution are as follows: for the first separated signal, by extracting signal energy distribution information, noise estimation is performed to obtain estimated noise of each audio signal; the estimated noise is input into a pre-trained noise recognition model, the noise characteristics of each audio signal and the corresponding noise reduction gain are output, and the corresponding audio signal is subjected to noise reduction processing according to the noise reduction gain to obtain a first noise reduction signal, thereby facilitating improving the accuracy of noise reduction.

[0090] According to some embodiments of the present invention, performing noise reduction processing on the corresponding audio signal according to the noise reduction gain to obtain a first noise reduction signal includes:

[0091] The audio signal is segmented to obtain P sub-signals, and each sub-signal is extracted to determine the maximum audio value and the minimum audio value in each sub-signal to obtain a feature matrix of the audio signal;

[0092] Performing denoising processing on the feature matrix according to the denoising gain to obtain a first denoised signal;

[0093]

[0094] Among them, Bi , t is the value of the ith row and t column of the feature matrix corresponding to the first denoised signal; Ai , t is the value of the i-th row and t-th column of the characteristic matrix of the audio signal; e is a natural constant, i = 1, 2, 3 ... P, t = 1, 2; A s,t is the value of the sth row and tth column of the characteristic matrix of the audio signal; A j,t is the value of the j-th row and t-th column of the feature matrix of the audio signal; H is the noise reduction gain corresponding to the audio signal.

[0095] The working principle and beneficial effects of the above technical solution are as follows: the audio signal is segmented to obtain P sub-signals, each sub-signal is extracted, the maximum audio value and the minimum audio value in each sub-signal are determined, and a feature matrix of the audio signal is obtained; the feature matrix is ​​subjected to denoising according to the denoising gain to obtain a first denoised signal; the corresponding audio signal is subjected to denoising according to the denoising gain to facilitate improving the denoising effect of each audio signal, and when performing specific denoising, the feature matrix is ​​subjected to denoising according to the denoising gain, thereby achieving denoising of the signal and obtaining an accurate first denoised signal.

[0096] According to some embodiments of the present invention, performing noise reduction processing on the second separated signal to obtain a second noise-reduced signal includes:

[0097] Analyze the second separated signal to determine frequency distribution information and amplitude change information of the second separated signal;

[0098] determining a target filter based on the frequency distribution information;

[0099] determining a suppression parameter of a noise suppressor according to the amplitude variation information;

[0100] The second separated signal is subjected to noise reduction processing according to the target filter and the noise suppressor to obtain a second noise reduction signal.

[0101] The working principle and beneficial effects of the above technical solution: In this embodiment, the second separated signal is analyzed to determine the frequency distribution information and amplitude change information of the second separated signal, including: using Fourier transform or other spectrum analysis methods to convert the second separated signal from the time domain to the frequency domain, and analyzing the spectrum diagram to identify the main frequency components in the signal and find the peaks with higher or lower frequencies, thereby obtaining the frequency distribution information of the second separated signal. Based on envelope detection or amplitude modulation analysis, the amplitude envelope of the second separated signal is extracted to obtain the overall amplitude change trend of the signal and the amplitude change characteristics of the signal, including the increase, decrease, stability or fluctuation of the amplitude.

[0102] In this embodiment, a time-frequency analysis method (such as wavelet transform, time-frequency distribution analysis, etc.) is used to simultaneously observe the changes of the second separated signal in the time domain and the frequency domain, and the target filter is determined according to the frequency distribution information. The target filter corresponds to the filter of the frequency distribution information, which is convenient for efficient filtering. While filtering, the suppression parameters of the noise suppressor are determined according to the amplitude change information; the noise is suppressed based on the noise suppressor. The second separated signal is subjected to noise reduction processing according to the target filter and the noise suppressor to obtain a second noise reduction signal. It is convenient to obtain an accurate second noise reduction signal.

[0103] According to some embodiments of the present invention, before classifying the target signal, the method further includes: evaluating the sound quality of the target signal, determining an evaluation value, and comparing it with a preset evaluation value, and judging whether the target signal is qualified based on the comparison result.

[0104] The working principle and beneficial effects of the above technical solution: Before classifying the target signal, it also includes: evaluating the sound quality of the target signal, determining the evaluation value, and comparing it with the preset evaluation value, and judging whether the target signal is qualified according to the comparison result. Check whether the fused signal is qualified, and if it is unqualified, it needs to be reprocessed, which improves the accuracy of the signal output.

[0105] According to some embodiments of the present invention, performing a plurality of transmission tasks based on a preset low-latency channel to perform signal transmission includes:

[0106] A matching relationship between the transmission task corresponding to each classification set and the corresponding preset low-latency channel is established, and signal transmission is performed based on the matching relationship as a matching result.

[0107] The working principle and beneficial effects of the above technical solution: different transmission tasks are transmitted based on the corresponding preset low-latency channels, realizing type matching transmission, improving the efficiency and accuracy of signal transmission, and reducing signal transmission delay.

[0108] According to some embodiments of the present invention, the further comprising:

[0109] Get the first time information of the car audio DSP receiving the audio signal;

[0110] Acquire second time information when a plurality of transmission tasks are executed based on a preset low-latency channel to transmit a signal;

[0111] Determine a processing delay of the vehicle audio DSP according to the first time information and the second time information, and compare it with a preset processing delay threshold;

[0112] When it is determined that the processing delay is greater than a preset processing delay threshold, an alarm prompt is issued.

[0113] The working principle and beneficial effects of the above technical solution are as follows: obtaining the first time information of the vehicle audio DSP receiving the audio signal; obtaining the second time information when the vehicle audio DSP performs several transmission tasks for signal transmission based on the preset low-latency channel; determining the processing delay of the vehicle audio DSP according to the first time information and the second time information, and comparing it with the preset processing delay threshold; when it is determined that the processing delay is greater than the preset processing delay threshold, issuing an alarm prompt. It is convenient to adjust the parameters of the vehicle audio DSP in time and perform fault detection to ensure low-latency signal processing.

[0114] According to some embodiments of the present invention, performing a plurality of transmission tasks based on a preset low-latency channel to perform signal transmission includes:

[0115] Determine the control relationship and transmission cost between the preset low-latency channels according to the preset low-latency channel control relationship diagram;

[0116] The transfer task G i Assigned to the preset low latency channel P j Perform the calculation, that is, assign the matrix elements x i，j =1, then preset low latency channel P j Complete the transmission task G i The transmission cost is

[0117]

[0118] in, For preset low latency channel P j The transfer fee for executing a transfer instruction; For the transmission task G i The number of transfer instructions included;

[0119] Preset low latency channel P j Complete the transmission task G i The transmission benefit obtained is

[0120]

[0121] in, The transmission fee paid for the car audio DSP, For preset low latency channel P j Due to the transmission task G i Delay compensation costs resulting from delayed completion;

[0122] Several preset low-latency channels complete the transmission task set G = {G1, G2, ... G m The total transfer income Q obtained from all tasks in} is:

[0123]

[0124] The total transmission cost M paid by several preset low-latency channels:

[0125]

[0126] Where n is the number of preset low-latency channels; m is the number of transmission tasks;

[0127] Based on the fixed transmission fee of each preset low-latency channel, the management goal of resource management for several transmission tasks according to the total transmission revenue and total transmission fee is A:

[0128]

[0129] Based on the genetic algorithm, the management target is approximately solved within the polynomial time complexity to obtain the solution result, which is used as the allocation relationship between several preset low-latency channels and several transmission tasks, and signal transmission is performed based on the allocation relationship.

[0130] The working principle and beneficial effects of the above technical solution are as follows: the control relationship and transmission cost between the preset low-latency channels are determined according to the preset low-latency channel control relationship diagram; the total transmission revenue obtained by all tasks in the transmission task set and the total transmission cost paid by several preset low-latency channels are determined, and a management goal for resource management of several transmission tasks based on the total transmission revenue and the total transmission cost is constructed based on the fixed transmission cost of each preset low-latency channel, and a corresponding function is constructed. Based on the genetic algorithm, the management goal is approximately solved within the polynomial time complexity to obtain the solution result, including: determining the specific problem of the management goal is to obtain the maximum transmission revenue. Encode the solution space of the problem, such as binary coding, real number coding, etc., to ensure that it is suitable for the processing of the genetic algorithm. Randomly generate an initial population, and each individual represents a solution to the problem. Calculate the fitness of each individual, that is, the value of the objective function. The higher the fitness, the better the individual's solution to the problem. Select a part of excellent individuals as parents to participate in the reproduction of the next generation of individuals according to the fitness. Select a certain proportion of individuals for crossover operation to generate a new group of individuals to increase the diversity of the population. Perform mutation operations on individuals and introduce new changes to help the population escape from the local optimal solution. Reselect a new population based on fitness to reproduce the next generation. Determine whether the algorithm has converged based on the set number of iterations or convergence conditions. When the algorithm stops iterating, output the final approximate solution as the solution result of the management target. Use the solution result as the allocation relationship between several preset low-latency channels and several transmission tasks, and perform signal transmission based on the allocation relationship. This is convenient for improving the efficiency and accuracy of signal transmission, while obtaining the maximum transmission benefit and improving the utilization rate of preset low-latency channel resources.

[0131] According to some embodiments of the present invention, an adaptive fusion method is used to perform signal fusion processing on the first noise reduction signal and the second noise reduction signal to obtain a target signal.

[0132] The working principle and beneficial effects of the above technical solution: The weight coefficients of each signal during signal fusion are calculated through an adaptive algorithm (such as the minimum mean square error method, the adaptive gradient method, etc.). The first noise reduction signal and the second noise reduction signal are linearly combined according to the determined weight coefficients to obtain the target signal. Signal fusion can be expressed as target signal = weight coefficient 1*first noise reduction signal + weight coefficient 2*second noise reduction signal. The fused target signal is analyzed, such as observing the spectral characteristics and time domain characteristics of the signal, to ensure that the fusion effect meets expectations and meets the requirements for signal characteristics. When it is determined that the requirements for signal characteristics are met, the target signal is obtained.

[0133] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A low-delay processing method for audio signals of a vehicle-mounted audio DSP, characterized in that: include: The vehicle-mounted audio DSP receives the audio signal, decomposes and processes the audio signal, and obtains a first separated signal and a second separated signal; Performing noise reduction processing on the first separated signal to obtain a first noise reduction signal; Performing noise reduction processing on the second separated signal to obtain a second noise reduction signal; Performing signal fusion processing on the first noise reduction signal and the second noise reduction signal to obtain a target signal; The target signal is classified to obtain several classification sets; each classification set generates a transmission task, and several transmission tasks are executed based on a preset low-latency channel to transmit the signal; Decomposing the audio signal to obtain a first separated signal and a second separated signal includes: Performing audio encoding on the audio signal, and determining a time domain feature of the audio signal based on the encoding result; Performing multi-scale mapping processing on the time domain features of the audio signal to determine the multi-scale features of the audio signal; Based on the multi-scale features, the audio signal is decomposed and processed by a binary tree decomposition method composed of a high-pass filter and a low-pass filter to obtain a plurality of sub-audio signals; The ratio of the average energy of the frequency line of each sub-audio signal to the average energy of the amplitude spectrum is calculated and compared with the preset separation threshold, and the sub-audio signal whose ratio is greater than the preset separation threshold is used as the first separation signal; the sub-audio signal whose ratio is less than or equal to the preset separation threshold is used as the second separation signal.

2. The method for low-delay processing of audio signals by a vehicle-mounted audio DSP as claimed in claim 1, characterized in that: Performing noise reduction processing on the first separated signal to obtain a first noise reduction signal includes: Extracting signal energy distribution information of frequency points in each audio signal frame in the first separation signal; Performing noise estimation on each audio signal frame according to signal energy distribution information to obtain estimated noise of each audio signal; Inputting the estimated noise into a pre-trained noise recognition model, and outputting the noise characteristics of each audio signal and the corresponding noise reduction gain; The corresponding audio signal is subjected to noise reduction processing according to the noise reduction gain to obtain a first noise reduction signal.

3. The method for low-delay processing of audio signals by a vehicle-mounted audio DSP as claimed in claim 2, characterized in that: Performing noise reduction processing on the corresponding audio signal according to the noise reduction gain to obtain a first noise reduction signal, including: The audio signal is segmented to obtain P sub-signals, and each sub-signal is extracted to determine the maximum audio value and the minimum audio value in each sub-signal to obtain a feature matrix of the audio signal; Performing denoising processing on the feature matrix according to the denoising gain to obtain a first denoised signal; Among them, B i,t is the value of the i-th row and t-th column of the feature matrix corresponding to the first denoised signal; i,t is the value of the i-th row and t-th column of the characteristic matrix of the audio signal; e is a natural constant, i = 1, 2, 3 ... P, t = 1, 2; A s,t is the value of the sth row and tth column of the characteristic matrix of the audio signal; A j,t is the value of the j-th row and t-th column of the feature matrix of the audio signal; H is the noise reduction gain corresponding to the audio signal.

4. The method for low-delay processing of audio signals of a vehicle-mounted audio DSP according to claim 1, characterized in that: Performing noise reduction processing on the second separated signal to obtain a second noise reduction signal includes: Analyze the second separated signal to determine frequency distribution information and amplitude change information of the second separated signal; determining a target filter based on the frequency distribution information; determining a suppression parameter of a noise suppressor according to the amplitude variation information; The second separated signal is subjected to noise reduction processing according to the target filter and the noise suppressor to obtain a second noise reduction signal.

5. The method for low-delay processing of audio signals of a vehicle-mounted audio DSP as claimed in claim 1, characterized in that: Before classifying the target signal, the method further includes: evaluating the sound quality of the target signal, determining an evaluation value, and comparing the evaluation value with a preset evaluation value, and judging whether the target signal is qualified according to the comparison result.

6. The method for low-delay processing of audio signals of a vehicle-mounted audio DSP as claimed in claim 1, characterized in that: Perform several transmission tasks based on the preset low-latency channel for signal transmission, including: A matching relationship between the transmission task corresponding to each classification set and the corresponding preset low-latency channel is established, and signal transmission is performed based on the matching relationship as a matching result.

7. The method for low-delay processing of audio signals of a vehicle-mounted audio DSP according to claim 1, characterized in that: Also includes: Get the first time information of the car audio DSP receiving the audio signal; Acquire second time information when a plurality of transmission tasks are executed based on a preset low-latency channel to transmit a signal; Determine a processing delay of the vehicle audio DSP according to the first time information and the second time information, and compare it with a preset processing delay threshold; When it is determined that the processing delay is greater than a preset processing delay threshold, an alarm prompt is issued.

8. The method for low-delay processing of audio signals of a vehicle-mounted audio DSP as claimed in claim 1, characterized in that: Perform several transmission tasks based on the preset low-latency channel for signal transmission, including: Determine the control relationship and transmission cost between the preset low-latency channels according to the preset low-latency channel control relationship diagram; The transfer task G i Assigned to the preset low latency channel P j Perform the calculation, that is, assign the matrix elements x i，j =1, then preset low latency channel P j Complete the transmission task G i The transmission cost is in, For preset low latency channel P j The transfer fee for executing a transfer instruction; For the transmission task G i The number of transfer instructions included; Preset low latency channel P j Complete the transmission task G i The transmission benefit obtained is in, The transmission fee paid for the car audio DSP, For preset low latency channel P j Due to the transmission task G i Delay compensation costs resulting from delayed completion; Several preset low-latency channels complete the transmission task set G = {G1, G2, ... G m The total transfer income Q obtained from all tasks in} is: The total transmission cost M paid by several preset low-latency channels: Where n is the number of preset low-latency channels; m is the number of transmission tasks; Based on the fixed transmission fee of each preset low-latency channel, the management goal of resource management for several transmission tasks according to the total transmission revenue and total transmission fee is A: Based on the genetic algorithm, the management target is approximately solved within the polynomial time complexity to obtain the solution result, which is used as the allocation relationship between several preset low-latency channels and several transmission tasks, and signal transmission is performed based on the allocation relationship.

9. The method for low-delay processing of audio signals of a vehicle-mounted audio DSP as claimed in claim 1, characterized in that: The first noise reduction signal and the second noise reduction signal are fused by an adaptive fusion method to obtain a target signal.

Images