Audio signal processing method and apparatus, and electronic device and readable storage medium

Abstract

An audio signal processing method and apparatus, and an electronic device and a readable storage medium, which belong to the technical field of audio. The audio signal processing method comprises: performing time-frequency transformation processing on time-domain audio signals of M audio channels, so as to obtain M frequency-domain audio signals (S201); determining M wind noise signal power spectra on the basis of power spectra of the M frequency-domain audio signals (S202); determining M wind noise suppression gains on the basis of the M wind noise signal power spectra and a wind noise suppression factor (S203); and performing wind noise suppression processing on the M frequency-domain audio signals on the basis of the M wind noise suppression gains (S204).

Description

Audio signal processing method, device, electronic device and readable storage medium

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to the Chinese patent application filed with the China Patent Office on September 21, 2023, with application number 202311231263.9 and invention name “Audio signal processing method, device, electronic device and readable storage medium”, the entire contents of which are incorporated by reference into this application. Technical Field

[0003] The present application belongs to the field of audio technology, and specifically relates to an audio signal processing method, device, electronic device and readable storage medium. Background Art

[0004] Currently, when electronic devices such as communication devices and recording devices collect sound, they are inevitably affected by wind noise in the environment, reducing the clarity of the sound collected. Since the formation of wind noise is complex and difficult to predict, how to deal with wind noise is a difficult problem in this field.

[0005] Generally, the traditional methods of wind noise processing mainly include single-channel wind noise processing and multi-channel wind noise processing. In the related art, when performing multi-channel wind noise processing, the method usually adopted is to output a suppression gain based on the low similarity between multiple channels, and apply the suppression gain to the audio signals of multiple channels to suppress the wind noise of the audio signals of multiple channels. However, due to the low similarity between multiple channels, such as the obvious differences in wind noise intensity and bandwidth of multiple channels, using the same suppression gain to suppress wind noise of audio signals of multiple channels will result in over-suppression of wind noise signals in weak wind channels among the multiple channels, resulting in loss of voice signals in the audio signal, thereby resulting in poor suppression effect of wind noise signals in the audio signal.

[0006] Summary of the Invention

[0007] The purpose of the embodiments of the present application is to provide an audio signal processing method, device, electronic device and readable storage medium, which can improve the wind noise signal suppression effect of multi-channel audio signals.

[0008] In a first aspect, an embodiment of the present application provides an audio signal processing method, the method comprising: performing time-frequency transformation processing on time-domain audio signals of M audio channels to obtain M frequency-domain audio signals; M is an integer greater than 1; determining M wind noise signal power spectra based on the power spectra of the M frequency-domain audio signals; a wind noise signal power spectrum is used to characterize the distribution of the power of a wind noise signal in a frequency-domain audio signal on each frequency component; determining M wind noise suppression gains based on the M wind noise signal power spectra and a wind noise suppression factor; the wind noise suppression factor is used to characterize the similarity between the signal power spectra of the M frequency-domain audio signals; and performing wind noise suppression processing on the M frequency-domain audio signals based on the M wind noise suppression gains.

[0009] In second aspect, an embodiment of the present application provides an audio signal processing device, which includes: a processing module, which is used to: perform time-frequency transformation processing on the time domain audio signals of M audio channels to obtain M frequency domain audio signals; M is an integer greater than 1; based on the power spectra of the M frequency domain audio signals obtained by the processing module, determine M wind noise signal power spectra; a wind noise signal power spectrum is used to characterize the distribution of the power of the wind noise signal in a frequency domain audio signal on each frequency component; based on the M wind noise signal power spectra and the wind noise suppression factor, determine M wind noise suppression gains; the wind noise suppression factor is used to characterize the similarity between the signal power spectra of the M frequency domain audio signals; based on the M wind noise suppression gains determined by the determination module, perform wind noise suppression processing on the M frequency domain audio signals.

[0010] In a third aspect, an embodiment of the present application provides an electronic device comprising a processor and a memory, wherein the memory stores programs or instructions that can be run on the processor, and when the programs or instructions are executed by the processor, the steps of the method described in the first aspect are implemented.

[0011] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0012] In a fifth aspect, an embodiment of the present application provides a chip, which includes a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the method described in the first aspect.

[0013] In a sixth aspect, an embodiment of the present application provides a computer program product, which is stored in a storage medium and executed by at least one processor to implement the method described in the first aspect.

[0014] In an embodiment of the present application, an audio signal processing device performs time-frequency transformation processing on the time domain audio signals of M audio channels to obtain M frequency domain audio signals, where M is an integer greater than 1. Based on the power spectra of the above M frequency domain audio signals, M wind noise signal power spectra are determined. A wind noise signal power spectrum is used to characterize the distribution of the power of a wind noise signal in a frequency domain audio signal on frequency components. Based on the above M wind noise signal power spectra and a wind noise suppression factor, M wind noise suppression gains are determined. The wind noise suppression factor is used to characterize the similarity between the signal power spectra of the above M frequency domain audio signals. Based on the above M wind noise suppression gains, wind noise suppression processing is performed on the above M frequency domain audio signals. Through this method, the audio signal processing device can respectively determine the wind noise signal power spectrum of each audio channel, and obtain the wind noise suppression gain of each audio channel based on the wind noise signal power spectrum and wind noise suppression factor of each audio channel. In this way, the wind noise suppression gain of each audio channel can be determined in a targeted manner to avoid over-suppression or erroneous suppression of the wind noise signal of the audio channel, so that the wind noise in the multi-channel audio signal can be accurately eliminated, thereby improving the suppression effect of the wind noise signal of the multi-channel audio signal, and at the same time avoiding the destruction of the stationarity characteristics of other noises, providing the possibility for speech enhancement algorithms such as beamforming to further improve speech quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] FIG1(A) is a schematic diagram of a WNS module deployment method in the related art;

[0016] FIG1(B) is a schematic diagram of another WNS module deployment method in the related art;

[0017] FIG1(C) is a schematic diagram of another WNS module deployment method in the related art;

[0018] FIG2 is a schematic diagram of the power spectrum of an audio signal after wind noise suppression processing by the WNS module;

[0019] FIG3 is a flow chart of an audio signal processing method according to an embodiment of the present application;

[0020] FIG4 is a schematic diagram of an estimation result of a wind noise signal power spectrum provided in an embodiment of the present application;

[0021] FIG5 is a schematic diagram showing the relationship between the low-frequency MSC and the presence of a wind noise signal according to an embodiment of the present application;

[0022] FIG6 is a schematic diagram of a wind noise signal power spectrum provided in an embodiment of the present application;

[0023] FIG7 is a flowchart of audio signal processing according to an embodiment of the present application;

[0024] FIG8 is a schematic diagram of the “multi-input multi-output” wind noise suppression effect in a dual-channel input scenario provided by an embodiment of the present application;

[0025] FIG9 is a schematic structural diagram of an audio signal processing device provided in an embodiment of the present application;

[0026] FIG10 is a schematic structural diagram of an electronic device provided in an embodiment of the present application;

[0027] FIG11 is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The following will be combined with the accompanying drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the embodiments described are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of this application.

[0029] The terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of this application can be implemented in an order other than that illustrated or described herein, and that the objects distinguished by "first," "second," and the like are generally of the same type, and do not limit the number of objects; for example, the first object can be one or more. In addition, the term "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0030] The terms "at least one" and "at least one of" in the specification and claims of this application refer to any one, any two, or a combination of more than two of the objects included. For example, at least one of a, b, and c can be represented by: "a", "b", "c", "a and b", "a and c", "b and c", and "a, b, and c", where a, b, and c can be single or multiple. Similarly, "at least two" means two or more, and its meaning is similar to "at least one".

[0031] The rapidly flowing air in the environment creates turbulence around the pickup device, which strikes the microphone diaphragm, causing the audio signal picked up by the device to contain "wind noise." Each "gust" of wind noise typically lasts from tens of milliseconds to several seconds, with its energy primarily concentrated in the low-frequency region and decaying toward the high-frequency region. The energy value has a high dynamic range and significant non-stationarity. The energy spectrum (or power spectrum) of wind noise collected by multi-microphone devices also exhibits significantly low similarity across audio channels.

[0032] Wind noise is strong and fluctuates significantly. In voice call scenarios, it seriously weakens the listener's ability to understand voice information and worsens the listening experience. Therefore, measures need to be taken to reduce the wind noise content in the audio signal. Implementation schemes can be divided into two categories: methods that require additional equipment and signal processing methods. The former includes the use of windproof balls, windproof ducts, etc., but it is obvious that some terminals do not have the conditions for deployment, such as mobile phones or tablets; the latter suppresses wind noise signals through mathematical calculation methods and has a wider range of application scenarios. This type of signal processing method is called wind noise suppression (WNS) algorithm.

[0033] The WNS algorithm is an important part of speech enhancement technology. Broadly speaking, the WNS algorithm consists of two steps:

[0034] (1) Detection: Detecting the presence of wind noise mainly relies on the signal characteristics of wind noise. Depending on the number of signal channels used, these can be divided into 1) single-channel features, including sub-band centroid, zero-crossing rate, short-time mean, etc.; and 2) dual-channel or multi-channel features, mainly based on the low similarity between wind noise channels, such as magnitude-squared coherence (MSC).

[0035] (2) Suppression: After detecting the presence of wind noise, the wind noise is eliminated. Classic noise suppression algorithms, such as Wiener filtering, are only effective for suppressing stationary noise and have limited ability to suppress non-stationary noise such as wind noise. The current industry consensus is to suppress wind noise based on the low similarity of the wind noise spectrum between channels.

[0036] Related art WNS algorithm solutions for multi-channel input signals typically output a single channel (or perform sub-band channel replication on a single-channel signal to output a "fake" multi-channel signal), such as a "multi-input, single-output" approach. A specific implementation is summarized as follows: based on low inter-channel similarity, a single-channel suppression gain is output, which is applied to the (preset) primary channel of the input multi-channel audio signal. A limitation of this design (taking a dual-channel input as an example) is that, considering embedding the WNS module in a dual-input, single-output speech enhancement algorithm chain, such as a mobile phone call algorithm, the placement of the dual-input, single-output WNS module is severely constrained. Figures 1(A) and 1(B) illustrate two dual-input, single-output WNS module deployment methods in a dual-channel speech enhancement algorithm chain. Regardless of whether the wind noise detection and suppression stages are separated, the single-channel suppression stage can only appear after the channel is switched from dual to single. The consequence of this deployment is that when the channel is in the dual state, wind noise remains present in the audio signal, interfering with the operation of other dual-channel input speech enhancement modules. Wind noise has strong energy and has the greatest impact on voice quality among all types of noise. It should be removed as far forward as possible.

[0037] In related art, Figure 1(C) shows a schematic diagram of another "dual-input, dual-output" WNS module design. As shown in Figure 1(C), a single-channel suppression gain is applied to both channels, allowing the wind noise suppression module to be placed closer to the front of the dual-channel speech enhancement algorithm chain. The disadvantage is that the low similarity between channels represents the "union" of wind noise in both channels, making it impossible to distinguish the strength of the wind noise in the two channels. If the wind noise intensity and bandwidth of the two channels are significantly different, but the same suppression gain is used to suppress them separately, the speech signal in the weak wind channel may be incorrectly suppressed. For example, channel 1 has wind noise in the bandwidth range of [0Hz, 1kHz], and channel 2 has wind noise in the bandwidth range of [0Hz, 2kHz]. Then the "low similarity feature between channels" can only reflect that these two channels have wind noise in the bandwidth of [0Hz, 2kHz]. It is impossible to distinguish which channel the wind noise in the bandwidth of [1kHz, 2kHz] belongs to, and it is unknown which channel the suppression gain in the bandwidth of [1kHz, 2kHz] should act on. For another example, the wind noise energy intensity of channel 1 is 80, and the wind noise energy intensity of channel 2 is 100. Then the "single-channel suppression gain" mapped by the "low similarity feature between channels" is likely to be -100, which is only suitable for channel 2 with stronger wind noise (wind noise intensity: 100-100=0). For channel 1, wind noise is over-suppressed (80-100=-20).

[0038] Another disadvantage of the wind noise suppression method shown in Figure 1 is that it will interfere with the operation of other modules. For example, the audio signal x collected by the microphone contains wind noise. k (t) can be modeled as:

[0039] Where t is the continuous time, k indicates the channel number; s k (t) is the speech signal, including the effect of the propagation path; the noise part includes stationary noise and wind noise Ideally, in a speech enhancement algorithm chain, the WNS module should only remove the noise part. Completely preserved The latter is suppressed by other stationary noise suppression modules. When the wind noise intensity of the two channels is significantly different, the above wind noise suppression method is difficult to control the suppression strength of WNS, which often leads to The features that characterize the stationarity of noise are destroyed, which interferes with the ability of the stationary noise suppression module to identify stationary noise and weakens its noise suppression effect.

[0040] Figure 2 is a schematic diagram of the power spectrum of an audio signal after wind noise suppression processing using the WNS module in related art. As shown in Figure 2, from top to bottom, they are input signal channel 1, input signal channel 2, output signal channel 1, and output signal channel 2. Input signal channel 1 has a large amount of wind noise, while input signal channel 2 has a small amount of wind noise. After processing by the WNS module, the wind noise signals of output signal channels 2 and 3 are reduced, but the stationary noise is also suppressed to non-stationary noise, resulting in significant loss of speech signal.

[0041] It should be noted that the solid-line frame 301 contains wind noise signals that need to be suppressed, and the dotted-line frames 302 and 303 contain stationary noise signals and speech signals that have been suppressed or incorrectly suppressed.

[0042] An embodiment of the present application provides a method for calculating the wind noise suppression gain of each of multiple channels. Time-frequency transform processing is performed on the time-domain audio signals of M audio channels to obtain M frequency-domain audio signals, where M is an integer greater than 1. Based on the power spectra of the above-mentioned M frequency-domain audio signals, M wind noise signal power spectra are determined. A wind noise signal power spectrum is used to characterize the distribution of the power of a wind noise signal in a frequency-domain audio signal on frequency components. Based on the above-mentioned M wind noise signal power spectra and a wind noise suppression factor, M wind noise suppression gains are determined. The wind noise suppression factor is used to characterize the similarity between the signal power spectra of the above-mentioned M frequency-domain audio signals. Based on the above-mentioned M wind noise suppression gains, wind noise suppression processing is performed on the above-mentioned M frequency-domain audio signals. Through this method, the wind noise signal power spectrum of each audio channel is determined separately, and the wind noise suppression gain of each audio channel is obtained based on the wind noise signal power spectrum and wind noise suppression factor of each audio channel. In this way, the wind noise suppression gain of each audio channel can be determined in a targeted manner to avoid over-suppression or erroneous suppression of the wind noise signal of the audio channel, thereby accurately eliminating the wind noise in the multi-channel audio signal, effectively ensuring the stability of other noises in the suppressed signal, and facilitating the operation of other modules.

[0043] The audio signal processing method provided in the embodiments of the present application can be executed by an electronic device or at least one of the functional modules and physical modules within the electronic device that are capable of implementing the audio signal processing method. The specific execution can be determined based on actual usage requirements and is not limited by the embodiments of the present invention. The audio signal processing method provided in the embodiments of the present application is described below using an audio signal processing device executing the audio signal processing method as an example.

[0044] The audio signal processing method provided by the embodiment of the present application is described in detail below through specific embodiments and application scenarios in conjunction with the accompanying drawings.

[0045] FIG3 is a flow chart of an audio signal processing method provided in an embodiment of the present application. As shown in FIG3 , the audio signal processing method may include the following steps S201 to S204:

[0046] Step S201: The audio signal processing apparatus performs time-frequency transformation on the time-domain audio signals of M audio channels to obtain M frequency-domain audio signals.

[0047] Wherein, M is an integer greater than 1.

[0048] Optionally, in an embodiment of the present application, the above-mentioned audio channel is used to transmit the collected time domain audio signal.

[0049] It should be noted that an audio channel refers to a dynamic path along which a sound signal can be transmitted. It exists in the form of digital coding to achieve multi-channel sound mixing.

[0050] Optionally, in an embodiment of the present application, the time-domain audio signal may be an audio signal collected by a microphone.

[0051] Exemplarily, the M audio channels may include at least two audio channels. Further, one audio channel is used to transmit a time domain audio signal collected by one microphone.

[0052] Optionally, in an embodiment of the present application, the frequency domain audio signal is an audio signal obtained by converting the collected original audio signal into the frequency domain.

[0053] Optionally, in an embodiment of the present application, the audio signal processing device may perform short-time Fourier transform (STFT) on the time-domain audio signals of the M audio channels to convert the time-domain audio signals of the M audio channels into frequency-domain audio signals.

[0054] For the convenience of description, the time-domain audio signals of the above-mentioned M audio channels may be collectively referred to as input signals.

[0055] For example, take M audio channels including two audio channels as an example. The time-domain audio signals collected by two microphones are represented as x1(t) and x2(t), respectively. After x1(t) is subjected to STFT, the frequency-domain audio signal X1 is obtained. The spectrum of the frequency-domain audio signal X1 is represented as X1[λ,μ]. After x2(t) is subjected to STFT, the frequency-domain audio signal X2 is obtained. The spectrum of the frequency-domain audio signal X2 is represented as X2[λ,μ]. Where λ represents the time frame number and μ represents the frequency point number.

[0056] Step S202: The audio signal processing apparatus determines M wind noise signal power spectra based on the power spectra of the M frequency-domain audio signals.

[0057] A wind noise signal power spectrum is used to characterize the power distribution of a wind noise signal in each frequency component in a frequency domain audio signal.

[0058] Optionally, in the embodiment of the present application, the audio signal apparatus may obtain M frequency-domain audio signal spectra, and obtain power spectra of the M frequency-domain audio signals according to the spectra of the M frequency-domain audio signals.

[0059] For example, in combination with the above example, after the time domain audio signal x1(t) is subjected to STFT to obtain the frequency domain audio signal X1, the spectrum of the frequency domain audio signal X1 is expressed as X1[λ,μ], and the spectrum of the frequency domain audio signal X1 is modulo and squared to obtain the power spectrum of the time domain audio signal x1(t). The power spectrum of the time domain audio signal x1(t) can be expressed as Φ 11 [λ,μ]; Similarly, the power spectrum of the time domain audio signal x2(t) can be calculated. The power spectrum of the time domain audio signal x2(t) can be expressed as Φ 22 [λ,μ].

[0060] It should be noted that the above Φ 11 [λ,μ] and Φ 22 [λ,μ] are the autopower spectra of the time-domain audio signal x1(t) and the time-domain audio signal x2(t), respectively. The cross-power spectrum of the time-domain audio signal x1(t) and the time-domain audio signal x2(t) can be expressed as Φ 12 [λ,μ].

[0061] Optionally, in this embodiment of the present application, one frequency domain audio signal corresponds to one wind noise signal power spectrum.

[0062] Optionally, in the embodiment of the present application, the audio signal processing apparatus may determine a wind noise signal power spectrum corresponding to a frequency domain audio signal according to the presence of wind noise and speech in the frequency domain audio signal.

[0063] Illustratively, when a frequency domain audio signal includes a wind noise signal and a speech signal, the audio signal processing apparatus may obtain a wind noise signal power spectrum corresponding to the frequency domain audio signal based on harmonic information of the frequency domain audio signal.

[0064] Exemplarily, in a case where a frequency-domain audio signal only includes a wind noise signal, the audio signal processing apparatus may use a power spectrum of the frequency-domain audio signal as a corresponding power spectrum of the wind noise signal.

[0065] Exemplarily, when a frequency-domain audio signal does not include a wind noise signal, the audio signal processing apparatus may set a wind noise signal power spectrum corresponding to the frequency-domain audio signal to a preset minimum value.

[0066] It should be noted that the explanation of the process of determining the M wind noise signal power spectra based on the power spectra of the M frequency-domain audio signals can be found in the following description and will not be repeated here.

[0067] Step S203: The audio signal processing apparatus determines M wind noise suppression gains according to the M wind noise signal power spectra and wind noise suppression factors.

[0068] The wind noise suppression factor is used to characterize the similarity between the signal power spectra of the M frequency-domain audio signals.

[0069] Optionally, in the embodiment of the present application, the audio signal processing apparatus may determine a wind noise suppression factor based on power spectra of M frequency-domain audio signals.

[0070] It should be noted that the explanation of the process of determining the wind noise suppression factor can be found below and will not be repeated here.

[0071] Optionally, in an embodiment of the present application, the audio signal processing apparatus may obtain M suppression floor factors according to M wind noise signal power spectra, and one suppression floor factor is used to suppress a wind noise signal of a time-domain audio signal.

[0072] Optionally, in the embodiment of the present application, the audio signal processing device fuses each suppression base factor with the wind noise suppression factor to obtain a wind noise suppression gain.

[0073] For example, taking the example that M's audio channels include two audio channels, audio channel 1 and audio channel 2, the frequency domain audio signal of audio channel 1 is X1, the frequency domain audio signal of audio channel 2 is X2, the suppression basis factor corresponding to the frequency domain audio signal X1 is B1, the suppression basis factor corresponding to the frequency domain audio signal X2 is B2, and the wind noise suppression factor is SGAR. By fusing the suppression basis factor and the wind noise similarity factor of the frequency domain audio signal X1, we can obtain the wind noise suppression gain of the frequency domain audio signal X1, expressed as G1, and obtain the wind noise suppression gain of the frequency domain audio signal X2, expressed as G2.

[0074] Step S204: the audio signal processing apparatus performs wind noise suppression processing on the M frequency-domain audio signals according to the M wind noise suppression gains.

[0075] Optionally, in the embodiment of the present application, the audio signal processing device may apply M wind noise suppression gains to the power spectra of M frequency-domain audio signals to obtain M processed frequency-domain audio signals.

[0076] For example, in combination with the above example, taking the example that the audio channel of M includes two audio channels, audio channel 1 and audio channel 2, the frequency domain audio signal of audio channel 1 is X1, the frequency domain audio signal of audio channel 2 is X2, and the power spectrum of the frequency domain audio signal X1 is Φ 11 , the wind noise suppression gain of the frequency domain audio signal X1 is G1, and the power spectrum of the frequency domain audio signal x2(t) is Φ 22 , the wind noise suppression gain of the frequency domain audio signal x2(t) is G2, and the power spectrum of the frequency domain audio signal X1 is Φ 11 Multiplying the corresponding wind noise suppression gain G1, the frequency domain audio signal after the wind noise suppression of the time domain audio signal of audio channel 1 is obtained, and the power spectrum of the frequency domain audio signal X2 is Φ 22 Multiplying the corresponding wind noise suppression gain G2 obtains the frequency domain audio signal after wind noise suppression is performed on the time domain audio signal of audio channel 2.

[0077] Optionally, in an embodiment of the present application, the audio signal processing apparatus performs time-frequency conversion on the M processed frequency-domain audio signal pairs to obtain M time-domain audio signals.

[0078] In the audio signal processing method provided in the embodiment of the present application, the audio signal processing device can respectively determine the wind noise signal power spectrum of each audio channel, and obtain the wind noise suppression gain of each audio channel based on the wind noise signal power spectrum and wind noise suppression factor of each audio channel. In this way, the wind noise suppression gain of each audio channel can be determined in a targeted manner to avoid over-suppression or erroneous suppression of the wind noise signal of the audio channel, thereby accurately eliminating the wind noise in the multi-channel audio signal, thereby improving the suppression effect of the wind noise signal of the multi-channel audio signal. Optionally, in the embodiment of the present application, the process of the above-mentioned step S202 can include the following steps S202a1 to S202a2, or the process of the above-mentioned step S202 can include the following step S202b, or the process of the above-mentioned step S202 can include the following step S202c:

[0079] Step S202a1: the audio signal processing apparatus obtains a harmonic frequency point set of a second frequency domain audio signal among the M frequency domain audio signals when the first frequency domain audio signal contains a wind noise signal and a speech signal.

[0080] Step S202a2: The audio signal processing device determines the signal power of the first frequency point of the first frequency domain audio signal as the first wind noise signal power corresponding to the first frequency point, and interpolates the signal power of the second frequency point of the frequency domain audio signal based on the first wind noise signal power to obtain the second wind noise signal power corresponding to the second frequency point.

[0081] Step S202a3: The audio signal processing device determines the wind noise signal power spectrum corresponding to the first frequency domain audio signal based on the first wind noise signal power and the second wind noise signal power; wherein the first frequency point includes at least one frequency point that does not belong to the harmonic frequency point set; and the second frequency point includes at least one frequency point that belongs to the harmonic frequency point set.

[0082] Step S202b: When there is a wind noise signal in the first frequency domain audio signal and no speech signal, the audio signal processing apparatus uses the power spectrum of the first frequency domain audio signal as the wind noise signal power spectrum corresponding to any frequency domain audio signal.

[0083] Step S202c: when there is no wind noise signal in the first frequency-domain audio signal, the audio signal processing apparatus uses a preset wind noise signal power spectrum as the wind noise signal power spectrum corresponding to the first frequency-domain audio signal.

[0084] Optionally, in an embodiment of the present application, the above-mentioned first frequency domain audio signal is any frequency domain audio signal among the above-mentioned M frequency domain audio signals, that is, the audio signal processing device can perform processing steps such as the first frequency domain audio signal on each frequency domain audio signal among the M frequency domain audio signals.

[0085] Optionally, in an embodiment of the present application, for a first frequency domain audio signal among M frequency domain audio signals, the presence of a wind noise signal and a voice signal in the first frequency domain audio signal can be first detected, and then, based on the presence of the wind noise signal and the voice signal, a method for determining the wind noise signal power spectrum corresponding to the first frequency domain audio signal can be determined.

[0086] The following embodiment exemplarily illustrates a process of determining a wind noise signal power spectrum corresponding to a first frequency-domain audio signal.

[0087] In some possible implementations, the first frequency-domain audio signal includes a wind noise signal and a speech signal. A harmonic frequency point set of the first frequency-domain audio signal is obtained by a harmonic estimation module and expressed as Ψ[λ], using the following formula:

[0088] Among them, μ FF Corresponding to the fundamental frequency (FF) of speech, μ Δ is a harmonic frequency range. When μ does not belong to the set Ψ[λ], the estimated wind noise spectrum frequency power is Φ kk [λ,μ]; conversely, interpolation is used to fill the spectrum at the position of speech harmonics. The calculation formula of the wind noise signal power spectrum is as follows:

[0089] Where λ represents the time frame number and μ represents the frequency point number.

[0090] For example, the first frequency point is a frequency point that does not belong to the set Ψ[λ], and the wind noise signal power corresponding to the first frequency point may be the signal power of the first frequency point, expressed as Φ kk [λ,μ]; the second frequency point mentioned above is the frequency point belonging to the set Ψ[λ], and the wind noise signal power corresponding to the second frequency point is obtained by interpolating the signal power of the second frequency point, which is expressed as

[0091] Exemplarily, after determining the wind noise signal powers corresponding to the first frequency point and the second frequency point of the first frequency domain audio signal, such as determining the first wind noise signal power and the second wind noise signal power, the wind noise signal power spectrum corresponding to the first frequency domain audio signal is obtained based on the first wind noise signal power and the second wind noise signal power.

[0092] Exemplarily, Figure 4 is the estimation result of the wind noise signal power spectrum provided in an embodiment of the present application. As shown in Figure 4, when both the wind noise signal and the speech signal exist in the first frequency domain audio signal, the spectrum of the speech harmonic position is filled in by interpolation. For example, the high value near 0.5kHz in the frequency domain audio signal power spectrum corresponds to a speech harmonic.

[0093] In some other possible implementations, when the first frequency domain audio signal contains a wind noise signal but does not contain a speech signal, the power spectrum of the first frequency domain audio signal is used as the corresponding wind noise signal power spectrum. The wind noise signal power spectrum is calculated as follows:

[0094] In some other possible implementations, the first frequency-domain audio signal includes only a speech signal but not a wind noise signal, and a preset wind noise signal power spectrum is used as the wind noise signal power spectrum corresponding to the first frequency-domain audio signal.

[0095] Exemplarily, the above-mentioned preset wind noise signal power spectrum may be a minimum value. Exemplarily, the estimated wind noise signal power spectrum is set to a minimum value, and the wind noise signal power spectrum calculation formula is as follows:

[0096] Here, ε is the minimum value, and 1 represents an all-1 vector with the same number of elements as the number of frequency points in a frame of spectrum.

[0097] In this way, the audio signal processing device determines the wind noise signal power spectrum corresponding to any frequency-domain audio signal according to the presence of the wind noise signal and the audio signal in the frequency-domain audio signal, thereby being able to accurately estimate the wind noise signal power spectrum.

[0098] Optionally, in the embodiment of the present application, before the above step S202, the audio signal processing method provided in the embodiment of the present application may further include the following steps A1 and A2:

[0099] Step A1: The audio signal processing device calculates wind noise presence flag information based on the amplitude square coherence MSC information of the first frequency domain audio signal and the proportion of high-energy low-MSC frequency points of the first frequency domain audio signal within a preset frequency range.

[0100] The high-energy-to-low-MSC proportion (HELMP) is used to characterize the energy attenuation characteristics of the wind noise signal.

[0101] Step A2: The audio signal processing device determines whether a wind noise signal exists in the first frequency domain audio signal according to the wind noise presence flag information.

[0102] Optionally, in the embodiment of the present application, the above-mentioned Magnitude-Squared Coherence (MSC) information is used to characterize the similarity between audio signals in multiple audio channels.

[0103] Optionally, in the embodiment of the present application, the audio signal processing device may first calculate the low frequency region MSC mean of the M frequency domain audio signals. To determine whether the wind noise signal exists in a certain frequency domain audio signal, and calculate the high energy-low MSC frequency point ratio ρ of each frequency domain audio signal k [λ], then the low-frequency region MSC mean The ratio of high energy-low MSC frequency points of each frequency domain audio signal is ρ k [λ] are fused, and finally the fused information is mapped to the wind noise flag information corresponding to the frequency domain audio signal of each audio channel.

[0104] Optionally, in an embodiment of the present application, when the wind noise existence flag information corresponding to the first frequency domain audio signal is 1, it is determined that a wind noise signal exists in the first frequency domain audio signal; when the wind noise existence flag information is 0, it is determined that no wind noise signal exists in the first frequency domain audio signal; or, when the wind noise existence flag information corresponding to the first frequency domain audio signal is within a preset numerical range, it is determined that a wind noise signal exists in the first frequency domain audio signal; when the wind noise existence flag information is not within the above-mentioned preset numerical range, it is determined that no wind noise signal exists in the first frequency domain audio signal.

[0105] The following explains the process of calculating the wind noise presence flag information by taking an example where the M audio channels include two audio channels and the M time-domain audio signals include x1(t) and x2(t).

[0106] For example, two microphones collect time-domain audio signals x1(t) and x2(t), and after STFT, the spectra X1[λ,μ] and X2[λ,μ] are obtained, where λ represents the time frame number and μ represents the frequency point number. The autopower spectrum Φ of X1[λ,μ] and X2[λ,μ] is 11 [λ,μ],Φ 22 [λ,μ], cross power spectrum Φ 12 The calculation formulas for [λ,μ] are:

[0107] in,(·) * represents the conjugate operation. After smoothing, it is expressed as:

[0108] Where α∈(0,1) is the smoothing coefficient. The MSC calculation formula is:

[0109] It should be noted that wind noise signals are primarily concentrated in the low-frequency region. If wind noise signals are present in at least one channel, the MSC in the low-frequency region will be low. For example, "the presence of wind noise" is a sufficient condition for "MSC in the low-frequency region." However, experiments have shown that "MSC in the low-frequency region" is not a sufficient condition for "the presence of wind noise," as shown in Figure 5.

[0110] For example, it is preliminarily determined whether wind noise may exist in a certain channel. Define the MSC mean of the low-frequency region of the frame-level feature The relationship indicating the presence of wind noise in both channels, The calculation formula is as follows:

[0111] Among them, β∈(0,1) is the smoothing coefficient, μ0 and μ1 respectively constrain the lower and upper bounds of the frequency range of the low frequency region. is a high value (eg, greater than the threshold Γ thre ), there must be no wind noise in the λth frame of both channels; otherwise, wind noise may exist in one channel or not, and further confirmation is required based on other wind noise characteristics.

[0112] For example, taking M audio channels including channel 1 and channel 2 as an example, the relationship between the presence of wind noise in the two channels can be: if there is wind noise in channel 1 or channel 2, then Output low value, but It cannot indicate which channel the wind noise is in. For example, “channel 1 has wind noise, channel 2 has no wind noise” or “channel 1 has no wind noise, channel 2 has wind noise” can be used. Outputs a low value. Therefore, the embodiment of the present application integrates MSC with the proportion of high-energy low-MSC frequency points to accurately determine whether there is a wind noise signal in any channel.

[0113] Through experiments, we found that the limitation of MSC in wind noise detection is that it has no energy dimension, and some other low-energy environmental noise may also cause The energy characteristics of wind noise can be used to distinguish other noises, such as attenuation characteristics, which can be modeled in two levels: within a frame of spectrum, the frequency points with the highest energy appear below 100 Hz and have high peak energy; within a limited frequency range, the frequency point energy attenuation is limited. Figure 6 shows the wind noise power spectrum for a frame, with the horizontal axis representing frequency and the vertical axis representing frequency point energy intensity.

[0114] It should be noted that the horizontal axis can be expressed in logarithmic form.

[0115] Further optionally, in the embodiment of the present application, before the above step A1, the audio signal processing method provided in the embodiment of the present application may further include the following steps A3 and A4:

[0116] Step A3: The audio signal processing apparatus calculates MSC information and signal power of each frequency point of the first frequency-domain audio signal within the first frequency range.

[0117] Step A4: The audio signal processing apparatus calculates the proportion of high-energy low-MSC frequency points of the first frequency-domain audio signal within the first frequency range according to the MSC information and signal power corresponding to each frequency point.

[0118] Optionally, in an embodiment of the present application, the first frequency range may be a preset frequency range.

[0119] For example, the High-Energy-to-Low-MSC Proportion (HELMP) is defined as ρ k [λ], characterizes the energy attenuation characteristics of wind noise, and is calculated as follows:

[0120] Wherein, μ′0 and μ′1 correspond to the aforementioned “limited frequency range”, such as the first frequency range; I{·} represents an indicator function, which is mapped to 1 when the condition in the brackets is met, and is otherwise mapped to 0.

[0121] For example, the energy judgment threshold of channel 1 is defined as and the energy judgment threshold of channel 2 The calculation formula is as follows:

[0122] It should be noted that the fundamental frequency of speech is generally higher than 100Hz, so the frequency points with the maximum energy value below 100Hz are usually not confused with the speech frequency points. Wind noise starts to decay from the highest value to the high frequency, but the decay rate is limited and will not decay by ξ times before the μ′1 frequency point. Therefore, if Φ kk The energy of [λ,μ] in the range [μ′0,μ′1] is lower than The frequency points are large and HELMP is small. If the energy decays too fast with the increase of frequency, it means that Φ kk If [λ,μ] does not meet the wind noise energy attenuation characteristics, then there is no wind noise in channel k; on the contrary, if HELMP is large, it means that the channel meets the wind noise characteristics. For some other environmental noises with smaller energy, p t Raising the threshold This causes HELMP to show a low value, indicating that no wind noise is present.

[0123] Finally, The information is fused with ρ1[λ] and ρ2[λ] respectively, and the calculation formula is as follows:

[0124] Smoothing the above formula yields Finally, it is mapped to the wind noise flag of each channel. The calculation formula is as follows:

[0125] For example, WN flag [k,λ]=1 corresponds to a “wind noise present” state, whereas [k,λ]=1 corresponds to a “wind noise absent” state.

[0126] For example, if the fused result is greater than the threshold, the wind noise flag WN is set. flag Mapping to 1 corresponds to the state of wind noise; otherwise, WN flag Mapping to 0 corresponds to no wind noise.

[0127] Optionally, in the embodiment of the present application, before the above step S202, the audio signal processing method provided in the embodiment of the present application may further include the following step B1:

[0128] Step B1: The audio signal processing apparatus detects whether a speech signal exists in a first frequency-domain audio signal among the M frequency-domain audio signals.

[0129] Optionally, in the embodiment of the present application, the audio signal processing apparatus may detect whether a voice signal exists in the first audio signal by using a Voice Activity Detection (VAD) algorithm or an Average Magnitude Difference Function (AMDF) algorithm.

[0130] It should be noted that the explanation of the VAD algorithm and the AMDF algorithm can be found in the relevant technology, and this application will not elaborate on this.

[0131] Exemplarily, the audio signal processing device can use a VAD algorithm to divide the first frequency domain audio signal into frequency bands and calculate the energy of each frequency band. Then, based on the energy of each frequency band and the energy of the noise signal estimated in each frequency band, it calculates the signal-to-noise ratio of each frequency band. Finally, it calculates the VAD decision threshold and determines whether there is a speech signal in the first frequency domain audio signal based on the VAD decision threshold.

[0132] It should be noted that the process of detecting whether a signal contains a speech signal through the VAD algorithm and the AMDF algorithm can be referred to the relevant technology, and the embodiments of the present application will not elaborate on this.

[0133] Optionally, in the embodiment of the present application, the audio signal processing apparatus may also use the characteristics of the speech signal itself to determine whether a speech signal exists in the first frequency domain audio signal.

[0134] Exemplarily, the audio signal detection device may obtain the harmonic feature of the first frequency domain audio signal and determine whether the harmonic feature meets the harmonic feature of the speech signal; if so, it is determined that the first frequency domain audio signal includes a speech signal.

[0135] It should be noted that the process of detecting whether a signal includes speech by using the harmonic characteristics of the signal can be referred to in related technologies and will not be described in detail here.

[0136] Optionally, in the embodiment of the present application, before the above step S203, the audio signal processing method provided in the embodiment of the present application may further include the following step C1:

[0137] Step C1: The audio signal processing device calculates the square ratio of the geometric mean to the arithmetic mean of the power spectra of the M frequency-domain audio signals to obtain a wind noise suppression factor.

[0138] Optionally, in the embodiment of the present application, a wind noise suppression factor is introduced based on the low similarity between the M audio channels.

[0139] Optionally, in an embodiment of the present application, the audio signal processing device may calculate the wind noise suppression factor according to the power spectra of M frequency-domain audio signals by using a Squared Geometric-Mean-to-Arithmetic-Mean Ratio (SGAR) algorithm.

[0140] For example, in combination with the above embodiment, taking M audio signals including channel 1 and channel 2 as an example, the power spectrum of the frequency domain audio signal of channel 1 and the power spectrum of the frequency domain audio signal of channel 2 are Φ 11 [λ,μ] and Φ 22 [λ,μ], the calculation formula of the wind noise suppression factor SGAR corresponding to channel 1 and channel 2 is as follows:

[0141] For example, for the frequency point μ with wind noise signal, Φ 11 [λ,μ] and Φ 22 If the difference between [λ,μ] is large, the value of SGAR[λ,μ] is small. With the help of SGAR[λ,μ], the speech frequency points in the audio signal can be effectively protected.

[0142] Exemplarily, in combination with the above embodiments, Figure 7 is a flow chart of the audio signal processing provided in an embodiment of the present application. As shown in Figure 7, the audio signal processing device obtains a multi-channel input signal, calculates the wind noise flag information of the input signal, calculates the wind noise suppression factor based on the wind noise flag information, and estimates the wind noise power spectrum. Then, based on the wind noise suppression factor and the wind noise power spectrum, the wind noise suppression gain is calculated. Finally, the input signal is subjected to wind noise suppression processing according to the wind noise suppression gain to obtain a processed output signal.

[0143] Optionally, in the embodiment of the present application, the process of step S204 may include the following steps S204a and S204b:

[0144] Step S204a: The audio signal processing apparatus calculates the suppression floor factor of each frequency-domain audio signal according to the M wind noise signal power spectra.

[0145] Step S204b: The audio signal processing apparatus calculates the wind noise suppression gain of each frequency domain audio signal according to the suppression floor factor and the wind noise suppression factor of each frequency domain audio signal to obtain M wind noise suppression gains.

[0146] Optionally, in the embodiment of the present application, the audio signal processing apparatus may calculate the suppression floor factor of each frequency-domain audio signal according to the frequency point energy of the wind noise signal power spectrum corresponding to the M frequency-domain audio signals.

[0147] It should be noted that for wind noise suppression, the ideal result is to suppress only the wind noise part of the input signal. However, when wind noise exists, the stationary noise It is very difficult to estimate and quantify the ideal expected result of wind noise suppression. If the MSC and other information are directly applied to the input signal, over-suppression will usually occur, and the residual wind noise will be less, but it will affect the smooth noise suppression effect of other modules. Therefore, it is necessary to design a reliable method for calculating the wind noise suppression gain that does not destroy the smooth noise. Due to the large difference in wind noise energy between channels, within the same time frame and the same frequency point, there will always be one channel in the dual channels with smaller wind noise energy. Taking advantage of this feature, the embodiment of the present application uses the energy level of the opposite channel as the basis for wind noise suppression in this channel, and lowers the energy level of the strong wind frequency point of the strong wind channel to the corresponding weak wind channel frequency point level.

[0148] Optionally, the above step S204a can be implemented by the following steps S204a1:

[0149] Step S204a1: The audio signal processing apparatus calculates a suppression floor factor of the second frequency-domain audio signal among the M frequency-domain audio signals based on the frequency-point energy of the wind noise signal power spectrum of the third frequency-domain audio signal and the frequency-point energy of the wind noise signal power spectrum of the M frequency-domain audio signals.

[0150] The second frequency domain audio signal is any one of the M frequency domain audio signals; and the third frequency domain audio signal is a frequency domain audio signal different from the second frequency domain audio signal among the M frequency domain audio signals.

[0151] Exemplarily, take M frequency-domain audio signals including a frequency-domain audio signal 1 of channel A and a frequency-domain audio signal 2 of channel B as an example. When calculating the suppression floor factor of the frequency-domain audio signal 1, the audio signal processing device calculates the sum of the frequency energy of the frequency-domain audio signal 1 and the frequency energy of the frequency-domain audio signal 2, and uses the ratio of the frequency energy of the audio signal 2 to the sum of the frequency energy of the audio signal 1 and the frequency energy of the frequency-domain audio signal 2 as the suppression floor factor of the frequency-domain audio signal 1. Similarly, when calculating the suppression floor factor of the frequency-domain audio signal 2, the audio signal processing device calculates the sum of the frequency energy of the frequency-domain audio signal 1 and the frequency energy of the frequency-domain audio signal 2, and uses the ratio of the frequency energy of the audio signal 1 to the sum of the frequency energy of the audio signal 1 and the frequency energy of the frequency-domain audio signal 2 as the suppression floor factor of the frequency-domain audio signal 2.

[0152] For example, in combination with the above embodiment, taking M audio channels including channel A and channel B as an example, the wind noise signal power spectrum corresponding to the frequency domain audio signal 1 of channel A and the wind noise signal power spectrum corresponding to the frequency domain audio signal 2 of channel B are respectively and Define the energy ratio of the wind noise signal power spectrum (Wind Noise Spectrum Energy Ratio, WNSER) B k As an inhibitory factor for channel A and channel B, B k The calculation formula is as follows:

[0153] It should be noted that B k represents the wind noise suppression factor, B1[λ,μ] represents the suppression floor factor of frequency domain audio signal 1, and B2[λ,μ] represents the suppression floor factor of frequency domain audio signal 2.

[0154] For example, assuming that there is no speech signal in channel A, the power spectrum intensity of the wind noise signal in channel A is Power spectrum intensity of wind noise signal in channel B According to the above formulas 21 and 22, we can calculate Furthermore, the above B1[λ,μ] and B2[λ,μ] can be used as preliminary wind noise suppression gains to apply the wind noise suppression base factor to the power spectrum Φ of the input frequency domain audio signal. 11 [λ,μ] and Φ 22 [λ,μ], and the wind noise signal strength of the frequency domain audio signal after wind noise suppression is obtained. The calculation formula is as follows: B1[λ,μ]×Φ 11 [λ,μ]=0.09×1=0.09 B2[λ,μ]×Φ 22 [λ,μ]=0.91×0.1=0.09

[0155] It can be seen that the wind noise signal strength of channel A and channel B is suppressed to 0.09.

[0156] It should be noted that when the wind noise frequency energy of channel A is stronger, The smaller the value, the smaller the B1[λ,μ] will be. For example, the suppression strength of this frequency point in channel A will be greater. By constraining the energy proportions B1[λ,μ] and B2[λ,μ], this method simply and effectively controls the amount of wind noise suppression, preventing over-suppression of the wind noise signal in the audio signal.

[0157] Optionally, in the embodiment of the present application, the audio signal processing device may calculate the wind noise suppression gain of each frequency domain audio signal according to the suppression floor factor and the above-mentioned wind noise suppression factor.

[0158] For example, in combination with the above embodiment, the suppression base factor and wind noise suppression factor corresponding to each channel in channel A and channel B are fused to obtain the wind noise suppression gain G1 of channel A and the wind noise suppression gain G2 of channel B. The calculation formulas are as follows: G1[λ,μ]=g(B1[λ,μ],SGAR[λ,μ]), (23) G2[λ,μ]=g(B2[λ,μ],SGAR[λ,μ]). (24)

[0159] It should be noted that the suppression floor factor corresponding to channel A is such as the suppression floor factor of frequency domain audio signal 1, and the suppression floor factor corresponding to channel B is such as the suppression floor factor of frequency domain audio signal 2.

[0160] In this way, by fusing the suppression base factor and the wind noise suppression factor, the wind noise suppression gain of each channel is obtained, so that the voice signal and non-wind noise frequency points in the frequency domain audio signal can be effectively protected when suppressing the wind noise signal.

[0161] Figure 8 is an example of the "multi-input and multi-output" wind noise suppression effect in a dual-channel input scenario provided by an embodiment of the present application. As shown in Figure 8, the input signal sampling frequency is 16kHz, the STFT length is 512 points, the STFT frame shift is 160 points, and from top to bottom are input signal channel 1, input signal channel 2, output signal channel 1, and output signal channel 2. Compared with the signals in the dotted box 302 and the dotted box 303 in Figure 2 corresponding to the wind noise suppression method of the related art mentioned above, the signals in the dotted box 302a and the solid box 303a retain the stationary noise and the voice signal (the white horizontal wavy lines in the figure). For example, only the wind noise signal is eliminated in the output signal channel, while the voice signal and other stationary noise are retained. It can be seen that the audio signal processing method provided by the embodiment of the present application only eliminates the wind noise signal in the input signal. Compared with the wind noise suppression method in the related art, the voice loss is significantly reduced, and the stationarity of other noises is effectively protected, ensuring that other noise suppression modules in the voice enhancement link can correctly identify and suppress stationary noise.

[0162] The above-mentioned method embodiments, or various possible implementation methods in each method embodiment, can be executed separately, or, under the premise that there is no contradiction, can also be executed in combination with each other. The specific implementation can be determined according to actual usage requirements, and the embodiments of the present application do not limit this.

[0163] The audio signal processing method provided in the embodiment of the present application can be executed by an audio signal processing device. In the embodiment of the present application, the audio signal processing device provided in the embodiment of the present application is described by taking the audio signal processing device executing the audio signal processing method as an example.

[0164] The audio signal processing device in the embodiment of the present application can be an electronic device or a component in the electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices other than a terminal. For example, the electronic device can be a mobile phone, a tablet computer, a laptop computer, a PDA, an in-vehicle electronic device, a mobile Internet device (MID), an augmented reality (AR) / virtual reality (VR) device, a robot, a wearable device, an ultra-mobile personal computer (UMPC), a netbook or a personal digital assistant (PDA), etc. It can also be a server, a network attached storage (NAS), a personal computer (PC), a television (TV), a teller machine or a self-service machine, etc., and the embodiment of the present application does not specifically limit it.

[0165] The audio signal processing device in the embodiment of the present application may be a device having an operating system. The operating system may be an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0166] The audio signal processing device provided in the embodiment of the present application can implement each process implemented in the method embodiments of Figures 1 to 8. To avoid repetition, they are not described here.

[0167] Figure 9 is a structural schematic diagram of an audio signal processing device provided in an embodiment of the present application. As shown in Figure 9, the audio signal processing device may include a processing module 801; the above-mentioned processing module 801 is used to: perform time-frequency transformation processing on the time domain audio signals of M audio channels to obtain M frequency domain audio signals; M is an integer greater than 1; based on the power spectra of the M frequency domain audio signals obtained by the above-mentioned processing module, M wind noise signal power spectra are determined; a wind noise signal power spectrum is used to characterize the distribution of the power of the wind noise signal in each frequency component of a frequency domain audio signal; based on the above-mentioned M wind noise signal power spectra and wind noise suppression factors, M wind noise suppression gains are determined; the above-mentioned wind noise suppression factor is used to characterize the similarity between the signal power spectra of the above-mentioned M frequency domain audio signals; based on the M wind noise suppression gains determined by the above-mentioned determination module, wind noise suppression processing is performed on the above-mentioned M frequency domain audio signals.

[0168] Optionally, in an embodiment of the present application, the processing module is specifically used to: for a first frequency domain audio signal among the M frequency domain audio signals, when a wind noise signal and a voice signal exist in the first frequency domain audio signal, obtain a harmonic frequency point set of the first frequency domain audio signal; determine the signal power of the first frequency point of the first frequency domain audio signal as the first wind noise signal power corresponding to the first frequency point, and based on the first wind noise signal power, interpolate the signal power of the second frequency point of the frequency domain audio signal to obtain a second wind noise signal power corresponding to the second frequency point; determine the wind noise signal power spectrum corresponding to the second frequency domain audio signal according to the first wind noise signal power and the second wind noise signal power; wherein the first frequency point includes at least one frequency point that does not belong to the harmonic frequency point set; and the second frequency point includes at least one frequency point that belongs to the harmonic frequency point set;

[0169] Alternatively, when a wind noise signal exists in the first frequency domain audio signal and no speech signal exists, the power spectrum of the first frequency domain audio signal is used as the power spectrum of the wind noise signal corresponding to the first frequency domain audio signal;

[0170] Alternatively, when there is no wind noise signal in the first frequency domain audio signal, a preset wind noise signal power spectrum is used as the wind noise signal power spectrum corresponding to the first frequency domain audio signal;

[0171] The first frequency-domain audio signal is any one of the M frequency-domain audio signals.

[0172] Optionally, in an embodiment of the present application, the above-mentioned processing module is also used to calculate wind noise existence flag information based on the amplitude square coherence MSC information of the above-mentioned first frequency domain audio signal and the proportion of high-energy low MSC frequency points of the above-mentioned first frequency domain audio signal within a preset frequency range before determining the power spectra of the M wind noise signals based on the power spectra of the above-mentioned M frequency domain audio signals; the above-mentioned proportion of high-energy low MSC frequency points is used to characterize the energy attenuation characteristics of the wind noise signal; the above-mentioned processing module is also used to determine whether there is a wind noise signal in the above-mentioned first frequency domain audio signal based on the above-mentioned wind noise existence flag information.

[0173] Optionally, in an embodiment of the present application, the above-mentioned processing module is further used to calculate the MSC information and signal power of each frequency point of the first frequency domain audio signal within the first frequency range before calculating the wind noise existence flag information based on the amplitude square coherent MSC information of the first frequency domain audio signal and the proportion of high-energy low-MSC frequency points of the first frequency domain audio signal within the first frequency range; the above-mentioned processing module is also used to calculate the proportion of high-energy low-MSC frequency points of the first frequency domain audio signal within the first frequency range based on the MSC information and signal power corresponding to each frequency point.

[0174] Optionally, in an embodiment of the present application, the above-mentioned processing module is also used to calculate the square ratio of the geometric mean to the arithmetic mean of the power spectra of the above-mentioned M frequency domain audio signals before determining the M wind noise suppression gains based on the above-mentioned M wind noise signal power spectra and wind noise suppression factors, so as to obtain the above-mentioned wind noise suppression factor.

[0175] Optionally, in an embodiment of the present application, the processing module is specifically configured to:

[0176] Calculate the suppression floor factor of each frequency domain audio signal based on the M wind noise signal power spectra;

[0177] The wind noise suppression gain of each frequency domain audio signal is calculated according to the suppression floor factor of each frequency domain audio signal and the wind noise suppression factor, to obtain the M wind noise suppression gains.

[0178] Optionally, in an embodiment of the present application, the processing module is specifically configured to calculate, for a second frequency domain audio signal among the M frequency domain audio signals, a suppression floor factor of the second frequency domain audio signal based on the frequency point energy of the wind noise signal power spectrum of the third frequency domain audio signal and the frequency point energy of the wind noise signal power spectrum of the M frequency domain audio signals;

[0179] The second frequency domain audio signal is any one of the M frequency domain audio signals; and the third frequency domain audio signal is a frequency domain audio signal different from the second frequency domain audio signal among the M frequency domain audio signals.

[0180] An audio signal processing device provided in an embodiment of the present application performs time-frequency transform processing on time-domain audio signals of M audio channels to obtain M frequency-domain audio signals, where M is an integer greater than 1. Based on the power spectra of the above-mentioned M frequency-domain audio signals, M wind noise signal power spectra are determined, where a wind noise signal power spectrum is used to characterize the distribution of the power of a wind noise signal in a frequency-domain audio signal on frequency components. M wind noise suppression gains are determined based on the above-mentioned M wind noise signal power spectra and a wind noise suppression factor, where the wind noise suppression factor is used to characterize the similarity between the signal power spectra of the above-mentioned M frequency-domain audio signals. Wind noise suppression processing is performed on the above-mentioned M frequency-domain audio signals based on the above-mentioned M wind noise suppression gains. Through this method, the audio signal processing device can respectively determine the wind noise signal power spectrum of each audio channel, and obtain the wind noise suppression gain of each audio channel based on the wind noise signal power spectrum and wind noise suppression factor of each audio channel. In this way, the wind noise suppression gain of each audio channel can be determined in a targeted manner to avoid over-suppression or erroneous suppression of the wind noise signal of the audio channel, thereby accurately eliminating the wind noise in the multi-channel audio signal, thereby improving the suppression effect of the wind noise signal of the multi-channel audio signal.

[0181] Optionally, as shown in Figure 10, an embodiment of the present application further provides an electronic device 900, including a processor 901 and a memory 902, where the memory 902 stores a program or instruction that can be run on the processor 901. When the program or instruction is executed by the processor 901, the various steps of the above-mentioned audio signal processing method embodiment are implemented and the same technical effect can be achieved. To avoid repetition, they are not repeated here.

[0182] It should be noted that the electronic devices in the embodiments of the present application include the mobile electronic devices and non-mobile electronic devices mentioned above.

[0183] FIG11 is a schematic diagram of the hardware structure of an electronic device implementing an embodiment of the present application.

[0184] The electronic device 100 includes but is not limited to components such as a radio frequency unit 101 , a network module 102 , an audio output unit 103 , an input unit 104 , a sensor 105 , a display unit 106 , a user input unit 107 , an interface unit 108 , a memory 109 , and a processor 110 .

[0185] Those skilled in the art will appreciate that the electronic device 100 may further include a power source (e.g., a battery) for powering various components. The power source may be logically connected to the processor 110 via a power management system, thereby enabling the power management system to manage charging, discharging, and power consumption. The electronic device structure shown in FIG11 does not limit the electronic device. The electronic device may include more or fewer components than shown, or may combine certain components, or have different component arrangements, which will not be described in detail here.

[0186] Among them, the above-mentioned processor 110 is used to: perform time-frequency transformation processing on the time domain audio signals of M audio channels to obtain M frequency domain audio signals; M is an integer greater than 1; determine M wind noise signal power spectra based on the power spectra of the above-mentioned M frequency domain audio signals; a wind noise signal power spectrum is used to characterize the distribution of the power of the wind noise signal in each frequency component of a frequency domain audio signal; determine M wind noise suppression gains based on the above-mentioned M wind noise signal power spectra and wind noise suppression factors; the above-mentioned wind noise suppression factor is used to characterize the similarity between the signal power spectra of the M frequency domain audio signals; and perform wind noise suppression processing on the above-mentioned M frequency domain audio signals based on the determined M wind noise suppression gains.

[0187] Optionally, in an embodiment of the present application, the processor 110 is specifically configured to: for a first frequency domain audio signal among the M frequency domain audio signals, when a wind noise signal and a voice signal exist in the first frequency domain audio signal, obtain a harmonic frequency point set of the first frequency domain audio signal; determine the signal power of the first frequency point of the first frequency domain audio signal as the first wind noise signal power corresponding to the first frequency point, and based on the first wind noise signal power, interpolate the signal power of the second frequency point of the frequency domain audio signal to obtain a second wind noise signal power corresponding to the second frequency point; determine a wind noise signal power spectrum corresponding to the second frequency domain audio signal according to the first wind noise signal power and the second wind noise signal power; wherein the first frequency point includes at least one frequency point that does not belong to the harmonic frequency point set; and the second frequency point includes at least one frequency point that belongs to the harmonic frequency point set;

[0188] Alternatively, when a wind noise signal exists in the first frequency domain audio signal and no speech signal exists, the power spectrum of the first frequency domain audio signal is used as the power spectrum of the wind noise signal corresponding to the first frequency domain audio signal;

[0189] Alternatively, when there is no wind noise signal in the first frequency domain audio signal, a preset wind noise signal power spectrum is used as the wind noise signal power spectrum corresponding to the first frequency domain audio signal;

[0190] The first frequency-domain audio signal is any one of the M frequency-domain audio signals.

[0191] Optionally, in an embodiment of the present application, the processor 110 is further configured to calculate wind noise presence flag information based on the amplitude-squared coherence MSC information of the first frequency domain audio signal and the proportion of high-energy, low-MSC frequency points of the first frequency domain audio signal within a preset frequency range before determining the power spectra of the M wind noise signals based on the power spectra of the M frequency domain audio signals; the proportion of high-energy, low-MSC frequency points is used to characterize the energy attenuation characteristics of the wind noise signal; the processor 110 is further configured to determine whether there is a wind noise signal in the first frequency domain audio signal based on the wind noise presence flag information.

[0192] Optionally, in an embodiment of the present application, the above-mentioned processor 110 is further used to calculate the MSC information and signal power of each frequency point of the first frequency domain audio signal within the first frequency range before calculating the wind noise presence flag information based on the amplitude square coherent MSC information of the first frequency domain audio signal and the proportion of high-energy low-MSC frequency points of the first frequency domain audio signal within the first frequency range; and calculate the proportion of high-energy low-MSC frequency points of the first frequency domain audio signal within the first frequency range based on the MSC information and signal power corresponding to each frequency point.

[0193] Optionally, in an embodiment of the present application, the above-mentioned processor 110 is also used to calculate the square ratio of the geometric mean and the arithmetic mean of the power spectra of the M frequency domain audio signals before determining the M wind noise suppression gains based on the above-mentioned M wind noise signal power spectra and wind noise suppression factors to obtain the wind noise suppression factor.

[0194] Optionally, in an embodiment of the present application, the processor 110 is specifically configured to: calculate a suppression floor factor of each frequency domain audio signal according to the M wind noise signal power spectra;

[0195] The wind noise suppression gain of each frequency domain audio signal is calculated according to the suppression floor factor of each frequency domain audio signal and the wind noise suppression factor, to obtain the M wind noise suppression gains.

[0196] Optionally, in an embodiment of the present application, the processor 110 is specifically configured to calculate, for a second frequency domain audio signal among the M frequency domain audio signals, a suppression floor factor of the second frequency domain audio signal based on the frequency point energy of the wind noise signal power spectrum of the third frequency domain audio signal and the frequency point energy of the wind noise signal power spectrum of the M frequency domain audio signals;

[0197] The second frequency domain audio signal is any one of the M frequency domain audio signals; and the third frequency domain audio signal is a frequency domain audio signal different from the second frequency domain audio signal among the M frequency domain audio signals.

[0198] An electronic device provided in an embodiment of the present application performs time-frequency transform processing on time-domain audio signals of M audio channels to obtain M frequency-domain audio signals, where M is an integer greater than 1. Based on the power spectra of the above-mentioned M frequency-domain audio signals, M wind noise signal power spectra are determined. A wind noise signal power spectrum is used to characterize the distribution of the power of a wind noise signal in a frequency-domain audio signal on frequency components. Based on the above-mentioned M wind noise signal power spectra and a wind noise suppression factor, M wind noise suppression gains are determined. The wind noise suppression factor is used to characterize the similarity between the signal power spectra of the above-mentioned M frequency-domain audio signals. Based on the above-mentioned M wind noise suppression gains, wind noise suppression processing is performed on the above-mentioned M frequency-domain audio signals. Through this method, the electronic device can determine the wind noise signal power spectrum of each audio channel respectively, and obtain the wind noise suppression gain of each audio channel based on the wind noise signal power spectrum and wind noise suppression factor of each audio channel. In this way, the wind noise suppression gain of each audio channel can be determined in a targeted manner to avoid over-suppression or erroneous suppression of the wind noise signal of the audio channel, thereby accurately eliminating the wind noise in the multi-channel audio signal, thereby improving the suppression effect of the wind noise signal of the multi-channel audio signal.

[0199] It should be understood that in an embodiment of the present application, the input unit 104 may include a graphics processing unit (GPU) 1041 and a microphone 1042, and the graphics processor 1041 processes the image data of a static picture or video obtained by an image capture device (such as a camera) in a video capture mode or an image capture mode. The display unit 106 may include a display panel 1061, and the display panel 1061 may be configured in the form of a liquid crystal display, an organic light emitting diode, etc. The user input unit 107 includes a touch panel 1071 and at least one of other input devices 1072. The touch panel 1071 is also called a touch screen. The touch panel 1071 may include two parts: a touch detection device and a touch controller. Other input devices 1072 may include, but are not limited to, a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a trackball, a mouse, and a joystick, which will not be repeated here.

[0200] The memory 109 can be used to store software programs and various data. The memory 109 may mainly include a first storage area for storing programs or instructions and a second storage area for storing data, wherein the first storage area may store an operating system, applications or instructions required for at least one function (such as a sound playback function, an image playback function, etc.). In addition, the memory 109 may include a volatile memory or a non-volatile memory, or the memory 109 may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), a static random access memory (SRAM), a dynamic random access memory (DRAM), a synchronous dynamic random access memory (SDRAM), a double data rate synchronous dynamic random access memory (DDRSDRAM), an enhanced synchronous dynamic random access memory (ESDRAM), a synchronous link dynamic random access memory (SLDRAM), and a direct memory bus random access memory (DRRAM). The memory 109 in the embodiment of the present application includes but is not limited to these and any other suitable types of memory.

[0201] Processor 110 may include one or more processing units. Optionally, processor 110 integrates an application processor and a modem processor. The application processor primarily handles operations related to the operating system, user interface, and application programs, while the modem processor primarily processes wireless communication signals, such as a baseband processor. It is understood that the modem processor may not be integrated into processor 110.

[0202] An embodiment of the present application further provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the various processes of the above-mentioned audio signal processing method embodiment are implemented and can achieve the same technical effect. To avoid repetition, they are not further described here.

[0203] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer-readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0204] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned audio signal processing method embodiment, and can achieve the same technical effect. To avoid repetition, it is not repeated here.

[0205] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0206] An embodiment of the present application provides a computer program product, which is stored in a storage medium and executed by at least one processor to implement the various processes of the above-mentioned audio signal processing method embodiment and can achieve the same technical effects. To avoid repetition, it is not described here.

[0207] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. In addition, it should be noted that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in the opposite order according to the functions involved. For example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0208] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the relevant technology, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0209] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also make many forms without departing from the purpose of this application and the scope of protection of the claims, all of which are within the protection of this application.

Claims

1. A method for processing an audio signal, the method comprising: Performing time-frequency transformation on the time-domain audio signals of the M audio channels to obtain M frequency-domain audio signals; M is an integer greater than 1; Based on the power spectra of the M frequency-domain audio signals, M wind noise signal power spectra are determined; a wind noise signal power spectrum is used to characterize the distribution of power of a wind noise signal in a frequency-domain audio signal on each frequency component; Determining M wind noise suppression gains according to the M wind noise signal power spectra and the wind noise suppression factor; the wind noise suppression factor is used to characterize the similarity between the signal power spectra of the M frequency domain audio signals; Wind noise suppression processing is performed on the M frequency-domain audio signals according to the M wind noise suppression gains.

2. The method according to claim 1, wherein: The determining M wind noise signal power spectra based on the power spectra of the M frequency-domain audio signals includes: For a first frequency domain audio signal among the M frequency domain audio signals, when a wind noise signal and a speech signal exist in the first frequency domain audio signal, obtaining a harmonic frequency point set of the first frequency domain audio signal; determining a signal power of a first frequency point of the first frequency-domain audio signal as a first wind noise signal power corresponding to the first frequency point, and performing interpolation processing on a signal power of a second frequency point of the frequency-domain audio signal based on the first wind noise signal power to obtain a second wind noise signal power corresponding to the second frequency point; Determine a wind noise signal power spectrum corresponding to the first frequency domain audio signal according to the first wind noise signal power and the second wind noise signal power; wherein the first frequency point includes at least one frequency point that does not belong to the harmonic frequency point set; and the second frequency point includes at least one frequency point that belongs to the harmonic frequency point set; Alternatively, when there is a wind noise signal in the first frequency domain audio signal and no speech signal, the power spectrum of the first frequency domain audio signal is used as the power spectrum of the wind noise signal corresponding to the first frequency domain audio signal; Alternatively, when there is no wind noise signal in the first frequency-domain audio signal, a preset wind noise signal power spectrum is used as the wind noise signal power spectrum corresponding to the first frequency-domain audio signal; The first frequency-domain audio signal is any frequency-domain audio signal among the M frequency-domain audio signals.

3. The method according to claim 2, wherein: Before determining the M wind noise signal power spectra based on the power spectra of the M frequency-domain audio signals, the method further includes: Calculating wind noise existence flag information according to the amplitude square coherent MSC information of the first frequency domain audio signal and the proportion of high energy low MSC frequency points of the first frequency domain audio signal within the first frequency range; the proportion of high energy low MSC frequency points is used to characterize the energy attenuation characteristics of the wind noise signal; According to the wind noise existence flag information, it is determined whether a wind noise signal exists in the first frequency domain audio signal.

4. The method according to claim 3, wherein: Before calculating the wind noise existence flag information according to the amplitude square coherent MSC information of the first frequency domain audio signal and the proportion of high energy low MSC frequency points of the first frequency domain audio signal within the first frequency range, the method further includes: Calculating MSC information and signal power of each frequency point of the first frequency domain audio signal within a first frequency range; According to the MSC information and signal power corresponding to each frequency point, a proportion of high-energy low-MSC frequency points of the first frequency-domain audio signal within the first frequency range is calculated.

5. The method according to claim 1, wherein: Before determining the M wind noise suppression gains according to the M wind noise signal power spectra and the wind noise suppression factors, the method further includes: The wind noise suppression factor is obtained by calculating the square of the ratio of the geometric mean to the arithmetic mean of the power spectra of the M frequency-domain audio signals.

6. The method according to claim 1, wherein: The determining M wind noise suppression gains according to the M wind noise signal power spectra and the wind noise suppression factors includes: Calculating the suppression floor factors of the M frequency-domain audio signals according to the M wind noise signal power spectra; The wind noise suppression gains of the M frequency domain audio signals are calculated according to the suppression floor factors of the M frequency domain audio signals and the wind noise suppression factors.

7. The method according to claim 6, wherein: The calculating, according to the M wind noise signal power spectra, the suppression floor factors of the M frequency domain audio signals comprises: For a second frequency domain audio signal among the M frequency domain audio signals, calculating a suppression floor factor of the second frequency domain audio signal according to frequency point energy of a wind noise signal power spectrum of a third frequency domain audio signal and frequency point energy of a wind noise signal power spectrum of the M frequency domain audio signals; The second frequency domain audio signal is any one of the M frequency domain audio signals; and the third frequency domain audio signal is a frequency domain audio signal different from the second frequency domain audio signal among the M frequency domain audio signals.

8. An audio signal processing device, the device comprising: Processing module; The processing module is used to: Performing time-frequency transformation on the time-domain audio signals of the M audio channels to obtain M frequency-domain audio signals; M is an integer greater than 1; Based on the power spectra of the M frequency-domain audio signals obtained by the processing module, M wind noise signal power spectra are determined; a wind noise signal power spectrum is used to characterize the distribution of power of a wind noise signal in a frequency-domain audio signal on each frequency component; Determining M wind noise suppression gains according to the M wind noise signal power spectra and the wind noise suppression factor; the wind noise suppression factor is used to characterize the similarity between the signal power spectra of the M frequency domain audio signals; According to the M wind noise suppression gains determined by the determination module, wind noise suppression processing is performed on the M frequency domain audio signals.

9. The device according to claim 8, wherein: The processing module is specifically used for: For a first frequency domain audio signal among the M frequency domain audio signals, when a wind noise signal and a speech signal exist in the first frequency domain audio signal, obtaining a harmonic frequency point set of the first frequency domain audio signal; determining a signal power of a first frequency point of the first frequency-domain audio signal as a first wind noise signal power corresponding to the first frequency point, and performing interpolation processing on a signal power of a second frequency point of the frequency-domain audio signal based on the first wind noise signal power to obtain a second wind noise signal power corresponding to the second frequency point; Determine a wind noise signal power spectrum corresponding to the second frequency domain audio signal according to the first wind noise signal power and the second wind noise signal power; wherein the first frequency point includes at least one frequency point that does not belong to the harmonic frequency point set; and the second frequency point includes at least one frequency point that belongs to the harmonic frequency point set; Alternatively, when there is a wind noise signal in the first frequency domain audio signal and no speech signal, the power spectrum of the first frequency domain audio signal is used as the power spectrum of the wind noise signal corresponding to the first frequency domain audio signal; Alternatively, when there is no wind noise signal in the first frequency-domain audio signal, a preset wind noise signal power spectrum is used as the wind noise signal power spectrum corresponding to the first frequency-domain audio signal; The first frequency-domain audio signal is any frequency-domain audio signal among the M frequency-domain audio signals.

10. The device according to claim 8, wherein: The processing module is further used to calculate the square ratio of the geometric mean to the arithmetic mean of the power spectra of the M frequency-domain audio signals before determining the M wind noise suppression gains according to the M wind noise signal power spectra and the wind noise suppression factor to obtain the wind noise suppression factor.

11. An electronic device comprising a processor and a memory, wherein the memory stores a program or instruction that can be run on the processor, and when the program or instruction is executed by the processor, the steps of the audio signal processing method according to any one of claims 1 to 7 are implemented.

12. A readable storage medium storing a program or an instruction, wherein the program or the instruction, when executed by a processor, implements the steps of the audio signal processing method according to any one of claims 1 to 7.

13. A chip, comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run a program or instruction to implement the method according to any one of claims 1 to 7.

14. A computer program product, wherein the program product is executed by at least one processor to implement the method according to any one of claims 1 to 7.

Images