CN118400652A

CN118400652A - Voice noise reduction method and earphone

Info

Publication number: CN118400652A
Application number: CN202311640538.4A
Authority: CN
Inventors: 杨钦雲; 李倩
Original assignee: Bestechnic Shanghai Co Ltd
Current assignee: Bestechnic Shanghai Co Ltd
Filing date: 2023-12-01
Publication date: 2024-07-26

Abstract

The application provides a voice noise reduction method and an earphone. The voice noise reduction method is applied to the earphone, the earphone at least comprises a first microphone and a second microphone, and the voice noise reduction method comprises the following steps: acquiring a target voice signal, wherein the target voice signal comprises a first voice signal acquired by the first microphone and a second voice signal acquired by the second microphone; calculating differential beam forming results of the first voice signal and the second voice signal in a plurality of null modes; the preset null directions of different null modes are different; taking the minimum amplitude value of the differential beam forming result corresponding to each null pattern as a target directivity gain result; and denoising the target voice signal by using the target directivity gain result. The voice noise reduction method can improve the array directivity and the noise reduction effect of noise reduction.

Description

Voice noise reduction method and earphone

Technical Field

The application relates to the field of voice noise reduction, in particular to a voice noise reduction method and an earphone.

Background

The microphone array signal processing can be applied to voice noise reduction, and when the microphone array signal processing technology is applied to some small-size earphones to reduce the voice noise, the problem that the array directivity and the noise reduction effect cannot reach expectations exists.

At present, a generalized sidelobe canceling algorithm is used for improving array directivity and noise reduction effects. In the generalized sidelobe canceling algorithm, fixed beam forming is performed on the voice signal collected by the microphone, but the currently used fixed beam forming algorithm has defects. For example, the delay-and-sum beamforming algorithm has poor directivity for low-frequency speech signals. The white noise gain of the super-directional beam forming algorithm at a low frequency is low, the beam response is reduced along with the increase of the frequency, so that the noise reduction effect is poor, the differential beam forming algorithm has high-pass characteristics on low-frequency signals, and has a frequency-invariant beam response mode in a full frequency band, the processed signals need to be compensated, the distance between array elements of a microphone is required to be small enough, and otherwise, frequency spectrum aliasing can occur.

Therefore, the fixed beam forming algorithm used in the current voice noise reduction has the problems of poor array directivity and/or poor noise reduction effect.

Disclosure of Invention

In view of the foregoing, the present application is directed to a voice noise reduction method and an earphone to improve array directivity and noise reduction effect for noise reduction of voice signals.

In a first aspect, an embodiment of the present application provides a method for noise reduction in a voice, which is applied to an earphone, where the earphone includes at least a first microphone and a second microphone, and a direction in which the first microphone points to the second microphone or a direction in which the second microphone points to the first microphone is an end-fire direction, and the method for noise reduction in a voice includes: acquiring a target voice signal, wherein the target voice signal comprises a first voice signal acquired by the first microphone and a second voice signal acquired by the second microphone; calculating differential beam forming results of the first voice signal and the second voice signal in a plurality of null modes; the preset null directions of different null modes are different; taking the minimum amplitude value of the differential beam forming result corresponding to each null pattern as a target directivity gain result; and denoising the target voice signal by using the target directivity gain result.

In the generalized sidelobe canceling algorithm, the larger the main lobe width is, the worse the array directivity is, the higher the sidelobe height is, the worse the noise suppression capability is, the main lobe width and the sidelobe height of the differential beam forming results corresponding to different directions are different.

In an embodiment, the predetermined nulling directions corresponding to the different nulling modes include a 90 ° direction, a 120 ° direction, and a 180 ° direction compared to the endfire direction.

In the embodiment of the application, the null refers to the differential wave beam forming device for carrying out directional wave trapping on signals, the value of the interference signal mapped to the end-shot direction is required to be calculated, cos90 DEG is 0, the absolute value of cos180 DEG is 1, and the absolute value of cos120 DEG is 0.5, so that the coverage area of the null is more complete by selecting the three directions, the differential wave beam forming result obtained by calculation can be effectively improved, the characteristics of main lobes and side lobes in the directions from 90 DEG to 180 DEG can be integrated, the height and the width of the side lobes obtained by calculation can be smaller, and the array directivity and the noise reduction effect of noise reduction can be effectively improved.

In one embodiment, the calculating the differential beamforming results of the first voice signal and the second voice signal in a plurality of null modes includes: aiming at each null pattern, aligning an interference signal of the second voice signal in a preset null direction corresponding to the null pattern with an interference signal of the first voice signal in a preset null direction corresponding to the null pattern; the interference signal is a signal in a non-end-emission direction in the first voice signal and the second voice signal; and calculating the difference between the first voice signal and the aligned second voice signal to obtain the differential beam forming result.

In the embodiment of the application, the interference signals of the first voice signal and the second voice signal in the null direction are aligned, so that the interference signals can be effectively eliminated when the differential beam forming result is calculated, and the accuracy of the differential beam forming result is improved.

In one embodiment, the method further comprises: acquiring a frequency and steering vector matrix calculation formula of the second voice signal; the aligning the interference signal of the second voice signal in the null direction corresponding to the preset null direction of the null mode with the interference signal of the first voice signal in the null direction corresponding to the preset null direction of the null mode includes: calculating a steering vector matrix corresponding to the null pattern based on a physical distance between the first microphone and the second microphone, a preset null direction corresponding to the null pattern, a frequency of the second voice signal and the steering vector matrix calculation formula; multiplying the second voice signal by the steering vector matrix to obtain a first product, wherein the first product represents that an interference signal of the second voice signal in a preset null direction corresponding to the null mode is aligned with an interference signal of the first voice signal in the preset null direction corresponding to the null mode.

In the embodiment of the application, the steering vector matrix can represent phase information, so that the steering vector matrix representing the phase information between the first voice signal and the second voice signal can be calculated based on the physical distance between the first microphone and the second microphone, the preset null direction corresponding to the null mode, the frequency of the second voice signal and the steering vector matrix calculation formula, and then the second voice signal is multiplied by the steering vector matrix, so that the interference signal of the second voice signal and the interference signal of the first voice signal are aligned in the preset null direction.

In an embodiment, the preset steering vector matrix calculation formula includes:

wherein steer _vector is a steering vector matrix, d is a physical distance between the first microphone and the second microphone, θ ₂ is the preset null direction, c is the sound propagation speed, f is a signal frequency of the second speech signal, j is a complex unit, exp is an exponential function based on a natural constant e.

In the guide vector matrix calculation formula provided by the embodiment of the application, each parameter is a known value and can be quickly obtained, so that the time and the power consumption for calculating the guide vector matrix can be effectively reduced.

In an embodiment, after the minimum amplitude value of the differential beam forming result corresponding to each null pattern is taken as the target directivity gain result, the method further includes: multiplying the target directivity gain result by the phase of the differential beam forming result corresponding to the 180-degree direction null pattern to obtain a second product; the second product is fed into an adaptive filter part of a generalized sidelobe canceling algorithm.

In the embodiment of the application, the steering vector matrix can be used for representing the phase difference information between the first voice signal and the second voice signal, the phase of the differential beam forming result in the direction of 180 degrees is multiplied by the target directivity gain result to obtain a second product, the second product can be used for noise reduction, and the second product is directly sent to the adaptive filtering part of the generalized sidelobe canceling algorithm for further noise reduction, so that the accuracy of noise reduction based on the second product can be further improved. ,

In one embodiment, before calculating the differential beamforming results of the first voice signal and the second voice signal in the plurality of null modes, the method includes: for any null pattern, calculating a first time difference between the arrival of the signal of the target speech signal in the end-fire direction at the first microphone and the second microphone; calculating a second time difference matrix of arrival of the interference signal in the null direction corresponding to the null mode between the first microphone and the second microphone; calculating to obtain a cut-off frequency based on the first time difference, the second time difference matrix and a preset cut-off frequency calculation formula; compensating signals lower than the cut-off frequency in the differential beam forming result corresponding to the null pattern to obtain a compensated differential beam forming result; the compensation is used to amplify the amplitude and phase of signals below the cutoff frequency in the differential beamforming result.

In the embodiment of the application, the differential beam former has the characteristic of high-pass filtering below the cut-off frequency, the signal below the cut-off frequency is weakened, and the data of the part of the differential beam forming result below the cut-off frequency is effective and should be completely reserved. Therefore, the differential beam forming result can be compensated, so that the amplitude value of the differential beam forming result is not lower than the first voice signal below the cut-off frequency, the data integrity is ensured, and the accuracy of the determined target directivity gain result is improved.

In one embodiment, the first time difference is calculated as follows: the second time difference matrix is calculated as follows: the preset cutoff frequency calculation formula comprises: Wherein d is a physical distance between the first microphone and the second microphone, θ ₁ is the end-fire direction, θ ₂ is the preset null direction, c is the sound propagation speed, τ1 is the first time difference, τ2 is the second time difference matrix, and ω _c is the preset cut-off frequency.

In the embodiment of the application, the first time difference matrix and the second time difference matrix can be calculated by the mode, and the preset cut-off frequency can be calculated by the first time difference matrix and the second time difference matrix. The parameters in the formula are easy to obtain, the sound propagation speed is a fixed value, the physical distance between the first microphone and the second microphone is also determined when the microphone array is designed, the mode of calculating the preset cut-off frequency is simple through the formula, and the possibility of overlarge power consumption is reduced.

In an embodiment, after compensating the signal below the cut-off frequency in the differential beam forming result corresponding to the null pattern, the method further includes: and limiting the compensated differential beam forming result so that the amplitude value of the limited differential beam forming result does not exceed the amplitude value of the first voice signal.

In the embodiment of the application, after the differential beam forming result is compensated, the amplitude value of the differential beam forming result may exceed the amplitude value of the original first voice signal, so that the data is inaccurate, the amplitude can be limited so as not to exceed the amplitude value of the first voice signal, the accuracy of the data is improved, and the accuracy of the target directional gain result is improved.

In a second aspect, an embodiment of the present application provides an earphone, including: the microphone array is used for collecting target voice signals and at least comprises a first microphone and a second microphone, wherein the direction of the first microphone pointing to the second microphone or the direction of the second microphone pointing to the first microphone is an end-shooting direction; a differential beamformer; a filtering unit; the processing unit is connected with the microphone array, the differential beam former and the filtering unit; the processing unit is used for acquiring a target voice signal, and the voice signal comprises a first voice signal acquired by the first microphone and a second voice signal acquired by the second microphone; the processing unit is further used for controlling the differential beam former to calculate differential beam forming results of the first voice signal and the second voice signal in a plurality of null modes; the preset null directions of different null modes are different; the processing unit is further used for taking the minimum amplitude value of the differential beam forming result corresponding to each null pattern as a target directivity gain result; the filtering unit is used for reducing noise of the target voice signal based on the target directivity gain result.

In one embodiment, the headphones are TWS (True Wireless Stereo, real Wireless stereo) headphones.

The TWS earphone is generally smaller in size, the distance between microphones is shorter, the arrangement is more compact, and compared with a single-channel voice enhancement mode, the TWS earphone is noise-reduced in the differential mode, and array directivity and noise reduction effect of TWS noise reduction can be effectively improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments of the present application will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and should not be considered as limiting the scope, and other related drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a block diagram of an earphone according to an embodiment of the present application;

FIG. 2 is a flowchart of a method for voice noise reduction according to an embodiment of the present application;

FIG. 3 is a schematic diagram illustrating a direction according to an embodiment of the present application;

FIG. 4 is a diagram illustrating the differential beam forming results according to an embodiment of the present application;

FIG. 5 is a schematic diagram of combining differential beam forming results according to an embodiment of the present application;

FIG. 6 is a schematic diagram illustrating the target directional gain result according to an embodiment of the present application.

Icon: a microphone array 110; a first microphone 111; a second microphone 112; a differential beamformer 120; a filtering unit 130; a processing unit 140; blocking matrix 150.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present application more apparent, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.

First, referring to fig. 1, fig. 1 is a schematic diagram of an earphone according to an embodiment of the present application. As shown in fig. 1, the earphone includes: a microphone array 110, a differential beamformer 120, a filtering unit 130 and a processing unit 140.

The microphone array 110 includes at least a first microphone 111 and a second microphone 112. The first microphone 111 and the second microphone 112 may be the same in type, structure, and the like, and are used for collecting voice signals, where, for illustration, the voice signal collected by the first microphone 111 is a first voice signal, and the voice signal collected by the second microphone 112 is a second voice signal.

In this embodiment, the direction in which the first microphone 111 of the microphone array 110 points to the second microphone 112 or the direction in which the second microphone 112 points to the first microphone 111 is the end-fire direction. In headphones, the extension of the end fire direction is usually aligned with the mouth, i.e. the end fire direction is also the direction pointing towards the mouth.

In this embodiment, the physical distance between the first microphone 111 and the second microphone 112 may be configured according to the earphone type, and the different earphone types have different sizes, which have different physical distance requirements for different microphones, for example, in the TWS earphone, the physical distance between the microphones in the microphone array 110 is less than 2.13cm.

The differential beam former 120 is connected to the first microphone 111 and the second microphone 112, and is configured to perform differential beam forming on the first voice signal and the second voice signal in response to the control of the processing unit 140, so as to obtain a differential beam forming result.

The filtering unit 130 is configured to filter the voice signal.

The processing unit 140 is connected to the first microphone 111, the second microphone 112, the differential beamformer 120, and the filtering unit 130, respectively.

In this embodiment, the processing unit 140 is configured to control the first microphone 111, the second microphone 112, the differential beam former 120, and the filtering unit 130, so that the earphone performs the voice noise reduction method provided in the embodiment of the present application, where the voice noise reduction method is developed later.

The embodiment provided by the application carries out voice noise reduction through a generalized sidelobe elimination algorithm, and the generalized sidelobe elimination algorithm comprises the following steps: performing fixed beam forming on the voice signals acquired by the two microphones in a frequency domain to obtain a processed voice signal Y1; multiplying the voice signals acquired by the two microphones by the blocking matrix 150 in the frequency domain to obtain a reference noise signal Y2; and finally, using an adaptive filtering algorithm, taking Y1 as an algorithm expected signal, taking Y2 as an algorithm input signal, and calculating an output signal to obtain a filtering result. Accordingly, in the earphone, a blocking matrix 150 may also be provided, the blocking matrix 150 being connected with the first microphone 111, the second microphone 112 and the filtering unit 130 to calculate a noise signal using the blocking matrix 150.

The specific principles and steps of the generalized sidelobe canceling algorithm described above, the structures of the differential beamformer 120, the blocking matrix 150 and the filtering unit 130 may refer to the prior art and are not further developed herein. Embodiments provided herein are primarily directed to adjusting a fixed beam forming section.

The earphone provided by the application applies a differential beam forming mode to reduce noise, and is more suitable for some earphones with smaller sizes. For example, the earphone provided by the application can be a TWS earphone, the TWS earphone has the characteristic of small size, and the size of the earphone is smaller, so that the distance between the microphones in the microphone arrangement is smaller, and compared with a single-channel voice enhanced noise reduction mode, the array directivity and the noise reduction effect can be better by applying the voice noise reduction method provided by the application in TWS.

Next, the embodiment of the present application further provides a voice noise reduction method, which can be applied to the processing unit 140 of the earphone provided in the foregoing embodiment. Referring to fig. 2, fig. 2 is a flowchart of a voice noise reduction method according to an embodiment of the application. The voice noise reduction method comprises the following steps:

s110, acquiring a target voice signal.

In this embodiment, the target voice signal includes a first voice signal collected by the first microphone and a second voice signal collected by the second microphone, and the first voice signal and the second voice signal are from the same sound source.

S120, calculating differential beam forming results of the first voice signal and the second voice signal in a plurality of null modes.

Null means that the signal is subjected to directional notching, the notching direction is the null direction, and in this embodiment, the null modes are different in the null direction, for example, the null direction is 90 °, 120 °, 135 °, 150 °, 180 °, and the like. The specific principles and implementations of nulling may be referenced to the prior art and are not further developed herein.

Referring to fig. 3, fig. 3 is a schematic diagram of a direction provided by an embodiment of the present application, in which MIC1 and MIC2 are a first microphone and a second microphone, respectively, and a direction pointing to the second microphone in the first direction is taken as an end-fire direction as an example. The 90-degree null direction is the direction perpendicular to the end-shot direction, the 180-degree null direction is the opposite direction of the end-shot direction, and the 120-degree null direction is the angle between the extension line in the direction and the end-shot direction of 120 degrees.

The differential beam signal is a frequency domain signal, and the value thereof is related to the cosine value, so that the angle may be an angle having the same absolute value as the cosine value, for example, 120 ° may correspond to 60 ° and 45 ° may correspond to 135 °, and the angle provided in the above embodiment is merely an example.

Accordingly, since the differential beam forming result is related to the cosine value, when the null direction of the null mode is selected, the directions of 90 degrees, 120 degrees and 180 degrees can be selected, and the absolute values of the cosine values of the three are respectively 0, 0.5 and 1, and the three values are increased uniformly, so that the calculated differential beam forming result is more uniform, and a larger range is covered. In some other embodiments, 135 °, 150 ° directions may be selected, and the above is merely an example, and should not be construed as limiting the present application.

In an embodiment of the present application, the process of calculating the differential beamforming result may include the following processes: aiming at each null pattern, aligning an interference signal of the second voice signal in the null pattern corresponding to the preset null direction with an interference signal of the first voice signal in the null pattern corresponding to the preset null direction; and calculating the difference between the first voice signal and the aligned second voice signal to obtain a differential beam forming result.

In this embodiment, the interference signal refers to a signal in a non-end-fire direction in the voice signal, where signals of the first voice signal and the second voice signal in a preset null direction are both interference signals.

The differential beam forming result is used for filtering the interference signal in the voice signal, and in the embodiment of the application, the differential beam forming result may be that the difference between the first voice signal and the second voice signal is calculated to filter the interference signal.

The first voice signal and the second voice signal have a certain physical distance between the first microphone and the second microphone respectively, so that a certain phase difference exists between the first voice signal and the second voice signal, if the interference signals between the first voice signal and the second voice signal in the same null direction need to be calculated and filtered, the interference signals of the first voice signal and the second voice signal in the null direction need to be aligned, otherwise, the calculated differential beam forming result is inaccurate.

Thus, for each null pattern, the second speech signal may be aligned with the first speech signal with respect to the interference signal in the null direction corresponding to the null pattern, and in an embodiment of the present application, the alignment process may include: calculating a steering vector matrix corresponding to the null pattern based on a physical distance between the first microphone and the second microphone, a preset null direction corresponding to the null pattern, a frequency of the second voice signal and a steering vector matrix calculation formula; and multiplying the second voice signal by the guide vector matrix to obtain a first product.

In this embodiment, the steering vector matrix is calculated from the frequency of the second speech signal, the distance between the first microphone and the second microphone, and the null direction, where the steering vector matrix may be used to characterize the phase difference between the first speech signal and the second speech signal in the frequency domain, and multiplying the second speech signal by the steering vector matrix may enable the second speech signal to correspond to the first speech signal in the null direction, that is, the first product may characterize that the interference signal of the second speech signal in the preset null direction corresponding to the null mode is aligned with the interference signal of the first speech signal in the preset null direction corresponding to the null mode.

In an embodiment of the present application, the preset steering vector matrix calculation formula may be:

In the earphone provided in the foregoing embodiment, the distance between the microphones in the microphone array is 2.13cm, and then the physical distance d between the first microphone and the second microphone may be 2.13cm, the sound propagation speed is 340m/s, θ ₂ is a preset null direction corresponding to the null mode, for example, θ ₂ may be 90 °, 180 °, 120 ° or the like.

In this embodiment, after the second voice signal is obtained, the frequency of the second voice signal may be calculated, so as to obtain the frequency of the second voice signal. The steering vector matrix calculation formula can be configured into a program in advance, or can be stored in a memory in advance and directly called.

Through the formula, the steering vector matrix for representing the phase information can be obtained through calculation, and then the second voice signal can be multiplied by the steering vector matrix, namely:

mic2_align＝mic2*steer_vector

Wherein mic2 _align is the aligned second speech signal, mic2 is the unaligned second speech signal, and steer _vector is the steering vector matrix.

Accordingly, the calculation formula of the differential beam forming result is: res=mic1-mic2 _align. Where res is the differential beamforming result.

The differential beamformer has the characteristic of high-pass filtering, and the differential beamforming result for frequencies below the cut-off frequency will be attenuated, thus compensating the differential beamforming result for such low frequencies with a preset gain, compensating the amplitude and phase of the signal below the cut-off frequency used in amplifying the differential beamforming result.

In one embodiment of the application, the cut-off frequency is calculated as follows: for any null mode, calculating a first time difference between arrival of a signal of the target voice signal in the end-fire direction at the first microphone and the second microphone; calculating a second time difference matrix of the interference signal in the null direction corresponding to the null mode reaching the first microphone and the second microphone; and calculating the cut-off frequency based on the first time difference, the second time difference matrix and a preset cut-off frequency calculation formula.

Wherein the signal reaching a first time difference between the first microphone and the second microphone may be converted into a relationship between the sound propagation speed, the physical distance between the first microphone and the second microphone, and the first time difference. Correspondingly, the first time difference is calculated as follows:

Wherein τ1 is the first time difference, d is the physical distance between the first microphone and the second microphone, θ1 is the end-fire direction, cos (θ1) =1.

Accordingly, the interference signal reaching the second time difference between the first microphone and the second microphone may also be converted into a relation between the sound propagation speed, the physical distance between the first microphone and the second microphone and the first time difference. Correspondingly, the second time difference matrix is calculated as follows:

wherein τ2 is the first time difference, d is the physical distance between the first microphone and the second microphone, and θ2 is a preset null direction.

The preset cutoff frequency calculation method comprises the following steps:

Wherein ω _c is a preset cutoff frequency.

If the differential beam forming result has a partial frequency omega satisfying 0 < omega _c, the phase and the amplitude of the frequency domain signal of omega in the range in the differential beam forming result are compensated. In this embodiment, the compensation filter may be used to compensate the partial frequency, and the differential beam forming result is input into the compensation filter, so as to obtain the differential beam forming result compensated by the compensation filter.

After compensation, the amplitude value of the differential beamforming result may exceed the amplitude value range of the first speech signal. Therefore, for the differential beamforming result calculated in each null mode, in the embodiment of the present application, the compensated differential beamforming result may be limited such that the amplitude value of the limited differential beamforming result does not exceed the amplitude value of the first speech signal.

Illustratively, if abs (res) > α×abs (mic 1), abs (res) =α×abs (mic 1), where abs is calculated as an amplitude value, α is an adjustable factor, res is a differential beam forming result, and mic1 is a first speech signal.

By limiting the amplitude value, the compensated differential beam forming result can keep a certain accuracy, and the accuracy of the subsequent calculation result is improved.

In the above embodiment, the calculation of the differential beamforming result may be the calculation by the differential beamformer.

S130, taking the minimum amplitude value of the differential beam forming result corresponding to each null pattern as a target directivity gain result.

As described previously, in the generalized sidelobe canceling algorithm, the signal for filtering includes the fixed-beamformed speech signal Y1 and the reference noise signal Y2 obtained by the blocking matrix. Therefore, after the differential beam forming result corresponding to each null pattern is obtained, the differential beam forming result corresponding to each null pattern may be processed to obtain the voice signal Y1.

Referring to the influence of the main lobe and the side lobe in the generalized side lobe elimination algorithm, the width of the main lobe influences the array directivity, the larger the width is, the worse the array directivity is, the higher the side lobe height influences the noise suppression capability, and the worse the suppression capability is, so in the embodiment of the application, the processing needs to reduce the width of the main lobe and the height of the side lobe so as to improve the array directivity and the noise suppression capability.

In the embodiment of the application, the amplitude value of the differential beam forming result corresponding to each null pattern can be taken in the frequency domain, the minimum amplitude value of the three amplitude values is calculated as the target directivity gain result, and the phase of the differential beam forming result corresponding to the null pattern in the 180-degree direction is multiplied, so that the frequency domain signal after differential beam forming, namely the voice signal Y1, can be obtained.

For example, referring to fig. 4, fig. 4 is a schematic diagram of a differential beam forming result provided in an embodiment of the present application, wherein a polar pattern corresponds to a preset null direction 90 °, a superstrate pattern corresponds to a preset null direction 120 °, and a cardioid pattern corresponds to a preset null direction 180 °. After the three graphs are integrated into one graph, a differential beam forming result shown in fig. 5 can be obtained, and after the minimum amplitude value of the three differential beam forming results is taken, a differential beam forming result shown in fig. 6 can be obtained. The main lobe width of the differential beam forming result shown in fig. 6 is smaller compared to the 180 ° and 120 ° and 90 ° null directions, and the side lobe height of the differential beam forming result shown in fig. 6 is lower compared to other null directions.

Therefore, compared with the differential beam forming result obtained by the unidirectional nulling, the differential beam forming result with smaller main lobe width and lower side lobe height can be obtained by the method provided by the embodiment of the application, and the differential beam forming result is used as a target directivity gain result so as to improve noise suppression capability and array directivity.

S140, noise reduction is carried out on the target voice signal by using the target directivity gain result.

After the target directivity gain result is obtained, the target directivity gain result and the phase of the differential beam forming result corresponding to the 180-degree direction null pattern can be multiplied to obtain a second product, and the second product is used as a voice signal Y1 in the generalized sidelobe canceling algorithm to reduce noise of the acquired signal. For specific principles and processes, reference is made to the prior art and will not be further developed herein.

The target directional gain result is a frequency domain amplitude value, and the complete frequency domain signal Y1 can be obtained by multiplying the frequency domain amplitude value by the phase, so that when the target directional gain result is utilized in the generalized sidelobe canceling algorithm, the target directional gain result can be multiplied by the phase of a differential beam forming result corresponding to the 180-degree directional null pattern to obtain a second product, and the second product is sent to the adaptive filtering algorithm part.

Illustratively, the conversion process includes:

res＝[res_90，res_120，res_180]

min_res＝min(abs(res))

fin_res＝min_res*angle(res_180)

Wherein fin_res represents a second product, that is, the frequency domain signal Y1, which is not required to be converted into a time domain signal in the generalized sidelobe canceling algorithm, the second product is directly sent to the filtering module 130, angle represents a phase of a differential beam forming result corresponding to the null pattern in the direction of 180 degrees, res represents a differential beam forming result in different null patterns, and the differential beam forming result is a matrix. res_90, res_120, res_180 represent differential beam forming results _, min_res in the directions of 90 °, 120 °, and 180 °, respectively, and the angle represents the phase of the differential beam forming result in the direction of 180 ° with the res matrix taking the minimum amplitude value as the target directivity gain result.

The second product can be used for noise reduction of a generalized sidelobe algorithm, and the input adaptive filter part can further reduce the noise of the second product, so that the accuracy of noise reduction based on the second product is improved.

In addition, if the second product is a frequency domain signal, if the output time domain signal of the differential fixed beamformer is to be directly obtained, the second product may be subjected to inverse fourier transform, and the frequency domain signal may be converted into a time domain signal, so that noise reduction is performed by the second product based on the time domain signal.

In the embodiment of the application, the differential beam forming results under different null modes are calculated, and the minimum amplitude value of the differential beam forming results corresponding to the different null modes is taken as the target directivity gain result, so that the main lobe width corresponding to the minimum amplitude value is narrower, and the side lobe height is lower, and therefore, the minimum amplitude value is used as the target directivity gain result for filtering, and the array directivity and the noise reduction effect of noise reduction of voice signals can be effectively improved.

The above embodiments can be freely combined without conflict, and the combined embodiments are covered in the protection scope of the present application.

The above detailed description of embodiments of the application provided in the accompanying drawings is not intended to limit the scope of the application as claimed, but is merely representative of selected embodiments of the application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.

It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further definition or explanation thereof is necessary in the following figures.

In the description of the present application, it should also be noted that, unless explicitly specified and limited otherwise, the term "connected" should be construed broadly, and may be a fixed connection, a removable connection, or an integral connection, for example; may be an electrical connection; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present application will be understood in specific cases by those of ordinary skill in the art.

The above description is only an example of the present application and is not intended to limit the scope of the present application, and various modifications and variations will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for noise reduction of speech applied to an earphone, the earphone at least comprising a first microphone and a second microphone, wherein the direction in which the first microphone points to the second microphone or the direction in which the second microphone points to the first microphone is an end-fire direction, the method for noise reduction of speech comprising:

acquiring a target voice signal, wherein the target voice signal comprises a first voice signal acquired by the first microphone and a second voice signal acquired by the second microphone;

Calculating differential beam forming results of the first voice signal and the second voice signal in a plurality of null modes; the preset null directions of different null modes are different;

Taking the minimum amplitude value of the differential beam forming result corresponding to each null pattern as a target directivity gain result;

and denoising the target voice signal by using the target directivity gain result.

2. The method of claim 1, wherein the predetermined nulling directions for different nulling patterns include 90 ° directions, 120 ° directions, and 180 ° directions as compared to the end-fire direction.

3. The method of voice noise reduction according to claim 1, wherein said calculating the differential beamforming results of the first voice signal and the second voice signal in a plurality of null modes comprises:

Aiming at each null pattern, aligning an interference signal of the second voice signal in a preset null direction corresponding to the null pattern with an interference signal of the first voice signal in a preset null direction corresponding to the null pattern; the interference signal is a signal in a non-end-emission direction in the first voice signal and the second voice signal;

And calculating the difference between the first voice signal and the aligned second voice signal to obtain the differential beam forming result.

4. A method of speech noise reduction according to claim 3, further comprising: acquiring a frequency and steering vector matrix calculation formula of the second voice signal;

The aligning the interference signal of the second voice signal in the null direction corresponding to the preset null direction of the null mode with the interference signal of the first voice signal in the null direction corresponding to the preset null direction of the null mode includes:

Calculating a steering vector matrix corresponding to the null pattern based on a physical distance between the first microphone and the second microphone, a preset null direction corresponding to the null pattern, a frequency of the second voice signal and the steering vector matrix calculation formula;

Multiplying the second voice signal by the steering vector matrix to obtain a first product, wherein the first product represents that an interference signal of the second voice signal in a preset null direction corresponding to the null mode is aligned with an interference signal of the first voice signal in the preset null direction corresponding to the null mode.

5. The method of voice noise reduction according to claim 4, wherein the predetermined steering vector matrix calculation formula includes:

6. The method for voice noise reduction according to claim 4, wherein after taking a minimum amplitude value of the differential beam forming result corresponding to each null pattern as a target directivity gain result, the method further comprises:

multiplying the target directivity gain result by the phase of the differential beam forming result corresponding to the 180-degree direction null pattern to obtain a second product;

the second product is fed into an adaptive filter part of a generalized sidelobe canceling algorithm.

7. The method of voice noise reduction according to any of claims 3-6, wherein prior to said calculating differential beamforming results for said first voice signal and said second voice signal in a plurality of null modes, said method comprises:

For any null pattern, calculating a first time difference between the arrival of the signal of the target speech signal in the end-fire direction at the first microphone and the second microphone;

Calculating a second time difference matrix of arrival of the interference signal in the null direction corresponding to the null mode between the first microphone and the second microphone;

calculating to obtain a cut-off frequency based on the first time difference, the second time difference matrix and a preset cut-off frequency calculation formula;

Compensating signals lower than the cut-off frequency in the differential beam forming result corresponding to the null pattern to obtain a compensated differential beam forming result; the compensation is used to amplify the amplitude and phase of signals below the cutoff frequency in the differential beamforming result.

8. The method of voice noise reduction according to claim 7,

The first time difference is calculated by the following method:

the second time difference matrix is calculated as follows:

the preset cutoff frequency calculation formula comprises:

wherein d is a physical distance between the first microphone and the second microphone, θ2 is the preset null direction, θ1 is the end-fire direction, c is the sound propagation speed, τ1 is the first time difference, τ2 is the second time difference matrix, and ω _c is the preset cut-off frequency.

9. The method for voice noise reduction according to claim 7, wherein after compensating the signal below the cut-off frequency in the differential beam forming result corresponding to the null pattern, the method further comprises:

and limiting the compensated differential beam forming result so that the amplitude value of the limited differential beam forming result does not exceed the amplitude value of the first voice signal.

10. An earphone, comprising:

The microphone array is used for collecting target voice signals and at least comprises a first microphone and a second microphone, wherein the direction of the first microphone pointing to the second microphone or the direction of the second microphone pointing to the first microphone is an end-shooting direction; a differential beamformer; a filtering unit;

the processing unit is connected with the microphone array, the differential beam former and the filtering unit;

The processing unit is used for acquiring a target voice signal, and the voice signal comprises a first voice signal acquired by the first microphone and a second voice signal acquired by the second microphone;

The processing unit is further used for controlling the differential beam former to calculate differential beam forming results of the first voice signal and the second voice signal in a plurality of null modes; the preset null directions of different null modes are different;

The processing unit is further used for taking the minimum amplitude value of the differential beam forming result corresponding to each null pattern as a target directivity gain result;

the filtering unit is used for reducing noise of the target voice signal based on the target directivity gain result.