WO2022027700A1

WO2022027700A1 - Wind noise resistance method

Info

Publication number: WO2022027700A1
Application number: PCT/CN2020/108097
Authority: WO
Inventors: 韩荣
Original assignee: 中科新声(苏州)科技有限公司
Priority date: 2020-08-07
Filing date: 2020-08-10
Publication date: 2022-02-10
Also published as: CN112019958B; CN112019958A

Abstract

A wind noise resistance method. A signal sample set of a first feedforward microphone and a second feedforward microphone is obtained, and a correlation coefficient feature and an energy feature are then obtained. Whether wind noise occurs in a usage environment is determined by means of comparison of a correlation coefficient and comparison of energy. Real-time wind noise is determined by means of updating a signal in real time. When wind noise occurs, outputs of the first feed-forward microphone and the second feed-forward microphone are turned down, and when the wind noise ends, the outputs are restored. According to the method, wind noise resistance processing is carried out according to the determination of wind noise, such that an adaptive wind noise resistance effect can be realized, and the influence of noise during a microphone application process can be reduced to the maximum extent.

Description

A method of anti-wind noise

technical field

The present invention relates to a method for resisting wind noise.

Background technique

In the process of microphone application, it is often required to be exposed to the external environment. For example, the feed-forward microphone of the noise-cancelling headset needs to be installed outside the headset to collect ambient noise signals. When there is a strong wind in the environment, the feed-forward microphone will collect wind noise, and the existence of wind noise will affect the normal signal processing process. In order to alleviate the influence of wind noise, the existing earphones reduce the collected wind noise by adding a windproof duct between the opening of the feedforward microphone of the earphone and the microphone or adjusting the position and orientation of the feedforward microphone. However, using these methods When the wind noise is reduced, it will still be collected by the feedforward microphone, which affects the noise reduction effect.

In order to improve the noise reduction effect, it is necessary to perform reliable wind noise detection before noise reduction, and optimize the working state of the system after detecting wind noise to improve the noise reduction effect when there is wind.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide an anti-wind noise method, which can perform reliable wind noise detection, optimize the system working state according to the judgment of wind noise, improve the noise reduction effect, and solve the problem of head-mounted and neck-mounted active noise reduction. When there is wind noise, the control effect of the noise earphone becomes poor, and it even produces additional noise.

In order to solve the above-mentioned technical problems, the technical scheme of the present invention is:

A method for anti-wind noise, comprising the following steps:

Using the first feed-forward microphone and the second feed-forward microphone to obtain a first set of signal samples and a second set of signal samples;

obtaining a first energy sample in the first signal sample set, obtaining a second energy sample in the second signal sample set, and obtaining a correlation coefficient sample in the first signal sample set and the second signal sample set;

Continuously update the first signal sample set and the second signal sample set to obtain the first energy sample set, the second energy sample set and the correlation coefficient sample set,

When in the correlation coefficient sample set, the number of correlation coefficient samples smaller than the first threshold reaches 0.5 to 1.0 times the number of samples in the correlation coefficient sample set, and when the first energy sample set and the second energy sample set, the first energy and the second energy sample value are both greater than the second threshold, judging the occurrence of wind noise;

When the number of correlation coefficient samples greater than the third threshold in the correlation coefficient sample set reaches 0.5 to 1.0 times the number of samples in the correlation coefficient sample set, it is determined that the wind noise is over.

When the occurrence of wind noise is detected, the controller reduces the signal output of the feedforward control unit corresponding to the first feedforward microphone and the second feedforward microphone, and maintains the signal output of the feedback control unit corresponding to the feedback microphone;

When the end of the wind noise is detected, the controller restores the signal output of the feedforward control units corresponding to the first feedforward microphone and the second feedforward microphone.

Preferably, when wind noise occurs, the controller reduces the signal output intensity of the feedforward control units corresponding to the first feedforward microphone and the second feedforward microphone to 0-0.8 times the original.

Preferably, the same sampling rate is set for sampling to obtain the first set of signal samples and the second set of signal samples.

Preferably, the first signal sample set and the second signal sample set are updated, and the sample repetition rate in the sample set is set to 20-80%.

Preferably, the correlation coefficient is solved by using the time domain correlation or frequency domain correlation solving formula, and the correlation coefficient sample is obtained.

Preferably, the energy solution is performed using a sum-of-square formula to obtain the first energy sample and the second energy sample.

Preferably, the first threshold value method is as follows: in the anechoic chamber, a sound source is set in front of the microphone, and a stable sound source is played. When the sound pressure level at the first feedforward microphone and the second feedforward microphone is are both 60dB, the first feedforward microphone and the second feedforward microphone obtain signals, and solve the correlation coefficient to obtain the correlation coefficient in the anechoic chamber, and cancel 0.1-0.5 times of the correlation coefficient in the anechoic chamber as the first threshold .

Preferably, the second threshold value method is: in the anechoic room, when the noise is the noise floor of the anechoic room, the signals of the first feedforward microphone and the second feedforward microphone are obtained, and the energy is calculated to cancel the sound. The energy in the chamber plus 5dB is used as the second threshold.

Preferably, the third threshold value method is as follows: in the anechoic chamber, a sound source is set in front of the microphone, and a stable sound source is played. When the sound pressure levels at the first feedforward microphone and the second feedforward microphone are both equal is 60dB, the first feedforward microphone and the second feedforward microphone obtain signals, and perform correlation coefficient calculation to obtain the correlation coefficient in the anechoic chamber, and cancel 0.2-0.9 times the correlation coefficient in the anechoic chamber as the third threshold.

Compared with the prior art, the present invention has the following advantages: the method for anti-wind noise obtains a set of signal samples obtained by two microphones at the same time point, performs feature analysis on the signal samples, Determine whether there is wind noise in the environment during use. Further, dynamically obtain the signal sample sets of the two microphones in real time, analyze the characteristics of the signal samples, and perform anti-wind noise processing according to the judgment of wind noise. When wind noise occurs, reduce the first feedforward microphone and the second feedforward microphone. The output of the microphone, when the wind noise ends, restores the output, which can achieve an adaptive anti-wind noise effect, minimize the impact of wind noise in the microphone application process, and improve the noise reduction effect.

Description of drawings

The drawings described herein are for explanatory purposes only and are not intended to limit the scope of the present disclosure in any way. In addition, the shapes and proportions of the components in the figures are only schematic and are used to help the understanding of the present invention, and do not specifically limit the shapes and proportions of the components of the present invention. Under the teachings of the present invention, those skilled in the art can select various possible shapes and proportions according to specific conditions to implement the present invention. In the attached image:

1 and 2 are flowcharts of a method for detecting wind noise according to Embodiment 1 of the present invention.

detailed description

In order to make those skilled in the art better understand the technical solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described The embodiments are only some of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

It should be noted that when an element is referred to as being "disposed on" another element, it can be directly on the other element or intervening elements may also be present. When an element is referred to as being "connected" to another element, it can be directly connected to the other element or intervening elements may also be present. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for the purpose of illustration only and do not represent the only embodiment.

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

Example 1

A method for detecting wind noise, comprising the steps of:

Referring to Figures 1 and 2, a first set of signal samples is obtained from the first feed-forward microphone; a second set of signal samples is obtained from the second feed-forward microphone, and the first set of signal samples and the second set of signal samples are obtained simultaneously .

The first feed-forward microphone and the second feed-forward microphone are wiredly connected.

Correlation and energy characteristic analysis is performed on the first set of signal samples and the second set of signal samples. First energy samples are obtained in the first signal sample set, second energy samples are obtained in the second signal sample set, and correlation coefficient samples are obtained in the first signal sample set and the second signal sample set.

Continuously update the first signal sample set and the second signal sample set, specifically, replace the original signal in the signal sample set with the newly collected signal sample to obtain the updated first signal sample set and the second signal sample set . Using the same method, perform correlation and energy feature analysis on the updated first signal sample set and the second signal sample set to obtain the first energy sample, the second energy sample and the correlation coefficient sample, thereby obtaining the first energy sample set , a second energy sample set and a correlation coefficient sample set.

When in the correlation coefficient sample set, the number of correlation coefficient samples smaller than the first threshold reaches 0.5 to 1.0 times the number of samples in the correlation coefficient sample set, and when the first energy sample set and the second energy sample set, the first energy and the second energy sample value is greater than the second threshold, it is judged that wind noise occurs.

When the first signal sample set and the second signal sample set are obtained from the first feed-forward microphone and the second feed-forward microphone, a sampling rate is set, for example, the sampling rate is 16 kHz. The two feedforward microphones sample at the same time, continuously collect external signals, and form their own signal streams. The signals are divided into signal windows of length L. For example, L=512, the first signal sample set with 512 samples is obtained and A second set of signal samples. Correlation and energy feature analysis are performed on the first signal sample set and the second signal sample set to obtain a first energy sample, a second energy sample and a correlation coefficient sample. The first feed-forward microphone and the second feed-forward microphone continuously collect signals, and the signals in the first signal sample set and the second signal sample set are continuously updated. In order to maintain data stability, the first signal is set The overlap rate of the signal samples in the sample set and the second signal sample set is 20-80%, preferably 50%. When the number of newly collected signals reaches half of the total number of signal sample sets, the original samples are replaced by the newly collected signals. The first half of the signal acquired in the set. The updated first signal sample set and the second signal sample set are subjected to correlation and energy feature analysis to obtain updated first energy samples, second energy samples and correlation coefficient samples, and the continuously obtained first energy samples form The first energy sample set, similarly, forms the second energy sample set and the correlation coefficient sample set.

When wind noise is detected, the correlation coefficient feature is solved for the first signal sample set and the second signal sample set by using a correlation solving formula to obtain correlation coefficient samples. At the same time, the energy characteristic solution is performed on the first signal sample set and the second signal sample set respectively by using an energy solution formula.

When the correlation coefficient feature is solved, the correlation coefficient solution can be obtained by using the time domain correlation solution formula or the frequency domain correlation solution formula to obtain the correlation coefficient sample.

The first signal sample set is represented by WLi, and the second signal sample set is represented by WRi.

When the correlation coefficient is solved by using the time domain correlation solution formula,

Time-domain correlation coefficient x=max(corr(WLi, WRi))/sqrt(WLi^2×WRi^2). Among them, corr is the function of the correlation coefficient, and max is the maximum value function.

When the correlation coefficient is solved using the frequency domain correlation solving formula,

First, the following formula is used to perform Fourier transform on the signal samples to obtain the frequency domain information,

F1=fft(WLi), F2=fft(WRi); wherein, F1 represents the frequency domain information of the first signal sample set, and F2 represents the frequency domain information of the second signal sample set.

Then, the frequency domain correlation coefficient is solved, and the frequency domain correlation coefficient WF=F1×F2. Each amplitude value in WF represents the correlation coefficient at each frequency. Therefore, the amplitude corresponding to the effective frequency band in the WF is taken as the correlation coefficient at the current moment. It is preferable to take the sum of squares of the correlation coefficients corresponding to frequency bands below 500 Hz.

When the energy characteristic is solved, the square sum formula is used to solve the energy. The square sum of all the signal samples in the first signal sample set is the first energy sample, and the square sum of all the signal samples in the second signal sample set is the second energy sample.

As the first signal sample set and the second signal sample set are continuously updated, the first energy sample set, the second energy sample set and the correlation coefficient sample set are obtained.

When the number of correlation coefficient samples in the correlation coefficient sample set smaller than the first threshold reaches 0.5 to 1.0 times the number of samples in the correlation coefficient sample set, and when the first energy sample set and the second energy sample set, the first energy and the second energy sample value is greater than the second threshold, it is judged that wind noise occurs.

When the number of correlation coefficient samples in the correlation coefficient sample set greater than the third threshold reaches 0.5-1.0 times the number of samples in the correlation coefficient sample set, it is determined that the wind noise is over.

The method for obtaining the first threshold value is as follows: in the anechoic chamber, a sound source is set in front of the microphone, and a stable sound source is played, and when the sound pressure levels at the first feedforward microphone and the second feedforward microphone are equal is 60dB, the first feedforward microphone and the second feedforward microphone obtain signals, and solve the correlation coefficient to obtain the correlation coefficient in the anechoic chamber, and cancel the correlation coefficient in the anechoic chamber 0.1-0.5 times as the first threshold.

The method for obtaining the second threshold value is: in the anechoic chamber, when the noise is the background noise of the anechoic chamber, the signals of the first feedforward microphone and the second feedforward microphone are obtained, and energy is calculated to cancel the noise in the anechoic chamber. Add 5dB to the energy as the second threshold.

The third threshold value method is: in the anechoic chamber, set a sound source in front of the microphone, and play a stable sound source, when the sound pressure level at the first feedforward microphone and the second feedforward microphone are both 60dB, The first feed-forward microphone and the second feed-forward microphone obtain signals, and perform correlation coefficient calculation to obtain the correlation coefficient in the anechoic chamber, and 0.2-0.9 times the correlation coefficient in the anechoic chamber is used as the third threshold.

The method for detecting wind noise can identify whether there is wind noise during the use of the microphone by judging the correlation and energy at the same time, with high accuracy and convenience.

Embodiment 2

A method for anti-wind noise, comprising the following steps:

A sampling rate is set, and the first feed-forward microphone and the second feed-forward microphone continuously collect external signals to form respective signal streams to obtain a first signal sample set and a second signal sample set. Wind noise is detected by using the method for detecting wind noise in the first embodiment, and when wind noise is detected, the controller reduces the signal output of the feedforward control units corresponding to the first feedforward microphone and the second feedforward microphone, The signal output of the feedback control unit corresponding to the feedback microphone is maintained, and when the end of the wind noise is detected, the controller restores the signal output of the feedforward control unit corresponding to the first feedforward microphone and the second feedforward microphone. Specifically, when wind noise occurs, the controller reduces the signal output intensity of the feedforward control units corresponding to the first feedforward microphone and the second feedforward microphone to 0 to 0.8 times the original.

The first feed-forward microphone and the second feed-forward microphone continuously collect external signals to form respective signal streams, and the signals are divided into signal windows of length L to obtain the first signal sample set and the second signal sample set. signal samples inside. The first feed-forward microphone and the second feed-forward microphone continuously collect external signals, and after new signals are collected, the first signal sample set and the second signal sample set are continuously updated. The overlap rate of the signal sample set and the second signal sample set is 50%. When the number of newly collected signals reaches half of the total number of signal sample sets, the newly collected signals replace the first half of the signals collected in the original sample set. Set the signal window length L=512, when a new signal is collected, the signal window will slide once every 256 signal points to update the signal samples in the signal sample set, the first signal sample set and the second signal The sample set is updated at the same time.

Embodiment 3

A device for detecting wind noise, comprising:

a first feedforward microphone, a second feedforward microphone and a controller,

the controller obtains a first set of signal samples from a first feedforward microphone;

obtaining a second set of signal samples from the second feedforward microphone, the first set of signal samples and the second set of signal samples obtained simultaneously;

The device for detecting wind noise includes a headset or a neck-worn headset.

The left and right pre-aural feedback microphones of the head-worn or neck-worn earphone can be directly connected, and the relationship between the signals at the left and right ears can be used to judge whether the wind noise exists, so as to realize the control of the wind noise. The first feedforward microphone and the second feedforward microphone are the left and right ear feedback microphones of the head-mounted or neck-mounted headset, and the signals obtained by the left and right feed-forward microphones are used to judge whether wind noise exists. When the occurrence of wind noise is detected, the controller in the earphone reduces the signal output of the feedforward control unit corresponding to the first feedforward microphone and the second feedforward microphone, and maintains the signal output of the feedback control unit corresponding to the feedback microphone. When the end of the wind noise is detected, the controller restores the signal output of the feedforward control units corresponding to the first feedforward microphone and the second feedforward microphone.

It should be understood that the above description is for purposes of illustration and not limitation. From reading the above description, many embodiments and many applications beyond the examples provided will be apparent to those skilled in the art. The scope of the present teachings should, therefore, be determined not with reference to the above description, but should instead be determined with reference to the preceding claims, along with the full scope of equivalents to which such claims are entitled. The disclosures of all articles and references, including patent applications and publications, are incorporated herein by reference for the purpose of being comprehensive. The omission of any aspect of the subject matter disclosed herein in the preceding claims is not intended to disclaim such subject matter, nor should the applicant be considered as not considering such subject matter as part of the disclosed subject matter.

Claims

A method for resisting wind noise, comprising the steps of:

Using the first feed-forward microphone and the second feed-forward microphone to obtain a first set of signal samples and a second set of signal samples;

obtaining a first energy sample in the first signal sample set, obtaining a second energy sample in the second signal sample set, and obtaining a correlation coefficient sample in the first signal sample set and the second signal sample set;

Continuously update the first signal sample set and the second signal sample set to obtain the first energy sample set, the second energy sample set and the correlation coefficient sample set,

When in the correlation coefficient sample set, the number of correlation coefficient samples smaller than the first threshold reaches 0.5 to 1.0 times the number of samples in the correlation coefficient sample set, and when the first energy sample set and the second energy sample set, the first energy and the second energy sample value are both greater than the second threshold, judging the occurrence of wind noise;

When the number of correlation coefficient samples greater than the third threshold in the correlation coefficient sample set reaches 0.5 to 1.0 times the number of samples in the correlation coefficient sample set, it is determined that the wind noise is over.

When the occurrence of wind noise is detected, the controller reduces the signal output of the feedforward control unit corresponding to the first feedforward microphone and the second feedforward microphone, and maintains the signal output of the feedback control unit corresponding to the feedback microphone;

When the end of the wind noise is detected, the controller restores the signal output of the feedforward control units corresponding to the first feedforward microphone and the second feedforward microphone.
The method for anti-wind noise according to claim 1, wherein when wind noise occurs, the controller reduces the signal output of the feedforward control units corresponding to the first feedforward microphone and the second feedforward microphone. The strength is 0 to 0.8 times of the original.
The method for anti-wind noise according to claim 1, wherein the first signal sample set and the second signal sample set are obtained by setting the same sampling rate for sampling.
The method for anti-wind noise according to claim 1, wherein the first signal sample set and the second signal sample set are updated, and the sample repetition rate in the sample set is set to be 20-80%.
The method for anti-wind noise according to claim 1, wherein a correlation coefficient is obtained by using a time domain correlation or frequency domain correlation solving formula to obtain a correlation coefficient sample.
The method for anti-wind noise according to claim 1, characterized in that the energy solution is performed using a sum of squares formula to obtain the first energy sample and the second energy sample.
The method for anti-wind noise according to claim 1, wherein the method for obtaining the first threshold value is as follows: in an anechoic chamber, a sound source is set in front of a microphone, and a stable sound source is played, and when the first threshold value is The sound pressure levels at the feed-forward microphone and the second feed-forward microphone are both 60dB, the first feed-forward microphone and the second feed-forward microphone obtain signals, and the correlation coefficient is solved to obtain the correlation coefficient in the anechoic chamber, and cancel the sound 0.1-0.5 times the correlation coefficient in the chamber was used as the first threshold.
The method for anti-wind noise according to claim 1, wherein the second threshold value method is: in the anechoic room, when the noise is the background noise of the anechoic room, obtain the first feedforward microphone and the second The signal of the two feedforward microphones, and the energy solution is carried out, and the energy in the cancellation chamber is added 5dB as the second threshold.
The method for anti-wind noise according to claim 1, wherein the third threshold value method is as follows: in an anechoic room, a sound source is set in front of the microphone, and a stable sound source is played. The sound pressure levels at the feed-forward microphone and the second feed-forward microphone are both 60dB, the first feed-forward microphone and the second feed-forward microphone obtain signals, and the correlation coefficient is solved to obtain the correlation coefficient in the anechoic chamber, and the anechoic chamber is cancelled. 0.2-0.9 times the correlation coefficient in as the third threshold.