CN111699700A - Audio signal processing method, apparatus and storage medium - Google Patents

Abstract

An audio signal processing method, apparatus and storage medium, by processing an analog audio signal to be processed using a plurality of preprocessing circuits to obtain a plurality of channels of digital audio signals (S101); performing gain compensation on the multiple paths of digital audio signals, and performing frequency domain conversion on the multiple paths of digital audio signals after the gain compensation to acquire multiple paths of frequency domain data (S102); determining frequency domain fusion data according to the multi-path frequency domain data (S103); and converting the frequency domain fusion data into a time domain audio signal, and acquiring an output audio signal according to the time domain audio signal (S104). Analog audio signals to be processed are processed through different preprocessing circuits, system comprehensive noises of the different preprocessing circuits are not related to each other and are also not related to the analog audio signals to be processed, the proportion of effective signals can be improved and the proportion of the system comprehensive noises can be reduced after the audio signals to be processed are fused in a frequency domain, and therefore the signal-to-noise ratio of a recording system is effectively improved.

Description

Audio signal processing method, device and storage medium

technical field

Embodiments of the present invention relate to the field of signal processing, and in particular, to an audio signal processing method, device, and storage medium.

Background technique

The dynamic range of the recording system is an important indicator to characterize the sound pressure level range of the recording, that is, the minimum sound and the maximum sound of the undistorted recording. The lower limit of the dynamic range is related to the system noise floor. The smaller the system noise floor is, the lower the lower limit of the dynamic range is. The noise floor of the system usually includes the noise floor of the microphone, the noise floor of the circuit, the external electromagnetic field crosstalk noise, and the digital quantization noise. When the acoustic signal is less than the system noise floor, the signal is completely submerged, so that it cannot be identified; when the acoustic signal is slightly larger than the noise floor, the signal-to-noise ratio of the effective acoustic signal is low. Therefore, reducing the noise floor of the recording system is an effective way to improve the recording quality of low-volume sound signals.

The recording system in the prior art generally converts the received acoustic signal into an electrical signal by a microphone, that is, the analog signal of the microphone is amplified by an analog amplifier, and the analog signal is converted into a digital signal by an analog-to-digital converter, and the subsequent digital signal is carried out. deal with. When the analog signal passes through the amplifier, it will introduce system noise such as circuit noise, electromagnetic crosstalk noise, and digital quantization noise (excluding microphone noise). In the prior art, reducing the system noise floor and improving the signal-to-noise ratio usually requires better hardware, circuit and structural design, which increases the design cost and design limitations.

SUMMARY OF THE INVENTION

Embodiments of the present invention provide an audio signal processing method, device, and storage medium, so as to effectively reduce the noise floor of the recording system and improve the signal-to-noise ratio of the recording.

A first aspect of the embodiments of the present invention provides an audio signal processing method, including:

Utilize a plurality of preprocessing circuits to process the analog audio signals to be processed to obtain multi-channel digital audio signals, wherein each of the plurality of preprocessing circuits includes an amplifier and an analog-to-digital converter;

Perform gain compensation on the multi-channel digital audio signal, and perform frequency domain conversion on the multi-channel digital audio signal after the gain compensation to obtain multi-channel frequency domain data;

Determine frequency domain fusion data according to the multi-channel frequency domain data;

Converting the frequency-domain fusion data into a time-domain audio signal, and obtaining an output audio signal according to the time-domain audio signal.

A second aspect of the embodiments of the present invention is to provide an audio signal processing device, including: a memory and a processor;

the memory is used to store program codes;

The processor calls the program code, and when the program code is executed, is configured to perform the following operations:

Utilize a plurality of preprocessing circuits to process the analog audio signals to be processed to obtain multi-channel digital audio signals, wherein each of the plurality of preprocessing circuits includes an amplifier and an analog-to-digital converter;

Perform gain compensation on the multi-channel digital audio signal, and perform frequency domain conversion on the multi-channel digital audio signal after the gain compensation to obtain multi-channel frequency domain data;

Determine frequency domain fusion data according to the multi-channel frequency domain data;

Converting the frequency-domain fusion data into a time-domain audio signal, and obtaining an output audio signal according to the time-domain audio signal.

A third aspect of the embodiments of the present invention provides a recording system, including:

a microphone for capturing audio analog signals; and

The audio processing device of the second aspect.

A fourth aspect of the embodiments of the present invention is to provide a computer-readable storage medium on which a computer program is stored, and the computer program is executed by a processor to implement the method described in the first aspect.

The audio signal processing method, device, and storage medium provided in this embodiment obtain multi-channel digital audio signals by using multiple preprocessing circuits to process analog audio signals to be processed, wherein each of the multiple preprocessing circuits pre- The processing circuit includes an amplifier and an analog-to-digital converter; performs gain compensation on the multi-channel digital audio signals, and performs frequency domain conversion on the multi-channel digital audio signals after the gain compensation to obtain multi-channel frequency domain data; according to the multi-channel digital audio signals The frequency domain data determines frequency domain fusion data; converts the frequency domain fusion data into a time domain audio signal, and obtains an output audio signal according to the time domain audio signal. In the method provided in this embodiment, the analog audio signal to be processed is processed by different preprocessing circuits, and considering that the system integrated noises of different preprocessing circuits are not correlated with each other and are not correlated with the analog audio signal to be processed, After fusion in the frequency domain, the proportion of effective signals can be increased, and the proportion of system integrated noise can be reduced, thereby effectively improving the signal-to-noise ratio of the recording system.

Description of drawings

In order to illustrate the technical solutions in the embodiments of the present invention more clearly, the following briefly introduces the accompanying drawings used in the description of the embodiments. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained from these drawings without creative labor.

1 is a flowchart of an audio signal processing method provided by an embodiment of the present invention;

2 is a schematic diagram of a parallel connection of multiple preprocessing circuits provided by an embodiment of the present invention;

3 is a flowchart of an audio signal processing method provided by another embodiment of the present invention;

4 is a flowchart of an audio signal processing method provided by another embodiment of the present invention;

5 is a flowchart of an audio signal processing method provided by another embodiment of the present invention;

6a is a time domain signal diagram of a digital audio signal of a preprocessing circuit;

Fig. 6b is the time-spectrogram of the digital audio signal of the preprocessing circuit;

Fig. 7a is the time domain signal diagram of output audio signal;

Fig. 7b is the time-spectrogram of the output audio signal;

FIG. 8 is a structural diagram of an audio signal processing device provided by an embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

It should be noted that when a component is considered to be "connected" to another component, it can be directly connected to another component or there may be an intervening component at the same time.

Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terms used herein in the description of the present invention are for the purpose of describing specific embodiments only, and are not intended to limit the present invention. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.

Some embodiments of the present invention will be described in detail below with reference to the accompanying drawings. The embodiments described below and features in the embodiments may be combined with each other without conflict.

Embodiments of the present invention provide an audio signal processing method. FIG. 1 is a flowchart of an audio signal processing method provided by an embodiment of the present invention. As shown in FIG. 1 , the audio signal processing method in this embodiment may include:

Step S101 , using a plurality of preprocessing circuits to process the analog audio signal to be processed to obtain a multi-channel digital audio signal, wherein each of the plurality of preprocessing circuits includes an amplifier and an analog-to-digital converter preprocessing circuit.

In this embodiment, as shown in FIG. 2 , multiple preprocessing circuits are connected in parallel, and each preprocessing circuit includes an amplifier and an analog-to-digital converter. It should be noted that the multiple preprocessing circuits include at least two preprocessing circuits. processing circuit; wherein the amplifier can be used to amplify the power of the analog audio signal, and the amplification factor is usually expressed by the gain. Can be used to convert analog audio signals into digital audio signals for subsequent signal processing.

In this embodiment, as shown in FIG. 2 , the analog audio signal x to be processed can be collected by a microphone, and then input to a plurality of parallel preprocessing circuits, respectively amplified by power with different gains, and converted into a digital audio signal x ₁ , ...x _i ..., x _I , where I is the number of preprocessing circuits, and system integrated noise is introduced in the process of analog audio signal transmission, amplification, and analog-to-digital conversion in different preprocessing circuits, The system integrated noise is equivalent to the system noise floor except the microphone noise floor, including circuit noise, electromagnetic crosstalk noise, digital quantization noise, etc., and the system integrated noise between the preprocessing circuits is usually uncorrelated with each other, and the difference between the preprocessing circuits. The system-integrated noise between , and the analog audio signal to be processed are independent of each other, and the digital audio signal can be expressed as x _i =G _i x+n _i , i=1, 2, . . . , I, where G _i is the preprocessing The analog gain of circuit i, n _i is the system integrated noise of preprocessing circuit i. That is, in this embodiment, the input of each preprocessing circuit is the same analog audio signal to be processed, and the output digital audio signal of each preprocessing circuit is different, thereby obtaining multiple digital audio signals.

In this embodiment, the analog audio signal to be processed may be cut into different segments with a predetermined duration as a frame signal, or a predetermined number of sampled data may be used as a frame signal after analog-to-digital conversion. Processing can be performed in units of one frame signal. In order to ensure the continuity of the signals, there may be a certain overlap between the signals of adjacent frames, that is, the tail of the signal of the previous frame and the head of the signal of the next frame have an overlap amount, so as to establish the correlation between the adjacent frames.

Step S102: Perform gain compensation on the multi-channel digital audio signals, and perform frequency domain conversion on the multi-channel digital audio signals after the gain compensation to obtain multi-channel frequency domain data.

In this embodiment, by performing gain compensation on the multi-channel digital audio signals, the gains of the multi-channel audio signals are realized to be the same, thereby facilitating the frequency domain fusion of the subsequent multi-channel digital audio signals under the same gain. One of the preprocessing circuits is selected as the reference preprocessing circuit, and the analog gain of its amplifier is used as the reference gain, and the multi-channel digital audio signal is subjected to gain compensation according to the reference gain. Further, frequency domain conversion is performed on each channel of digital audio signals in the multi-channel digital audio signals after the gain compensation, so as to obtain frequency domain data corresponding to each channel of digital audio signals. In this embodiment, the multi-channel digital audio signals after the gain compensation are subjected to frequency domain conversion to obtain multi-channel frequency-domain data, thereby realizing the fusion of the multi-channel frequency-domain data in the frequency domain. The frequency domain conversion method may adopt Fourier transform (such as discrete Fourier transform), Laplace transform, Z transform, etc. The specific frequency domain conversion process will not be repeated here.

Step S103: Determine frequency domain fusion data according to the multi-channel frequency domain data.

In this embodiment, the system integrated noise between the preprocessing circuits is generally uncorrelated, and the system integrated noise between the preprocessing circuits and the analog audio signal to be processed are uncorrelated with each other. During fusion, the effective signal and system integrated noise can be identified according to the similarity between each frequency domain data. Specifically, at a certain sampling point of the frequency domain data during fusion, if the degree of correlation is greater, the frequency domain data will be fused. The higher the ratio of valid signals at the sampling point, the higher the ratio of valid signals in the frequency-domain fusion data, and the lower the ratio of system integrated noise.

Step S104: Convert the frequency-domain fusion data into a time-domain audio signal, and obtain an output audio signal according to the time-domain audio signal.

In this embodiment, after the frequency-domain fusion data is obtained, the frequency-domain fusion data can be converted into a time-domain audio signal, and the conversion method can adopt the inverse transform of Fourier transform (such as discrete Fourier transform), Lapp The inverse transformation of the Lass transform, the inverse transformation of the Z transform, etc. will not be repeated here. After the conversion is completed, the output audio signal can be obtained according to the time-domain audio signal, and in the process of obtaining the output audio signal according to the time-domain audio signal, operations such as compression and noise reduction can be performed on the time-domain audio signal. In addition, if the above-mentioned audio signal processing flow is performed in units of one frame of signal in this embodiment, in the process of obtaining the output audio signal according to the time-domain audio signal, it is necessary to splicing each frame signal to establish the correlation between adjacent frames. Specifically, the time-domain superimposition processing may be performed on the time-domain audio signal of the current frame and the time-domain audio signal of the previous frame before the previous frame.

In the audio signal processing method provided by this embodiment, a plurality of preprocessing circuits are used to process an analog audio signal to be processed to obtain a multi-channel digital audio signal, wherein each of the plurality of preprocessing circuits includes an amplifier and a an analog-to-digital converter; performing gain compensation on the multi-channel digital audio signals, and performing frequency domain conversion on the multi-channel digital audio signals after the gain compensation to obtain multi-channel frequency domain data; determining the frequency domain according to the multi-channel frequency domain data domain fusion data; converting the frequency domain fusion data into a time domain audio signal, and obtaining an output audio signal according to the time domain audio signal. In the method provided in this embodiment, the analog audio signal to be processed is processed by different preprocessing circuits, and considering that the system integrated noises of different preprocessing circuits are not correlated with each other and are not correlated with the analog audio signal to be processed, After fusion in the frequency domain, the proportion of effective signals can be increased, and the proportion of system integrated noise can be reduced, thereby effectively improving the signal-to-noise ratio of the recording system.

On the basis of any of the foregoing embodiments, the step S102 to perform gain compensation on the multi-channel digital audio signal may include:

Gain compensation is performed on each channel of digital audio signal according to the gain compensation coefficient corresponding to each channel of digital audio signal in the multichannel digital audio signal, wherein the gain compensation coefficient corresponding to each channel of digital audio signal in the multichannel digital audio signal is different from each other .

In this embodiment, in order to realize that the gains of the multi-channel audio signals are the same and facilitate the frequency domain fusion of the subsequent multi-channel digital audio signals under the same gain, the gain corresponding to each channel of the digital audio signals in the multi-channel digital audio signals can be obtained first. compensation coefficient, and then perform gain compensation for each channel of digital audio signal according to its corresponding gain compensation coefficient. Since the analog gains of the amplifiers included in each preprocessing circuit may be different, if the multi-channel digital audio signals amplified by different analog gains are adjusted to the same gain, the gain compensation coefficients corresponding to each channel of digital audio signals will be different. Are not the same.

Further, the gain compensation coefficient corresponding to each channel of digital audio signal is determined according to the reference gain and the analog gain of the amplifier included in the preprocessing circuit corresponding to each channel of digital audio signal, wherein the reference gain is a plurality of preprocessing steps. The analog gain of the amplifier included in the reference preprocessing circuit in the circuit.

In this embodiment, any one of the multiple preprocessing circuits can be selected as the reference preprocessing circuit, and the analog gain of the amplifier included in the reference preprocessing circuit can be used as the reference gain. The analog gain of the amplifier included in the preprocessing circuit corresponding to the signal, and then the ratio between the reference gain and the analog gain is used as the gain compensation coefficient corresponding to the digital audio signal, and then when the digital audio signal is subjected to gain compensation, The digital audio signal after the gain compensation can be obtained by multiplying the digital audio signal by the gain compensation coefficient.

On the basis of any of the foregoing embodiments, as shown in FIG. 3 , the determining of frequency domain fusion data according to the multi-channel frequency domain data includes:

Step S201, obtaining the correlation parameter between every two channels of frequency domain data in the multi-channel frequency domain data;

Step S202: Determine frequency domain fusion data according to the correlation parameter between the two channels of frequency domain data and the two channels of frequency domain data corresponding to the correlation parameter.

In this embodiment, for any two channels of frequency domain data in the multichannel frequency domain data, the correlation parameters at each sampling point of the two channels of frequency domain data are obtained, and then the two channels of frequency domain data are obtained according to the correlation parameters. Each sampling point is superimposed to obtain the superposition result of the two channels of frequency domain data. If the correlation parameter of a certain sampling point is larger, the superposition result of the sampling point is relatively larger, and if the correlation parameter is smaller, the The superposition result of the sampling points is relatively smaller; finally, the superposition results of any two channels of frequency domain data in the multi-channel frequency domain data are accumulated, and then the frequency domain fusion data can be obtained.

In this embodiment, the correlation parameters at each sampling point of any two channels of frequency domain data can be obtained by the following formula:

Among them, C _ij (n) is the correlation parameter of the frequency domain data X′ _i and X′ _j at the nth sampling point, and X′ _j (n) is the signal of the nth sampling point of the frequency domain data X′ _j , X′ _i (n) is the signal of the n-th adopted point of the frequency domain data X′ _i , I is the number of preprocessing circuits, the operation min(a, b) is the smaller value of a and b, and the operation | ·| represents the modulus of a complex number. 0≤C _ij (n) _l ≤ 1, a larger C _ij (n) indicates that the signals of the frequency domain data X′ _i and X′ _j at the nth sampling point are more similar, and the ratio of valid signals is higher.

It should be noted that the correlation parameter is not limited to the correlation parameter obtained by the above formula, and the ratio of the signals of any two channels of frequency domain data at each sampling point may also be used as the correlation parameter.

Further, the frequency domain fusion data can be obtained by the following formula:

In this formula, the signals of the nth sampling point of any two channels of frequency domain data X' _i and X' _j can be superimposed according to the correlation parameter, that is, X' _i (n)+X' _j (n), Then multiply the correlation parameter C _ij (n) to obtain the superposition result of the two channels of frequency domain data at the nth sampling point. The larger the correlation parameter, the larger the superposition result of the sampling point. is small, the superposition result of the sampling point is relatively small; finally, the superposition results of any two channels of frequency domain data in the multi-channel frequency domain data are accumulated, and then the frequency domain fusion data can be obtained, which improves the efficiency of the frequency domain fusion data. The ratio of the signal to the system reduces the ratio of the integrated noise of the system, thereby improving the signal-to-noise ratio.

Further, since the frequency domain fusion data usually has symmetry, when obtaining the frequency domain fusion data, only the unilateral spectrum data of the frequency domain fusion data can be obtained, and by converting the unilateral spectral data into the multilateral spectrum data, as the frequency domain Fusion data.

On the basis of any of the above embodiments, the method further includes:

performing phase compensation on the multi-channel frequency domain data;

The determining the frequency domain fusion data according to the multi-channel frequency domain data includes:

The frequency domain fusion data is determined according to the multi-channel frequency domain data after phase compensation.

In this embodiment, the multi-channel digital audio signal may have a certain time offset due to the circuit design or sampling logic of multiple pre-processing circuits, and the frequency domain corresponds to the phase offset, that is, the multi-channel frequency domain. There may be a phase offset between the data, and in the above embodiment, when the frequency domain fusion data is obtained from the multichannel frequency domain data, the similarity comparison needs to be performed on the same sampling points of the multichannel frequency domain data. Therefore, in this embodiment, before the frequency-domain fusion data is determined according to the multi-channel frequency-domain data, it is necessary to perform phase compensation on the multi-channel frequency-domain data. The frequency domain data determines the frequency domain fusion data.

Further, as shown in FIG. 4 , the performing phase compensation on the multi-channel frequency domain data includes:

Step S301, determining the phase difference or time difference between each channel of frequency domain data in the multichannel frequency domain data and the reference frequency domain data in the multichannel frequency domain data;

Step S302: Perform phase compensation on the corresponding one channel and multiple channels of frequency domain data according to the phase difference or time difference.

In this embodiment, any channel of frequency-domain data in the multi-channel frequency-domain data can be selected as the reference frequency-domain data, and the phase difference or time difference between each channel of frequency-domain data in the multi-channel frequency-domain data and the reference frequency-domain data is obtained, Then phase compensation is performed on the frequency domain data of each channel according to the corresponding phase difference or time difference of the channel. The phase compensation can use linear phase compensation.

It should be noted that, if there is no time delay between multiple preprocessing circuits, phase compensation is not required.

On the basis of any of the foregoing embodiments, the time-domain audio signal is a time-domain audio signal of the current frame, wherein the acquiring the output audio signal according to the time-domain audio signal includes:

performing superimposition processing on the current frame time domain audio signal and the historical frame time domain audio signal obtained before the current frame time domain audio signal to obtain the current frame time domain fusion audio signal;

The output audio signal is determined according to the time-domain fusion audio signal of the current frame.

In this embodiment, the above-mentioned audio signal processing procedures are performed in units of one frame signal. In order to ensure the continuity of the signals, there may be a certain overlap between adjacent frame signals, that is, the tail of the previous frame signal and the next frame signal are overlapped. The headers of the frame signals have an amount of overlap, thereby establishing a correlation between adjacent frames. Therefore, after converting the current frame frequency domain fusion data into the current frame time domain audio signal in S104, the overlapping part of the current frame time domain audio signal and the previous frame time domain audio signal can be overlapped and superimposed. The non-overlapping part of the time domain audio signal and the previous frame time domain audio signal does not perform superposition operation, thereby obtaining the current frame time domain fusion audio signal, and then the output audio signal can be determined according to the current frame time domain fusion audio signal.

On the basis of any of the above-mentioned embodiments, performing frequency domain conversion on the multi-channel digital audio signals after gain compensation to obtain multi-channel frequency domain data includes:

Windowing is performed on the multi-channel digital audio signals after the gain compensation;

Perform frequency domain conversion according to the windowed multi-channel digital audio signals to obtain multi-channel frequency domain data.

In this embodiment, an analysis window function is used to perform windowing processing on each channel of digital audio signals, wherein the analysis window function includes but is not limited to a sine window function, a rectangular window function, a triangular window function, a Hanning window function (Hanning), a Gaussian window function Window function (Gaussian), Blackman window function (Blackman), Chebyshev window function (Chebyshev), Hamming window function (Hamming), flat top window function (Flap Top), Kaiser window function (Kaiser). Of course, the windowing process may not be performed, which is equivalent to adding a rectangular window. .

Further, the obtaining the output audio signal according to the time domain audio signal includes:

performing windowing processing on the time-domain audio signal;

The output audio signal is determined from the time-domain audio signal after the windowing process.

In this embodiment, the time-domain audio signal is windowed by using a synthesis window function, wherein the synthesis window function includes but is not limited to a sine window function, a rectangular window function, a triangular window function, a Hanning window function (Hanning), a Gaussian window function Gaussian, Blackman, Chebyshev, Hamming, Flap Top, Kaiser. Of course, the windowing process may not be performed, which is equivalent to adding a rectangular window.

Embodiments of the present invention provide an audio signal processing method. FIG. 5 is a flowchart of an audio signal processing method provided by another embodiment of the present invention. As shown in FIG. 5 , on the basis of the foregoing embodiment, the method in this embodiment may include:

Step S401 , using a plurality of preprocessing circuits to process the analog audio signal to be processed to obtain a multi-channel digital audio signal.

Wherein, each of the plurality of preprocessing circuits includes an amplifier and an analog-to-digital converter, and the analog gains of the amplifiers of each preprocessing circuit may be different from each other, and may of course be the same.

The preprocessed analog audio signal x( _t ), t is the sampling discrete time series t=1, 2, . In the audio signal, the analog gains of the amplifiers included in the i-th preprocessing circuit are respectively G _i , where i=1, 2, . , Amplification, and analog-to-digital conversion, the system integrated noise is introduced, including circuit noise, electromagnetic crosstalk noise, digital quantization noise, etc. The system integrated noise of each channel is n _i (t), that is, the digital audio signal output by each channel. x _i (t) is as follows:

x _i (t)=G _i x(t)+n _i (t) i=1,2,...,I

Step S402 , extracting N sampling points at every interval of M sampling points for each digital audio signal in the multi-channel digital audio signal, as a frame of digital audio signal.

In this embodiment, the first frame of digital audio signal of the i-th preprocessing circuit is denoted as:

x _i (t) _l , t=1, 2, . . . , N.

Among them, N is called the frame length, M is called the frame shift, 0<M<N, that is, the last NM sampling points of the l-1th frame are the same as the first NM sampling points of the lth frame, 0.005f _s <N < f _s , and N is a power of 2.

Step S403 , performing gain compensation and windowing processing on the multi-channel digital audio signal.

In this embodiment, in order to achieve the same gain of the multi-channel audio signals, the gain compensation coefficient corresponding to each channel of digital audio signals in the multi-channel digital audio signals can be obtained. In this embodiment, one channel is selected from multiple preprocessing circuits. As the reference preprocessing circuit, without loss of generality, the first preprocessing circuit is used as the reference preprocessing circuit, and the analog gain G ₁ of the amplifier it includes is used as the reference gain to obtain the digital audio signal of a certain channel of digital audio signal. The analog gain G _i of the amplifier included in the corresponding preprocessing circuit takes the ratio G ₁ /G _i between the reference gain and the analog gain as the gain compensation coefficient corresponding to the digital audio signal of the channel, and then according to the gain compensation coefficient for the channel. Digital audio signal for gain compensation.

In this embodiment, multi-channel digital audio signals can also be windowed, wherein the analysis window function includes but is not limited to a sine window function, a rectangular window function, a triangular window function, a Hanning window function (Hanning), a Gaussian window function ( Gaussian), Blackman, Chebyshev, Hamming, Flap Top, Kaiser. Of course, the windowing process may not be performed, which is equivalent to adding a rectangular window.

It should be noted that, in this embodiment, the gain compensation may be performed first and then the windowing process may be performed, or the windowing process may be performed first and then the gain compensation may be performed, or the gain compensation and the windowing process may be performed simultaneously. The lth frame digital audio signal x′ _i (t) _l of the i-th preprocessing circuit after gain compensation and windowing processing is:

Among them, h _ana (t) is the N-point analysis window function.

Step S404: Perform frequency domain conversion on the multi-channel digital audio signals after the gain compensation to obtain multi-channel frequency domain data.

In this embodiment, discrete Fourier transform is performed on x′ _i (t) _l to obtain the i-th frequency domain data X _i (n) _l :

为虚数单位。本实施例中可采用快速傅里叶变换进行加速计算。Among them, n is a discrete spectrum sequence, e is a natural constant,

is an imaginary unit. In this embodiment, fast Fourier transform may be used to perform accelerated calculation.

Step S405: Perform phase compensation on the multi-channel frequency domain data.

In this embodiment, any channel of frequency-domain data in the multi-channel frequency-domain data can be selected as the reference frequency-domain data. Without loss of generality, the first channel of frequency-domain data is used as the reference frequency-domain data. The phase difference between each channel of frequency domain data and the reference frequency domain data. In this embodiment, the phase difference P _{d, i} (n) _l of the ith frequency domain data X _i (n) _l of the l th frame and the reference frequency domain data at the n th sampling point can be obtained according to the following formula:

Here, the operation (·) ^* represents the conjugate of the complex number, and the operation angle(·) represents the phase of the complex number.

Then, phase compensation is performed on the corresponding one-way multi-channel frequency-domain data according to the phase difference, and the i-th multi-channel frequency-domain data linear compensation phase spectrum P _{c of the 1 th frame, i} (n) _l is:

进而第l帧第i路相位补偿复数谱P_i(n)_l为：Among them, 1≤n ₁ ≤n ₂ ≤N/2, the parameters between P _d,i (n ₁ ) _l and P _d,i (n ₂ ) _l are used for linear phase fitting. This formula is a linear phase fitting formula with a constant term of 0, P _{d, i} (n) _l is the phase value of the nth sampling point, and the slope after linear fitting is

Then the i-th phase compensation complex spectrum P _i (n) _l of the l-th frame is:

Therefore, the multi-channel frequency domain data X′ _i (n) _l after phase compensation is:

X′ _i (n) _l =P _i (n) _l X _i (n) _l

It should be noted that, if there is no time delay between multiple preprocessing circuits, no phase compensation is required, that is, P _i (n) _l =1. In addition, if the time difference between each frequency domain data is fixed and known, that is, the time difference between multiple preprocessing circuits is fixed and known, for example, the time difference between the i-th frequency domain data and the first frequency domain data is ahead is τ _i , the phase compensation can be performed according to the time difference, and the linear compensation phase spectrum P _{c of the i-th multi-channel frequency-domain data of the 1 th frame above, i} (n) _l can be:

P _{c, i} (n) _l = 2πf _s τ _i n/N n = 1, 2, ..., N

Step S406: Determine frequency domain fusion data according to the multi-channel frequency domain data after phase compensation.

In this embodiment, the system integrated noise between the preprocessing circuits is usually uncorrelated with each other, and the system integrated noise between the preprocessing circuits and the analog audio signal to be processed are uncorrelated with each other. At the same time, the effective signal and the system integrated noise can be identified according to the similarity between the frequency domain data. Specifically, for the frequency domain data X′ _i (n) _l and X′ _j (n) _l after any two phase compensation , obtain the correlation parameter C _ij (n) _l of the two channels of frequency domain data at the nth sampling point:

Among them, the operation min(a, b) is the smaller value of a and b, and the operation |·| represents the modulus of the complex number. 0≤C _ij (n) _l ≤ 1, a larger C _ij (n) indicates that the signals of the frequency domain data X′ _i and X′ _j at the nth sampling point are more similar, and the ratio of valid signals is higher.

Further, the 1st frame unilateral frequency domain fusion data can be obtained by the following formula:

Among them, 1≤n≤N/2+1.

Due to the symmetry of the frequency domain fusion data, the unilateral spectral data can be converted into multilateral spectral data, and as the frequency domain fusion data, the frequency domain fusion data X _syn (n) _l of the first frame is as follows:

Step S407: Convert the frequency domain fusion data into a time domain audio signal.

In this embodiment, the 1st frame frequency domain fusion data X _syn (n) ₁ is processed through the inverse transform of discrete Fourier transform to obtain the 1 th frame time domain audio signal x′ _syn (t) ₁ :

Specifically, the inverse transform of the fast Fourier transform can be used to accelerate the calculation.

Step S408, performing windowing processing on the time-domain audio signal.

In this embodiment, windowing is performed on the time-domain audio signal x′ _syn (t) _l of the lth frame to obtain a windowed time-domain audio signal

Among them, h _syn (t) is an N-point synthesis window function, wherein the synthesis window function includes but is not limited to a sine window function, a rectangular window function, a triangular window function, a Hanning window function (Hanning), a Gaussian window function (Gaussian), Black Blackman, Chebyshev, Hamming, FlapTop, Kaiser. Of course, the windowing process may not be performed, which is equivalent to adding a rectangular window.

Step S409, performing superposition processing on the current frame time domain audio signal and the historical frame time domain audio signal obtained before the current frame time domain audio signal to obtain the current frame time domain fusion audio signal, and according to the current frame time domain fusion audio signal Get the output audio signal.

In this embodiment, the overlapping part of the current frame time domain audio signal and the previous frame time domain audio signal can be overlapped and superimposed to obtain the current frame time domain fusion audio signal:

Among them, x _syn (t+M) _l-1 is the time domain signal of the 1-1th frame overlapping and accumulating N points; if l=1, x _syn (t+M) _l-1 =0. Therefore, after the lth frame overlaps and accumulates M points, the current frame output audio signal y(n) _l of the lth frame corresponding to the overlapped M point is output as:

y(n) _l = x _syn (t) _l t = 1, 2, ..., M

After the output audio signal y(n) _l is obtained, it can be used as the output audio signal, and the recording can be saved or played in real time.

倍，信噪比提高了

倍。若合理配置各预处理电路放大器的增益以及增加相关性检测，则抑制系统底噪的效果更佳。In the above embodiment, there is a special case that the analog gains of the amplifiers included in all the preprocessing circuits are the same, the system integrated noise of each preprocessing circuit obeys a normal distribution, and the noise energy is expected to be consistent, and the correlation coefficient C _ij ( k) are all 1, then the noise energy of the output audio signal is expected to be the sum of the noise energy of the single preprocessing circuit

times, the signal-to-noise ratio is improved

times. If the gain of each preprocessing circuit amplifier is properly configured and the correlation detection is increased, the effect of suppressing the noise floor of the system will be better.

The specific principles and implementation manners of the audio signal processing method provided by the embodiments of the present invention are similar to those of the foregoing embodiments, and details are not described herein again.

In the audio signal processing method provided by this embodiment, a plurality of preprocessing circuits are used to process an analog audio signal to be processed to obtain a multi-channel digital audio signal, wherein each of the plurality of preprocessing circuits includes an amplifier and a an analog-to-digital converter; performing gain compensation on the multi-channel digital audio signals, and performing frequency domain conversion on the multi-channel digital audio signals after the gain compensation to obtain multi-channel frequency domain data; determining the frequency domain according to the multi-channel frequency domain data domain fusion data; converting the frequency domain fusion data into a time domain audio signal, and obtaining an output audio signal according to the time domain audio signal. In the method provided in this embodiment, the analog audio signal to be processed is processed by different preprocessing circuits, and considering that the system integrated noises of different preprocessing circuits are not correlated with each other and are not correlated with the analog audio signal to be processed, After fusion in the frequency domain, the proportion of effective signals can be increased, and the proportion of system integrated noise can be reduced, thereby effectively improving the signal-to-noise ratio of the recording system.

A specific example is used as an example below. In this example, there are two preprocessing circuits (that is, I=2), and the gains of the two preprocessing circuits are the same, that is, G ₁ =G ₂ . The time difference between them is fixed and is 1 sampling time interval, that is, τ ₂ =1/f _s . The test environment is an ordinary quiet room, and the audio source is a voice signal. The signal sampling rate is f _s =48kHz, the analysis frame length is N=2048, and the frame shift is M=1024. Both the analysis window and the synthesis window are sinusoidal windows:

Since the gains of the two preprocessing circuits are the same, the digital audio signals (time domain signals) of the two preprocessing circuits are similar, as shown in Figure 6a (only one of them is shown), the digital audio signals of the two preprocessing circuits have The time spectrum is also similar, as shown in Figure 6b (only one of them is shown), and after processing by the audio signal processing method of this embodiment, the time domain signal and time spectrum of the output audio signal are shown in Figure 7a and Figure 7b respectively. Show. Comparing the time-domain signals of the digital audio signals in Fig. 6a and Fig. 7a, it can be seen that in the speech signal segment, the difference between the two is not obvious, while in the noise segment, the noise of the output audio signal of this embodiment is significantly smaller than that of the unprocessed digital audio signal noise. Comparing the time spectrum of the digital audio signal in Fig. 6b and Fig. 7b, the signal in Fig. 6b contains white noise covering the whole frequency band, which is the system noise floor, and the output audio signal of this embodiment in Fig. 7b is obviously better than that in Fig. 6b. In the noise section, the system noise floor has been greatly suppressed in the whole frequency band. By comparison, the frequency below 500Hz in the unprocessed digital audio signal is a background noise signal with more low-frequency components, and the output audio signal of this embodiment can basically completely retain the effective signal; The frequency noise components are basically the system noise floor, and the output audio signal in this embodiment can suppress the noise by more than 10 dB. In the speech segment, the effective spectral information can be retained clearly and completely, and even if the system noise floor is mixed between the speech spectra, it can also have a great noise suppression effect.

Embodiments of the present invention provide an audio signal processing device. FIG. 8 is a structural diagram of an audio signal processing device provided by an embodiment of the present invention. As shown in FIG. 8 , the audio signal processing device 50 includes a memory 52 and a processor 51 .

The memory 52 is used to store program codes;

The processor 51 calls the program code, and when the program code is executed, is configured to perform the following operations:

Utilize a plurality of preprocessing circuits to process the analog audio signals to be processed to obtain multi-channel digital audio signals, wherein each of the plurality of preprocessing circuits includes an amplifier and an analog-to-digital converter;

Perform gain compensation on the multi-channel digital audio signal, and perform frequency domain conversion on the multi-channel digital audio signal after the gain compensation to obtain multi-channel frequency domain data;

Determine frequency domain fusion data according to the multi-channel frequency domain data;

Converting the frequency-domain fusion data into a time-domain audio signal, and obtaining an output audio signal according to the time-domain audio signal.

On the basis of any of the above embodiments, when the processor 51 performs gain compensation on the multi-channel digital audio signal, the processor 51 is configured to:

Gain compensation is performed on each channel of digital audio signal according to the gain compensation coefficient corresponding to each channel of digital audio signal in the multichannel digital audio signal, wherein the gain compensation coefficient corresponding to each channel of digital audio signal in the multichannel digital audio signal is different from each other .

On the basis of any of the above embodiments, the gain compensation coefficient corresponding to each channel of digital audio signal is determined according to the reference gain and the analog gain of the amplifier included in the preprocessing circuit corresponding to each channel of digital audio signal, wherein the The reference gain is an analog gain of an amplifier included in the reference preprocessing circuit among the plurality of preprocessing circuits.

On the basis of any of the foregoing embodiments, when the processor 51 determines the frequency domain fusion data according to the multi-channel frequency domain data, the processor 51 is configured to:

Obtain the correlation parameter between every two channels of frequency domain data in the multi-channel frequency domain data;

The frequency domain fusion data is determined according to the correlation parameter between the two channels of frequency domain data and the two channels of frequency domain data corresponding to the correlation parameter.

On the basis of any of the foregoing embodiments, the processor 51 is further configured to:

performing phase compensation on the multi-channel frequency domain data;

When the processor 51 determines the frequency domain fusion data according to the multi-channel frequency domain data, the processor 51 is configured to:

The frequency domain fusion data is determined according to the multi-channel frequency domain data after phase compensation.

On the basis of any of the foregoing embodiments, when the processor 51 performs phase compensation on the multi-channel frequency domain data, the processor 51 is configured to:

determining the phase difference between each channel of frequency domain data in the multichannel frequency domain data and the reference frequency domain data in the multichannel frequency domain data;

Phase compensation is performed on the corresponding one channel and multiple channels of frequency domain data according to the phase difference.

On the basis of any of the above embodiments, the time-domain audio signal is a time-domain audio signal of the current frame, wherein when the processor 51 acquires the output audio signal according to the time-domain audio signal, the processor 51 is configured as:

performing superimposition processing on the current frame time domain audio signal and the historical frame time domain audio signal obtained before the current frame time domain audio signal to obtain the current frame time domain fusion audio signal;

The output audio signal is determined according to the time-domain fusion audio signal of the current frame.

On the basis of any of the above embodiments, when the processor 51 performs frequency domain conversion on the multi-channel digital audio signals after gain compensation to obtain multi-channel frequency domain data, the processor 51 is configured to:

Windowing is performed on the multi-channel digital audio signals after the gain compensation;

Perform frequency domain conversion according to the windowed multi-channel digital audio signals to obtain multi-channel frequency domain data.

On the basis of any of the foregoing embodiments, when the processor 51 acquires the output audio signal according to the time-domain audio signal, the processor 51 is configured to:

performing windowing processing on the time-domain audio signal;

The output audio signal is determined from the time-domain audio signal after the windowing process.

The specific principles and implementation manners of the audio signal processing device provided by the embodiments of the present invention are similar to those of the above-mentioned embodiments, and details are not described herein again.

The audio signal processing device provided in this embodiment obtains multi-channel digital audio signals by processing analog audio signals to be processed by using multiple preprocessing circuits, wherein each of the multiple preprocessing circuits includes an amplifier and a an analog-to-digital converter; performing gain compensation on the multi-channel digital audio signals, and performing frequency domain conversion on the multi-channel digital audio signals after the gain compensation to obtain multi-channel frequency domain data; determining the frequency domain according to the multi-channel frequency domain data domain fusion data; converting the frequency domain fusion data into a time domain audio signal, and obtaining an output audio signal according to the time domain audio signal. In this embodiment, the analog audio signal to be processed is processed by different preprocessing circuits, and considering that the system integrated noises of different preprocessing circuits are not correlated with each other and are not correlated with the analog audio signal to be processed. After the fusion in the domain, the ratio of effective signals can be increased and the ratio of system integrated noise can be reduced, thereby effectively improving the signal-to-noise ratio of the recording system.

An embodiment of the present invention provides a recording system, where the recording system includes: a microphone for collecting analog audio signals; and the audio processing device 50 described in the above-mentioned embodiments.

In addition, this embodiment also provides a computer-readable storage medium on which a computer program is stored, and the computer program is executed by a processor to implement the method described in the foregoing embodiment.

In the several embodiments provided by the present invention, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored, or not implemented. On the other hand, the shown or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, indirect coupling or communication connection of devices or units, and may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution in this embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware, or may be implemented in the form of hardware plus software functional units.

The above-mentioned integrated units implemented in the form of software functional units can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (processor) to execute the methods described in the various embodiments of the present invention. some steps. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and other media that can store program codes .

Those skilled in the art can clearly understand that, for the convenience and brevity of the description, only the division of the above functional modules is used for illustration. The internal structure is divided into different functional modules to complete all or part of the functions described above. For the specific working process of the apparatus described above, reference may be made to the corresponding process in the foregoing method embodiments, and details are not described herein again.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: The technical solutions described in the foregoing embodiments can still be modified, or some or all of the technical features thereof can be equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the embodiments of the present invention. scope.

Claims (20)

1. an audio signal processing method, is characterized in that, comprises: Utilize a plurality of preprocessing circuits to process the analog audio signals to be processed to obtain multi-channel digital audio signals, wherein each of the plurality of preprocessing circuits includes an amplifier and an analog-to-digital converter preprocessing circuit; Perform gain compensation on the multi-channel digital audio signal, and perform frequency domain conversion on the multi-channel digital audio signal after the gain compensation to obtain multi-channel frequency domain data; Determine frequency domain fusion data according to the multi-channel frequency domain data; Converting the frequency-domain fusion data into a time-domain audio signal, and obtaining an output audio signal according to the time-domain audio signal. 2. The method according to claim 1, wherein the performing gain compensation on the multi-channel digital audio signal comprises: Gain compensation is performed on each channel of digital audio signal according to the gain compensation coefficient corresponding to each channel of digital audio signal in the multichannel digital audio signal, wherein the gain compensation coefficient corresponding to each channel of digital audio signal in the multichannel digital audio signal is different from each other . 3. method according to claim 2 is characterized in that, the gain compensation coefficient corresponding to each road digital audio signal is determined according to the analog gain of the amplifier that the preprocessing circuit corresponding to each road digital audio signal comprises reference gain and described each road digital audio signal , wherein the reference gain is an analog gain of an amplifier included in the reference preprocessing circuit in the plurality of preprocessing circuits. 4. The method according to any one of claims 1-3, wherein the determining the frequency-domain fusion data according to the multi-channel frequency-domain data comprises: Obtain the correlation parameter between every two channels of frequency domain data in the multi-channel frequency domain data; The frequency domain fusion data is determined according to the correlation parameter between the two channels of frequency domain data and the two channels of frequency domain data corresponding to the correlation parameter. 5. The method according to any one of claims 1-4, wherein the method further comprises: performing phase compensation on the multi-channel frequency domain data; The determining the frequency domain fusion data according to the multi-channel frequency domain data includes: The frequency domain fusion data is determined according to the multi-channel frequency domain data after phase compensation. 6. The method according to claim 5, wherein the performing phase compensation on the multi-channel frequency domain data comprises: determining the phase difference or time difference between each channel of frequency domain data in the multichannel frequency domain data and the reference frequency domain data in the multichannel frequency domain data; Phase compensation is performed on the corresponding one channel and multiple channels of frequency domain data according to the phase difference or time difference. 7. The method according to any one of claims 1-6, wherein the time-domain audio signal is a current-frame time-domain audio signal, wherein the acquiring an output audio signal according to the time-domain audio signal, include: performing superimposition processing on the current frame time domain audio signal and the historical frame time domain audio signal obtained before the current frame time domain audio signal to obtain the current frame time domain fusion audio signal; The output audio signal is determined according to the time-domain fusion audio signal of the current frame. 8. The method according to any one of claims 1-7, wherein the performing frequency domain conversion on the multi-channel digital audio signals after the gain compensation to obtain multi-channel frequency domain data, comprising: Windowing is performed on the multi-channel digital audio signals after the gain compensation; Perform frequency domain conversion according to the windowed multi-channel digital audio signals to obtain multi-channel frequency domain data. 9. The method according to any one of claims 1-8, wherein the acquiring an output audio signal according to the time-domain audio signal comprises: performing windowing processing on the time-domain audio signal; The output audio signal is determined from the time-domain audio signal after the windowing process. 10. An audio signal processing device, comprising: a memory and a processor; the memory is used to store program codes; The processor calls the program code, and when the program code is executed, is configured to perform the following operations: Utilize a plurality of preprocessing circuits to process the analog audio signals to be processed to obtain multi-channel digital audio signals, wherein each of the plurality of preprocessing circuits includes an amplifier and an analog-to-digital converter; Perform gain compensation on the multi-channel digital audio signal, and perform frequency domain conversion on the multi-channel digital audio signal after the gain compensation to obtain multi-channel frequency domain data; Determine frequency domain fusion data according to the multi-channel frequency domain data; Converting the frequency-domain fusion data into a time-domain audio signal, and obtaining an output audio signal according to the time-domain audio signal. 11. The device according to claim 10, wherein when the processor performs gain compensation on the multi-channel digital audio signal, the processor is configured to: Gain compensation is performed on each channel of digital audio signal according to the gain compensation coefficient corresponding to each channel of digital audio signal in the multichannel digital audio signal, wherein the gain compensation coefficient corresponding to each channel of digital audio signal in the multichannel digital audio signal is different from each other . 12. The device according to claim 11, wherein the gain compensation coefficient corresponding to each channel of digital audio signal is determined according to the reference gain and the analog gain of the amplifier included in the preprocessing circuit corresponding to each channel of digital audio signal , wherein the reference gain is an analog gain of an amplifier included in the reference preprocessing circuit in the plurality of preprocessing circuits. 13. The device according to any one of claims 10-12, wherein when the processor determines the frequency-domain fusion data according to the multi-channel frequency-domain data, the processor is configured to: Obtain the correlation parameter between every two channels of frequency domain data in the multi-channel frequency domain data; The frequency domain fusion data is determined according to the correlation parameter between the two channels of frequency domain data and the two channels of frequency domain data corresponding to the correlation parameter. 14. The device according to any one of claims 10-13, wherein the processor is further configured to: performing phase compensation on the multi-channel frequency domain data; When the processor determines the frequency domain fusion data according to the multi-channel frequency domain data, the processor is configured to: The frequency domain fusion data is determined according to the multi-channel frequency domain data after phase compensation. 15. The device according to claim 14, wherein when the processor performs phase compensation on the multi-channel frequency domain data, the processor is configured to: determining the phase difference or time difference between each channel of frequency domain data in the multichannel frequency domain data and the reference frequency domain data in the multichannel frequency domain data; Phase compensation is performed on the corresponding one channel and multiple channels of frequency domain data according to the phase difference or time difference. 16. The device according to any one of claims 10-15, wherein the time-domain audio signal is a time-domain audio signal of a current frame, wherein the processor obtains and outputs an output according to the time-domain audio signal audio signal, the processor is configured to: performing superimposition processing on the current frame time domain audio signal and the historical frame time domain audio signal obtained before the current frame time domain audio signal to obtain the current frame time domain fusion audio signal; The output audio signal is determined according to the time-domain fusion audio signal of the current frame. 17. The device according to any one of claims 10-16, wherein when the processor performs frequency domain conversion on the multi-channel digital audio signals after gain compensation to obtain multi-channel frequency domain data, the The processor is configured to: Windowing is performed on the multi-channel digital audio signals after the gain compensation; Perform frequency domain conversion according to the windowed multi-channel digital audio signals to obtain multi-channel frequency domain data. 18. The device according to any one of claims 10-17, wherein when the processor obtains the output audio signal according to the time-domain audio signal, the processor is configured to: performing windowing processing on the time-domain audio signal; The output audio signal is determined from the time-domain audio signal after the windowing process. 19. A recording system, characterized in that, comprising: a microphone for capturing audio analog signals; and An audio processing device as claimed in any one of claims 10-18. 20. A computer-readable storage medium, characterized in that a computer program is stored thereon, and the computer program is executed by a processor to implement the method according to any one of claims 1-9.

Images