WO2023077252A1

WO2023077252A1 - Fxlms structure-based active noise reduction system, method, and device

Info

Publication number: WO2023077252A1
Application number: PCT/CN2021/128037
Authority: WO
Inventors: 阎钰; 梁羽贤; 邓翔宇; 殷福亮; 陈喆; 张立斌
Original assignee: 华为技术有限公司; 大连理工大学
Priority date: 2021-11-02
Filing date: 2021-11-02
Publication date: 2023-05-11
Also published as: CN118251717A

Abstract

The present application discloses an FxLMS structure-based active noise reduction system, a method, and a device. The device comprises: a primary path adaptive filter, a secondary path adaptive filter, and a subband filter set, the subband filter set being used to adjust a filter coefficient of the secondary path adaptive filter on the basis of a target playback sound. On the basis of an existing FxLMS structure, a subband filter set is added to the system, so that, in a process during which an active noise reduction device plays a target sound (such as a music signal), the subband filter set is used to whiten the target playback sound to eliminate correlation (this is due to the fact that the correlation in a music signal is strong, an eigenvalue is close to zero, a step size needs to be set smaller, and therefore a convergence speed of the adaptive filter is severely affected), so as to obtain a real-time filter coefficient of the secondary path adaptive filter. Compared with an offline method for modeling a secondary path using white noise, real-time online modeling is performed directly using target playback sound, improving user comfort. The method is more practical, and can be dynamically and automatically adjusted in real time.

Description

An active noise reduction system, method and device based on FxLMS structure

technical field

The present application relates to the field of audio processing, and in particular to an active noise reduction system, method and device based on an FxLMS structure.

Background technique

Commonly used noise reduction systems (also called noise reduction devices) are mainly divided into three types according to different working positions: noise reduction at the sound source, noise reduction during transmission, and noise reduction at the human ear. In order to actively eliminate noise, in the noise reduction of headphones, people have proposed "active noise cancellation" technology, which eliminates the noise in the environment by generating a signal opposite to the noise near the human ear. The principle is to generate reverse sound waves equal to the external noise through the noise reduction system to neutralize the noise, thereby achieving noise reduction. In active noise reduction equipment (such as noise reduction headphones), it involves digital-to-analog conversion (D\A), signal circuits, analog-to-digital conversion (A\D), filter circuits, speakers, microphones and other electronic equipment and speakers The combination of physical channels such as the actual pipes between the microphones and the microphones is called the secondary channel.

Since everyone has different head sizes, different angles of wearing noise reduction equipment, and different environments, the noise reduction effect will inevitably be affected by these factors. In recent years, with the rapid development of noise-canceling headphones and other active noise-canceling devices, it has become more and more urgent to establish a suitable online modeling model. A typical noise reduction system adopts the structure of X-filtered least mean square error (filtered-X least mean square, FxLMS) algorithm (which can be referred to as FxLMS structure or FxLMS algorithm). The block diagram of the current mainstream FxLMS structure is shown in Figure 1 As shown, the FxLMS structure includes the primary path adaptive filter W(z), the secondary path adaptive filter

(for simulating the secondary path), and the connection/computation relationship between the two has been determined, as shown in the block diagram in Figure 1, where the meanings of the symbols are x(n): noise signal; P(z): primary path, also called primary path; d(n): primary noise signal of error microphone; S(z): secondary path, also called secondary path; y(n): primary The output signal of the channel adaptive filter W(z); y'(n): the secondary noise signal through the secondary channel; x'(n): the noise signal generated by the estimated secondary channel, which can also be called Estimated signal; e(n): error signal of the system.

However, the existing FxLMS structure first uses white noise to model the secondary path offline, and then directly uses the modeled secondary path coefficients (ie, the filter coefficients of the secondary path adaptive filter) to remove music and other processing. , and then perform adaptive LMS adjustment for noise reduction on the primary channel. However, in practice, the secondary channel is not fixed, but will change according to the wearer's wearing method and wearing time, and offline design in advance will affect the noise reduction effect. In addition, the FxLMS structure is not effective for high-frequency noise reduction, because the high-frequency audio signal propagates faster, and the filter processing time is too long compared to the propagation time, so that it is impossible to play the reverse sound wave that can offset the external noise. Therefore, for High-frequency noise cancellation is poor.

Contents of the invention

The embodiment of the present application provides an active noise reduction system, method and device based on the FxLMS structure. The noise reduction system adds a sub-band filter bank on the basis of the existing FxLMS structure, so that the active noise reduction device (such as , noise-canceling earphones) in the process of playing the target sound (such as music signal), the sub-band filter bank is used to whiten the target sound to eliminate its correlation (this is because the correlation of the music signal is strong, and the eigenvalue close to zero, the step size needs to be set smaller, which seriously affects the convergence speed of the adaptive filter) to obtain the filter coefficients of the real-time secondary path adaptive filter, compared with the off-line method of using white noise for secondary path modeling , directly using the target playback sound to model in real time on-line improves user comfort, is more practical, and can be adjusted dynamically in real time.

Based on this, the embodiment of the present application provides the following technical solutions:

In the first aspect, the embodiment of the present application first provides an active noise reduction system based on the FxLMS structure, which can be used in the field of active noise reduction. The system includes: primary path adaptive filter W(z), secondary path adaptive filter device

and a subband filter bank, wherein the subband filter bank includes M bandpass filters, and different bandpass filters have different passbands, which can be called G ₀ (z), G 1 (z), G ₁ (z), ..., G _M-1 (z), M≥1. Specifically, the primary path adaptive filter W(z) is used to simulate the primary path, wherein the primary path represents the transmission of the external ambient sound (which may be called the first signal) at the reference microphone of the active noise reduction device to the active noise reduction The physical path through which the loudspeaker of the device passes, so this primary path can also be called the air path; the secondary path adaptive filter

Used to simulate the secondary path, where the secondary path represents the circuit path through which the second signal is transmitted to the loudspeaker, and the second signal is the output signal of the primary path adaptive filter W(z) and the target playback sound (such as a headphone Played music) calculated signal, for example, the operation can be the output signal of the primary path adaptive filter W(z) and the target playback sound; the subband filter bank is used to adjust the sub-band based on the target playback sound stage path adaptive filter

filter coefficients.

In the above-mentioned embodiments of the present application, the noise reduction system provided by the embodiment of the present application is based on the existing FxLMS structure, and a sub-band filter bank is added, so that the target sound can be played on the active noise reduction device (such as noise reduction headphones) (e.g., music signal), use the sub-band filter bank to whitenize the target playing sound to obtain the filter coefficients of the real-time secondary path adaptive filter, compared with using white noise for the secondary path The offline mode of modeling directly uses the target playback sound to model online in real time, which improves user comfort, is more practical, and can be adjusted dynamically in real time.

In a possible implementation of the first aspect, the active noise reduction system may further include a sampling module, which can perform up-sampling or down-sampling on the signal, specifically, it can be used to: (1) filter the sub-band The output signal of each band-pass filter in the filter group is down-sampled to obtain a down-sampled signal. Assuming that there are M band-pass filters, a total of M down-sampled signals are obtained, and the obtained M down-sampled signals are used as secondary channel adaptive filter

input signal. (2) Adaptive filter for the secondary path

The output signal is up-sampled to obtain an up-sampled signal. One up-sampled signal corresponds to one down-sampled signal. Assuming that there are M band-pass filters, there are M down-sampled signals and M up-sampled signals.

In the above embodiments of the present application, the active noise reduction system provided in the embodiment of the present application further includes a sampling module for down-sampling and/or up-sampling. In the embodiment of the present application, there are two purposes of down-sampling: 1 , Broaden the specific frequency bands passed by each bandpass filter to achieve white noise, and the white noise signal is used to adjust the secondary channel adaptive filter

coefficient; 2, reduce the secondary path adaptive filter

calculation amount. The purpose of upsampling is the opposite operation to downsampling, in order to restore the broadened frequency band to facilitate subsequent calculations.

In a possible implementation manner of the first aspect, the input signals of the sub-band filter bank may include the following types: the target playing sound, the up-sampling signal output by the sampling module, and the system error signal.

In the above embodiments of the present application, it is specifically stated that at different stages, the input signals of the sub-band filter bank can be different input signals, that is, different signals can share the same sub-band filter bank, which is flexible, and Saves hardware cost.

In a possible implementation of the first aspect, when the input signal of the sub-band filter bank is the target playing sound, the sampling module is specifically configured to: The signal (which may be referred to as the first output signal) is down-sampled, for example, m times of down-sampling is performed to obtain a down-sampled signal (which may be referred to as the first down-sampled signal), wherein one first output signal corresponds to one first Downsampling signal, assuming that the sub-band filter bank includes M bandpass filters, namely G ₀ (z), G ₁ (z), ..., G _M-1 (z), then there are M band-pass filters in total An output signal and M first downsampling signals. It should be noted that, in the embodiment of this application, the obtained M first downsampled signals are used as the secondary path adaptive filter

The input signal, the secondary path adaptive filter

The output signal obtained based on the input signal may be referred to as a second output signal, and M second output signals can be obtained accordingly, and the second output signal is used to adjust the error signal of the entire system.

In the above-mentioned embodiments of the present application, it has been specifically explained that in the case where the input signal of the sub-band filter bank is to play sound, the function of the output signal of the sub-band filter bank, that is, for adjusting the error signal of the entire system, has Realizability.

In a possible implementation of the first aspect, when the input signal of the subband filter bank is the error signal of the entire system, the sampling module is specifically configured to: filter each bandpass in the subband filter bank The signal (may be referred to as the third output signal) output by the device is down-sampled, for example, m times of down-sampling is performed to obtain a down-sampled signal (may be referred to as the second down-sampled signal), wherein one third output signal corresponds to one The second down-sampling signal, the obtained M second down-sampling signals are used as the secondary path adaptive filter

The input signal, the secondary path adaptive filter

The output signal obtained based on the input signal can be called the fourth output signal, and M fourth output signals can be obtained accordingly, and the fourth output signal is used to adjust the secondary path adaptive filter

The filter coefficients of , that is, the coefficients used to estimate the secondary path.

In the above embodiments of the present application, in the case that the input signal of the subband filter bank is the error signal of the whole system, the function of the output signal of the subband filter bank, that is, for estimating the secondary channel coefficient, is specifically explained. It is achievable.

In a possible implementation of the first aspect, when the input signal of the sub-band filter bank is the up-sampled signal output by the sampling module, the error signal of the whole system is based on each band in the sub-band filter bank The signal output by the pass filter (which may be referred to as the fifth output signal) and the third signal are calculated, wherein the third signal is a combination of the first signal (i.e. external noise) and the second signal (i.e. target playback sound and primary The output signal of the channel adaptive filter W(z) after calculation) is a signal obtained by calculation.

In the above-mentioned embodiments of the present application, the error signal of the system is not only related to the external noise, but also related to the real-time playing target sound. Online modeling improves user comfort.

In a possible implementation of the first aspect, the adjustment method of the filter coefficients of each band-pass filter in the sub-band filter bank may be an ant colony algorithm, or a gradient update method. No limit.

In the above implementation manners of the present application, there may be multiple implementation manners for adjusting the filter coefficients of each bandpass filter in the subband filter bank, which have flexibility and wide applicability.

In the second aspect, the embodiment of the present application also provides an active noise reduction method based on the FxLMS structure, which is applied to an active noise reduction device. The method may include: first, the active noise reduction device obtains a first signal through a primary path, and the first The signal is used to represent the ambient sound at the reference microphone of the active noise reduction device, and the primary path represents a physical path through which the first signal is transmitted to the speaker of the active noise reduction device, which may also be called an air path. In addition, the active noise reduction device also needs to obtain the estimated signal corresponding to the first signal through the secondary path adaptive filter, that is, in this case, the first signal is used as the input signal of the secondary path adaptive filter, and the estimated signal As the output signal of the secondary path adaptive filter, the secondary path adaptive filter is used to simulate the secondary path, and the filter coefficients of the secondary path adaptive filter (ie, the secondary path coefficients) are determined by the The band filter bank is adjusted based on the target playing sound (eg, playing music signal). After the active noise reduction device obtains the first signal and the estimated signal, it will further obtain the second signal through the secondary path, wherein the secondary path represents the circuit path through which the second signal is transmitted to the speaker of the active noise reduction device, and the second The signal is the output signal of the primary path adaptive filter and the signal after the operation of the target playback sound. The primary path adaptive filter is used to simulate the primary path, and the feedback signal of the primary path adaptive filter is obtained from the error signal and the estimated signal. After obtaining the first signal and the second signal, the active noise reduction device performs calculations on the first signal and the second signal to obtain a third signal, and finally, the third signal is played through a speaker of the active noise reduction device.

In the above-mentioned embodiments of the present application, on the basis of the active noise reduction system based on the FxLMS structure, the process of the active noise reduction method based on the FxLMS structure is described in detail. The method uses a sub-band filter bank to play the target Sound white noise, eliminating its correlation to obtain real-time filter coefficients of the secondary channel adaptive filter, compared with the offline method of using white noise for secondary channel modeling, directly using the target playback sound to improve real-time online modeling It improves user comfort, is more practical, and can be adjusted dynamically in real time.

In a possible implementation manner of the second aspect, the subband filter bank includes M bandpass filters, M≥1, and the method may further include: for each bandpass filter in the subband filter bank The output signal is down-sampled to obtain a down-sampled signal. Assuming that there are M band-pass filters, a total of M down-sampled signals are obtained, and the obtained M down-sampled signals are used as the input signal of the secondary channel adaptive filter. And/or, upsampling the output signal of the secondary path adaptive filter to obtain an upsampling signal, one upsampling signal corresponds to one downsampling signal, assuming that there are M bandpass filters, there are also M downsampling signal and M upsampled signals.

In the above-mentioned embodiments of the present application, there are two purposes of downsampling: 1. To widen the specific frequency bands passed by each bandpass filter to achieve white noise. The white noise signal is used to adjust the secondary channel The coefficient of the adaptive filter; 2. Reduce the calculation amount of the secondary path adaptive filter. The purpose of upsampling is the opposite operation to downsampling, in order to restore the broadened frequency band to facilitate subsequent calculations.

In a possible implementation manner of the second aspect, the input signals of the subband filter bank may include the following types: the target playing sound, the up-sampling signal output by the sampling module, and the system error signal.

In a possible implementation of the second aspect, when the input signal of the sub-band filter bank is the target playback sound, the output signal of each bandpass filter in the sub-band filter bank is down-sampled , to obtain the downsampling signal is specifically: downsampling the signal output by each bandpass filter in the subband filter bank (which may be referred to as the first output signal), for example, performing m times downsampling to obtain the downsampling signal (may be referred to as the first down-sampling signal), wherein one first output signal corresponds to one first down-sampling signal, assuming that the sub-band filter bank includes M band-pass filters, respectively G ₀ (z), G ₁ (z), . . . , G _M-1 (z), there are M first output signals and M first downsampling signals in total. It should be noted that in the embodiment of the present application, the obtained M first downsampled signals are used as input signals of the secondary path adaptive filter, and the output signal obtained by the secondary path adaptive filter based on the input signals may be It is called the second output signal, and M second output signals can be obtained accordingly, and the second output signals are used to adjust the error signal of the entire system.

In a possible implementation of the second aspect, when the input signal of the subband filter bank is the error signal of the entire system, the output signal of each bandpass filter in the subband filter bank is Downsampling to obtain the downsampling signal is specifically: downsampling the signal (which may be referred to as the third output signal) output by each bandpass filter in the subband filter bank, for example, performing m times downsampling to obtain the downsampling The sampling signal (may be referred to as the second downsampling signal), wherein one third output signal corresponds to one second downsampling signal, and the obtained M second downsampling signals are used as the input signal of the secondary path adaptive filter, The output signal obtained by the secondary path adaptive filter based on the input signal can be called the fourth output signal, and M fourth output signals can be obtained accordingly, and the fourth output signal is used to adjust the secondary path adaptive filter The filter coefficients of the filter, that is, the coefficients used to estimate the secondary path.

In a possible implementation of the second aspect, when the input signal of the sub-band filter bank is the up-sampled signal output by the sampling module, the error signal of the whole system is based on each band in the sub-band filter bank The signal output by the pass filter (which may be referred to as the fifth output signal) and the third signal are calculated, wherein the third signal is a combination of the first signal (i.e. external noise) and the second signal (i.e. target playback sound and primary The signal obtained by calculating the output signal of the channel adaptive filter).

In a possible implementation of the second aspect, the adjustment method of the filter coefficients of each bandpass filter in the subband filter bank may be an ant colony algorithm, or a gradient update method. No limit.

A third aspect of the embodiments of the present application provides an active noise reduction device, and the device has a function of implementing the method of the second aspect or any possible implementation manner of the second aspect. This function may be implemented by hardware, or may be implemented by executing corresponding software on the hardware. The hardware or software includes one or more modules corresponding to the above functions.

The fourth aspect of the embodiment of the present application provides an active noise reduction device, which may include a memory, a processor, and a bus system, wherein the memory is used to store programs, and the processor is used to call the programs stored in the memory to execute the first embodiment of the present application. The second aspect or any one of the possible implementation methods of the second aspect.

The fifth aspect of the embodiment of the present application provides a computer-readable storage medium, the computer-readable storage medium stores instructions, and when it is run on a computer, the computer can execute any one of the above-mentioned second aspect or the second aspect. method of possible implementation.

The sixth aspect of the embodiments of the present application provides a computer program, which, when running on a computer, causes the computer to execute the method of the above-mentioned second aspect or any possible implementation manner of the second aspect.

The seventh aspect of the embodiment of the present application provides a chip, the chip includes at least one processor and at least one interface circuit, the interface circuit is coupled to the processor, and the at least one interface circuit is used to perform the function of sending and receiving, and send instructions to At least one processor, at least one processor is used to run computer programs or instructions, which has the function of realizing the method of the second aspect or any possible implementation mode of the second aspect above, and this function can be realized by hardware or by software Realization can also be achieved through a combination of hardware and software, where the hardware or software includes one or more modules corresponding to the above functions. In addition, the interface circuit is used to communicate with other modules outside the chip.

Description of drawings

Fig. 1 is a schematic diagram of the block diagram of the current mainstream FxLMS structure;

FIG. 2 is a schematic diagram of the architecture of the active noise reduction device provided by the embodiment of the present application;

FIG. 3 is a schematic diagram of an active noise reduction device that can deploy an active noise reduction system based on the FxLMS structure provided by the embodiment of the present application;

FIG. 4 is another schematic diagram of an active noise reduction device that can deploy an active noise reduction system based on the FxLMS structure provided by the embodiment of the present application;

FIG. 5 is a schematic block diagram of an active noise reduction device provided by an embodiment of the present application;

FIG. 6 is a schematic diagram of an active noise reduction system based on the FxLMS structure provided by the embodiment of the present application;

FIG. 7 is a schematic diagram of the primary channel adaptive filter W(z) provided by the embodiment of the present application;

FIG. 8 is a schematic diagram of the principle of an adaptive filter for estimating primary channel coefficients provided by an embodiment of the present application;

Figure 9 shows the existing secondary path adaptive filter

A block diagram of;

Figure 10 is the secondary path adaptive filter provided by the embodiment of the present application

A structure diagram of

FIG. 11 is a schematic diagram of a subband filter bank adjustment method provided in an embodiment of the present application;

Fig. 12 is a schematic flow chart of the ant colony algorithm provided by the embodiment of the present application;

FIG. 13 is a schematic flowchart of an active noise reduction method based on the FxLMS structure provided by the embodiment of the present application;

FIG. 14 is a comparison diagram of a noise reduction result between the active noise reduction method provided by the embodiment of the present application and the traditional FxLMS method;

Fig. 15 is a schematic diagram of the original signal adjusted and optimized by the ant colony algorithm;

Fig. 16 is a schematic diagram of a signal obtained after subband decomposition and synthesis without an adjustment filter;

Fig. 17 is a schematic diagram of a signal obtained after filter subband decomposition and synthesis;

FIG. 18 is a schematic diagram of a computer-readable storage medium provided by an embodiment of the present application;

FIG. 19 is a schematic structural diagram of an active noise reduction system based on an FxLMS structure provided by an embodiment of the present application.

Detailed ways

The terms "first", "second" and the like in the specification and claims of the present application and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific sequence or sequence. It should be understood that the terms used in this way can be interchanged under appropriate circumstances, and this is merely a description of the manner in which objects with the same attribute are described in the embodiments of the present application. Furthermore, the terms "comprising" and "having", as well as any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, product, or apparatus comprising a series of elements is not necessarily limited to those elements, but may include elements not expressly included. Other elements listed explicitly or inherent to the process, method, product, or apparatus.

The embodiment of the present application involves a lot of relevant knowledge about filters, noise reduction, etc. In order to better understand the solution of the embodiment of the present application, the following first introduces related terms and concepts that may be involved in the embodiment of the present application. It should be understood that the interpretation of related concepts may be limited due to the specific conditions of the embodiment of the application, but it does not mean that the application is limited to the specific conditions, and there may be differences in the specific conditions of different embodiments. Specifically, there is no limitation here.

(1) Adaptive filter (adaptive filter)

An adaptive filter refers to a filter that uses an adaptive algorithm to change the parameters and structure of the filter according to changes in the environment. In general, the structure of the adaptive filter is not changed. The coefficients of the adaptive filter are time-varying coefficients updated by the adaptive algorithm. That is, its coefficients are automatically and continuously adapted to a given signal to obtain the desired response. The most important feature of an adaptive filter is its ability to work effectively in unknown environments and to track the time-varying characteristics of the input signal.

The mathematical principle of the adaptive filter is: based on the estimation of the statistical characteristics of the input and output signals, a specific algorithm is adopted to automatically adjust the filter coefficients to achieve an algorithm or device with the best filtering characteristics. Adaptive filters can be either continuous or discrete. The discrete domain adaptive filter is composed of a group of tapped delay lines, variable weight coefficients and mechanisms for automatically adjusting coefficients. For each sample value of the input signal sequence x(n), the adaptive filter updates and adjusts the weighting coefficients according to a specific algorithm, so that the average value of the output signal sequence y(n) compared with the expected output signal sequence d(n) The square error is the minimum, that is, the output signal sequence y(n) is close to the expected signal sequence d(n).

Adaptive filters can be applied to automatic equalization, echo cancellation, antenna array beamforming in the communication field, and parameter identification, noise elimination, and spectrum estimation of signal processing in other related fields. For different applications, only the added input signal and the expected signal are different, but the basic principle is the same.

(2) filter bank

A filter bank is a set of filters that share a common input signal or have a common output signal. For example, assuming that the filter bank is a filter bank with a common input signal s(n), then after s(n) passes through this set of filters (assuming that there are M filters), the obtained y ₀ (n) , y ₁ (n), . . . , y _M-1 (n) are sub-band signals, and their frequency spectrums are ideally non-overlapping.

(3) Analysis filter bank

With a common input signal, the one that obtains M subband signals is called an analysis filter bank (assuming it includes M filters).

(4) Sub-band signal

Referring to frequency division multiplexing in the communication field, assuming that between 60kHZ and 72kHZ, there are three channels occupying the frequency ranges of 60-64kHZ, 64-68kHZ, and 68-72kHZ, then the frequencies occupied by these three channels The ranges are called subbands.

It should be noted here that, in this article, "microphone" refers to a device for collecting sound and converting it into a corresponding electrical signal. "Speaker" means a device used to convert electrical signals into sound based on audio data. The term "environmental sound" herein refers to the sound existing in the external environment where the active noise reduction device is located and collected by the active noise reduction device, which may be a combination of one or more sounds, such as speech, music, noise, etc. .

Embodiments of the present application are described below in conjunction with the accompanying drawings. Those of ordinary skill in the art know that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

First, the architecture of the active noise reduction device involved in the embodiment of the present application will be described. For details, please refer to FIG. 2. FIG. 2 is a schematic diagram of the architecture of the active noise reduction device provided by the embodiment of the present application. A microphone (referred to as a reference microphone), an error microphone (referred to as an error microphone), a speaker, and an active noise reduction system (that is, the ANC described in Figure 2, which is the active noise reduction system based on the FxLMS structure provided by the embodiment of the present application ), where the reference microphone is installed close to the sound source, and the error microphone is installed close to the human ear.

To further understand the active noise reduction device described in the embodiment of the present application, FIG. 3 shows a schematic diagram of an active noise reduction device 10 that can deploy the active noise reduction system based on the FxLMS structure provided by the embodiment of the present application. In one embodiment, the active noise reduction device 10 may be, for example, an audio playback device that is in contact with the ear, such as a true wireless stereo (true wireless stereo, TWS) earphone. The active noise reduction device 10 may include a pair of earphones, and the two

earphones

11 and 12 are configured substantially identically to each other. Therefore, only one earphone 11 is schematically described. The earphone 11 includes an external reference microphone 13, a processor 17 inside the earphone 11, a first ear-in or ear-contacting part inside the earphone 11 (relative to the external reference microphone 13 exposed to the environment). Loudspeaker 15 and residual microphone 14. The reference microphone 13 is configured to detect or collect sounds of the external environment.

Although a possible configuration of the active noise reduction device 10 is shown in FIG. 3 , this is illustrative only and does not limit the scope of the present disclosure. For example, in some embodiments, the two

earphones

11 and 12 may only have one processor 17, and wireless signals are transmitted through wireless transmission methods such as Bluetooth signal transmission to realize the pairing of the two

earphones

11 and 12 to a single processor. 17 shares. In another embodiment, the two

earphones

11 and 12 can also share a single reference microphone 13 .

In one embodiment, the external reference microphone 13 of the active noise reduction device 10 collects ambient sound, and performs acoustic-electric conversion to generate a continuous electrical signal and transmits it to the processor 17 . The processor 17 predicts or estimates the ambient sound at a subsequent time based on the received signal, and generates an inverse signal representing the ambient sound at the subsequent time and transmits it to the first speaker 15 . In this context, "inverted signal" refers to a signal that is manipulated after inverting the audio signal, eg by directly inverting the sign of the audio sample point or by further processing. In contrast, a signal that is not inverted may be referred to as a "normal phase signal". The out-of-phase signal played by the speaker is used to cancel the direct sound (in-phase sound) directly inside the active noise reduction device 10 to a certain extent, so as to reduce the sound perceived by the ear. The first loudspeaker 15 plays the anti-phase sound based on the received anti-phase signal, so as to offset the direct ambient sound directly from the environment to the ANC device 10 at a subsequent moment, so as to achieve the effect of noise reduction.

Although a schematic configuration of the active noise reduction device 10 is shown in FIG. 3 as a TWS earphone, it is understood that the scope of the present disclosure is not limited thereto. For example, another possible configuration of an active noise reduction device is illustrated in FIG. 4 as a headset, namely around-ear headphones. As shown in FIG. 4 , the active noise reduction device 20 may include a pair of ear cup parts, and the two ear cup parts are configured substantially identically to each other. Therefore, only one cap part is schematically depicted. The earmuff part includes an external first microphone 13, a second microphone 19 and a processor 17 inside the earmuff part. The ear cup portion also includes a first residual microphone 14, a second residual microphone 16, a first loudspeaker 15 and a second residual microphone 14 located inside the ear cup portion (relative to the first microphone 13 and the second microphone 19 exposed to the environment). Two loudspeakers 18 . Both the first microphone 13 and the second microphone 19 are configured to detect or collect the sound of the external environment, and the first microphone 13 and the second microphone 19 may operate simultaneously or alternately and may collect the same or different sounds. In one embodiment, the first microphone 13 may have an internal first filter to only collect the sound of the first frequency, and the second microphone 19 may have an internal second filter to only collect the sound of the second frequency. For example, the first frequency is a low frequency and the second frequency is a mid to high frequency. By capturing sounds for different frequencies, more ambient sound details can be obtained for better sound estimation, resulting in better noise reduction width and depth.

In one embodiment, the external first microphone 13 and second microphone 19 of the active noise reduction device 20 collect ambient sound, and perform acoustic-electric conversion to generate a continuous electrical signal and transmit it to the processor 17 . The processor 17 predicts or estimates the ambient sound at a subsequent time based on the received signal, and generates an inverse signal representing the ambient sound at a subsequent time and transmits it to the built-in first speaker 15 and second speaker 18 . The first speaker 15 and the second speaker 18 play the anti-phase sound based on the received anti-phase signal to offset the direct ambient sound from the environment directly to the active noise reduction device 20 at a subsequent moment, so as to achieve the effect of noise reduction.

Alternatively, besides the active noise reduction device 10 and the active noise reduction device 20 as described above, the active noise reduction device can also be other types of active noise reduction devices that transmit audio through bone conduction. The application does not give an example.

In addition, FIG. 5 also shows a schematic block diagram of the active noise reduction device provided by the present application. It should be understood that the active noise reduction device 100 shown in FIG. 5 is only exemplary, for example, to illustrate a possible implementation of the active noise reduction device 10 in FIG. A possible implementation of the noise reduction device 20 should not constitute any limitation on the functions and scope of the implementation described in this application. In one embodiment, the active noise reduction device 100 may include a processor 110, a wireless communication module 160, an antenna 1, an audio module 170, a speaker module 170A, a microphone module 170C, buttons 190, An internal memory 121 , a universal serial bus (universal serial bus, USB) interface 130 , a charging management module 140 , and a power management module 141 .

The processor 110 may be, for example, the processor 17 in FIG. 3 , and may include one or more processing units. For example, the processor 110 may include an application processor (application processor, AP), a modem processor, a graphics processor ( graphics processing unit (GPU), image signal processor (image signal processor, ISP), controller, video codec, digital signal processor (digital signal processor, DSP), baseband processor and/or neural network processor ( neural-network processing unit, NPU), etc. In some embodiments, the different processing units may be separate devices. In other embodiments, different processing units can also be integrated in one or more processors. The controller can generate an operation control signal according to the instruction opcode and timing signal, and complete the control of fetching and executing the instruction.

A memory may also be provided in the processor 110 for storing instructions and data. The internal memory 121 may be used to store computer-executable program codes including instructions. The internal memory 121 may include an area for storing programs and an area for storing data. The processor 110 executes various functional applications and data processing of the active noise reduction device 100 by executing instructions stored in the internal memory 121 and/or instructions stored in a memory provided in the processor.

The active noise reduction device 100 may implement audio functions through the audio module 170 , the speaker module 170A, the microphone module 170C, an application processor, and the like. Such as music playback, recording, etc. The audio module 170 is used to convert digital audio information into analog audio signal output, and is also used to convert analog audio input into digital audio signal. The audio module 170 may also be used to encode and decode audio signals. In some embodiments, the audio module 170 may be set in the processor 110 , or some functional modules of the audio module 170 may be set in the processor 110 .

The speaker module 170A, also referred to as a "horn", is used to convert audio electrical signals into sound signals. The active noise reduction device 100 can listen to music through the speaker module 170A, or listen to a hands-free call. The microphone module 170C is also called "microphone" or "microphone". When making a call or sending a voice message, the user can approach the microphone module 170C to make a sound through the mouth, and input the sound signal to the microphone module 170C.

In some other embodiments, the active noise reduction device 100 may further include an antenna 2 and a mobile communication module 150 . In addition to the above components, the active noise reduction device 100 may also include an external memory interface 120, a battery 142, a receiver 170B, an earphone interface 170D, a sensor module 180, a motor 191, an indicator 192, and a camera 193, which are shown in dotted lines and dotted lines. , a display screen 194, and one or more of a subscriber identification module (subscriber identification module, SIM) card interface 195. The sensor module 180 may include a pressure sensor 180A, a gyro sensor 180B, an air pressure sensor 180C, a magnetic sensor 180D, an acceleration sensor 180E, a distance sensor 180F, a proximity light sensor 180G, a fingerprint sensor 180H, a temperature sensor 180J, a touch sensor 180K, an ambient light sensor One or more of 180L and bone conduction sensor 180M. The sensor module 180 may also include other types of sensors not listed.

It can be understood that the structure illustrated in the embodiment of the present invention does not constitute a specific limitation on the active noise reduction device 100 . In other embodiments of the present application, the active noise reduction device 100 may include more or fewer components than shown in the figure, or combine some components, or separate some components, or arrange different components. The illustrated components can be realized in hardware, software or a combination of software and hardware.

On the basis of the above-mentioned active noise reduction equipment, the active noise reduction system based on the FxLMS structure provided by the embodiment of the present application will be described next. Please refer to Figure 6 for details. A schematic diagram of the active noise reduction system of , which is applied to active noise reduction equipment, and may specifically include: primary path adaptive filter W(z)601, secondary path adaptive filter

602 and a subband filter bank 603, wherein the subband filter bank 603 includes M bandpass filters, and different bandpass filters have different passbands, which can be called G ₀ (z), G ₁ ( z), ..., G _M-1 (z), M≥1.

Specifically, the primary path adaptive filter W(z) 601 is used to simulate the primary path, where the primary path represents the external ambient sound (which may be called the first signal) at the reference microphone of the active noise reduction device is transmitted to the active noise reduction device. The physical path passed by the loudspeaker of the noisy device, so the primary path can also be called the air path; the secondary path adaptive filter

602, used to simulate the secondary path, wherein the secondary path represents the circuit path through which the second signal is transmitted to the speaker, and the second signal is the output signal of the primary path adaptive filter W(z) 601 and the target playback sound ( For example, the calculated signal of the music played by the earphones, for example, the calculation can be the addition of the output signal of the primary path adaptive filter W(z) 601 and the target playback sound; the subband filter bank 603 is used to Target Playback Sound Adjustment Secondary Path Adaptive Filter

602 filter coefficients.

It should be noted that, in some embodiments of the present application, the active noise reduction system may further include a sampling module 604, and the sampling module 604 may perform up-sampling or down-sampling on the signal, specifically, it may be used for:

(1) Downsampling the output signal of each bandpass filter in the subband filter bank 603 .

The output signal of each band-pass filter in the sub-band filter bank 603 is down-sampled to obtain a down-sampled signal. Assuming that there are M band-pass filters, M down-sampled signals are obtained in total, and the obtained M down-sampled signals The signal is then used as a secondary path adaptive filter

602 input signal.

In the embodiment of the present application, there are two purposes of downsampling: 1. Broaden the specific frequency band passed by each bandpass filter to realize white noise. The white noise signal is used to adjust the secondary channel self- adaptive filter

coefficient; 2, reduce the secondary path adaptive filter

calculation amount.

(2) Adaptive filter for the secondary path

The output signal of 602 is up-sampled.

Adaptive filter for secondary path

The output signal of 602 is up-sampled to obtain an up-sampled signal. One up-sampled signal corresponds to one down-sampled signal. Assuming that there are M band-pass filters, there are M down-sampled signals and M up-sampled signals.

In order to facilitate the understanding of the active noise reduction system based on the FxLMS structure provided by the embodiment of the present application, the primary path adaptive filter W(z) and the secondary path adaptive filter included in the active noise reduction system are as follows:

The subband filter bank is described in detail:

1. Primary path adaptive filter W(z)

Please refer to FIG. 7 for details. FIG. 7 is a schematic diagram of the primary path adaptive filter W(z) provided by the embodiment of the present application. The primary path adaptive filter W(z) can be a FIR filter (other filtering The device can also be used, this application is not limited to this, it is only shown here), the role is to characterize the estimated real primary path P(z) (including the passive noise reduction material of the active noise reduction device and the connection between the earmuffs and the speaker The air passage between them) for subsequent noise reduction steps, the order can be set to 128 (other orders are also possible, this application does not limit this, here is only for illustration), assuming the actual primary passage order At about tens of orders, the filter order that is redundantly set during estimation will approach 0 after operation. The primary path adaptive filter W(z) is an adaptive filter based on the LMS algorithm, and the LMS adaptive filter The schematic diagram of the circuit breaker is shown in Figure 7, x(n) is the input signal (that is, the first signal mentioned above, that is, external noise), and y(n) is the output result of the primary path adaptive filter W(z) , used for subsequent noise reduction operations, d(n) is the output signal of the real primary channel, that is, the signal to be denoised, and e(n) is the error signal, used to adjust the filter coefficients, that is, the primary channel from The filter coefficients of the adaptive filter W(z) can be updated according to the following formula (1), and repeated iterations make the final error signal e(n) as small as possible.

y(n)=x(n)*w(n)

e(n)=y(n)-d(n) (1)

w(n+1)=w(n)+μe(n)x(n)

Among them, w(n) is the filter coefficient of the primary channel adaptive filter W(z) of the nth round, and w(n+1) is the filter coefficient of the primary channel adaptive filter W(z) of the n+1th round Filter coefficient, μ is the step size coefficient.

It should be noted that, in the embodiment of this application, since the signal after passing through the noise reduction filter W(z) will pass through the secondary channel S(z) and then be played out through the speaker, if the influence of this part is not considered, it may It will cause the adjusted filter coefficient of the primary path adaptive filter W(z) (also known as the primary path coefficient) to obtain a reverse sound wave amplitude and phase that are different from the value of the external noise after reaching the speaker through the primary path , which affects the cancellation effect, so in order to counteract the interference effect of the secondary channel on noise reduction, the estimated secondary channel adaptive filter

The filter coefficients (also referred to as secondary path coefficients) are placed in the primary path adaptive filter W(z) part, specifically as shown in Figure 8, Figure 8 is the self-assessment of primary path coefficients provided by the embodiment of the present application A schematic diagram of the adaptive filter principle. At this time, the update formula of the primary channel adaptive filter W(z) is shown in the following formula (2):

y(n)=x(n)*w(n)

e(n)=y(n)*s(n)-d(n) (2)

in,

secondary path adaptive filter for the nth round

s(n) is the coefficient of the actual secondary path (i.e. electrical path), w(n) is the filter coefficient of the nth round primary path adaptive filter W(z), w(n +1) is the filter coefficient of the primary path adaptive filter W(z) of the n+1th round, and μ is the step size coefficient.

Second, the secondary path adaptive filter

In order to facilitate the understanding of the secondary path adaptive filter described in the embodiment of the present application

First introduce the block diagram of the known secondary path adaptive filter, please refer to Figure 9 for details, and Figure 9 shows the existing secondary path adaptive filter

The block diagram, under the block diagram shown in Figure 9, the secondary path adaptive filter

The filter coefficients of are adjusted using the following formula (3):

e ₁ (n)=d(n)+y(n)*s(n)

Among them, y(n) is the output signal of the adaptive filter W(z) in the primary channel, d(n) is the signal to be de-noised obtained by passing the external noise through the primary channel, and S(z) is the secondary channel (i.e. path),

is the secondary path adaptive filter moved to the primary path, e ₁ (n) is the residual error signal received by the loudspeaker, e ₂ (n) the overall filter coefficient adjustment signal (ie error signal), music(n) Play the sound for the target, μ is the step factor,

secondary path adaptive filter for the nth round

filter coefficients,

Secondary path adaptive filter for round n+1

The filter coefficients of s(n) are the coefficients of the actual secondary path (ie, the electrical path).

Because the basic framework of the FxLMS structure is used to construct the active noise reduction system, before the input signal passes through the main filter (that is, the primary path adaptive filter W(z)), it must first pass through the moved secondary path adaptive filter

Since the main filter and the secondary path adaptive filter

The adjustment process will affect each other, which will greatly affect the accuracy of the adjustment. However, both the secondary pathway and the primary pathway exist objectively, and both require adaptive estimation. The current general practice is to set the secondary path adaptive filter

The filter coefficients (i.e., the secondary channel coefficients) are known, and then only the main filter is adjusted, or a period of white noise is passed through the system to obtain the secondary channel coefficients, and then the secondary channel coefficients are kept fixed, or when When the stage path changes, a section of white noise is passed into the system to model again. Considering that the secondary path in the actual system is time-varying, the constant setting will affect the active noise reduction effect, so online modeling is required.

While modeling the secondary path requires a suitable modeling signal, the best modeling signal is white noise, because when modeling as a full-band signal, the applicability of the secondary path to signals of all frequencies that are allowed to pass can be fully considered sex. However, in actual use, since the modeling signal will be played by the speaker and received by the microphone, considering the comfort of the user experience, the actually used modeling signal is the music signal played by the user. However, music signals are characterized by rich harmonics, high energy at harmonics, and low energy at non-harmonics, and the adjustment principle of the adaptive filter is that the selection of the step size during adjustment is related to the characteristic value of the input signal. The correlation of the signal is strong, the eigenvalue is close to zero, and the step size needs to be set small, which seriously affects the convergence speed of the adaptive filter. In addition, according to the principle of LMS, the update of the filter coefficients of the two adaptive filters is adjusted by using the error signal, and due to the reverse in-phase sound wave cancellation of the speaker playback and external noise, when the cancellation effect is good The residual noise obtained by referring to the microphone is the error used to update the filter coefficients, and the overall principle of the system is to reduce the error signal by adjusting the adaptive filter coefficients. According to the principle of LMS algorithm, the secondary path adaptive filter

The update formula of is shown in the following formula (4):

Among them, e(n) is the error signal, x(ni) is the input music signal, the expected gradient is simplified, and the adaptive filter update function can be obtained

The music signal and the error signal in practical applications are not completely irrelevant, so using the simplified update function will cause the filter coefficients of the adaptive filter to be updated inaccurately.

To sum up, the active noise reduction system based on the FxLMS structure is not suitable for direct modeling with music signals. Therefore, in the embodiment of this application, the music signal is whitened by sub-band decomposition, and its correlation is eliminated to better The secondary path modeling is carried out, so in the embodiment of this application, the secondary path adaptive filter

The actual structure adopted is shown in Figure 10, and Figure 10 is the secondary path adaptive filter provided by the embodiment of the present application

and the subband filter bank (G ₀ (z), G ₁ (z), ..., G _M-1 (z)), the structural block diagram between the sampling module, in Fig. 10, the subband filter bank ( G ₀ (z), G ₁ (z), ..., G _M-1 (z)) can decompose the input broadband signal into corresponding frequency bands; and in order to achieve white noise, the embodiment of the present application adopts The module down-samples the output signal of each band-pass filter in the sub-band filter bank, that is, to widen the specific frequency band passed by each band-pass filter to achieve white noise. The white-noise signal is used for Tuning Secondary Path Adaptive Filters

The coefficient; at the same time, the downsampling operation can also reduce the secondary path adaptive filter

calculation amount. After the above sub-band decomposition and down-sampling operations, the signals on the respective frequency bands can be considered to be similar to white noise (white noise is not directly used for synchronization, because what the user listens to is the target playback sound, directly playing white noise will cause user experience Not good), and then pass the signals of each frequency band through the same secondary channel adaptive filter in sequence

The filtered signals of each frequency band are respectively obtained, and then the sampling module performs up-sampling to restore the original sampling rate, and then passes through the same band-pass filter and sums to restore the decomposed frequency bands. Theoretically, the result is The target playback sound after the adaptive filtering of the simulated secondary channel coefficients is calculated with e ₁ (n) to obtain the error signal e ₂ (n) used to adjust the whole system, because it is used to pass through the secondary channel adaptive filter

The target playback sound is decomposed and downsampled, so the error e ₂ (n) used to adjust the filter should also be operated in the same way.

It should be noted that, as can be seen from the system block diagram in FIG. 10 , the input signals of the sub-band filter bank may include the following types: target playback sound, up-sampling signal output by the sampling module, and system error signal. The following is an explanation based on different situations:

(1) When the input signal of the subband filter bank is the target playback sound

In the case that the input signal of the sub-band filter bank is the target playing sound, the sampling module is used for: down-sampling the output signal (which may be referred to as the first output signal) of each band-pass filter in the sub-band filter bank Sampling, for example, performing m-fold downsampling to obtain a downsampled signal (which may be referred to as a first downsampled signal), wherein one first output signal corresponds to one first downsampled signal, assuming that the subband filter bank includes a band There are M pass filters, respectively G ₀ (z), G ₁ (z), ..., G _M-1 (z), and there are M first output signals and M first downsampling signals in total. It should be noted that, in the embodiment of this application, the obtained M first downsampled signals are used as the secondary path adaptive filter

The input signal, the secondary path adaptive filter

The output signal obtained based on the input signal can be called the second output signal, and M second output signals can be obtained accordingly, and the second output signal is used to adjust the error signal e ₂ (n) of the whole system.

(2) When the input signal of the subband filter bank is the upsampling signal output by the sampling module

In the case that the input signal of the sub-band filter bank is the error signal e ₂ (n) of the whole system, the sampling module is used to: output the signal of each band-pass filter in the sub-band filter bank (which can be referred to as The third output signal) is down-sampled, for example, down-sampled by m times to obtain a down-sampled signal (which may be referred to as a second down-sampled signal), wherein a third output signal corresponds to a second down-sampled signal, and the obtained The M second downsampled signals are used as the secondary path adaptive filter

The input signal, the secondary path adaptive filter

(3) The case where the input signal of the subband filter bank is the error signal of the system

In the case that the input signal of the subband filter bank is the upsampling signal output by the sampling module, the error signal e ₂ (n) of the whole system is based on the signal output by each bandpass filter in the subband filter bank (which can be It is called the fifth output signal) and the third signal to obtain, wherein, the third signal is the first signal (i.e. external noise) and the second signal (i.e. the target playing sound and the primary path adaptive filter W(z ) The output signal of the calculated signal) The signal obtained by performing the calculation.

In the implementation of the above embodiments in this application, the active noise reduction system can simultaneously adjust the filter coefficients of the primary path adaptive filter W(z) and the secondary path adaptive filter

In addition, compared with using white noise for secondary channel modeling, directly using the target playback sound modeling to be listened to by the user improves user comfort and is more practical.

(4) Subband filter bank

In the embodiment of the present application, the subband filter bank divides the frequency band to be processed into equal intervals through M bandpass filter banks G ₀ (z), G ₁ (z), ..., G _M-1 (z) For small frequency bands, M≥1, the initial value of the filter is a physically realizable filter bank generated by the formula, but the overlapping process between the filter banks is not flat at the junction of the frequency and the frequency by using a mathematical method. That is, there is a loss of energy at the frequency junction. That is, after the target playback sound is decomposed and synthesized by the sub-band filter bank, a certain distortion is generated. In order to prevent the target playback sound from being distorted after sub-band decomposition/synthesis, a certain adjustment of the sub-band filter bank is considered. The block diagram of the adjustment method can be shown in Figure 11. Based on the block diagram in Figure 11, the following formula (5) can be used for fine-tuning:

err(n)=x(n)-out(n) (5)

Among them, x(n) is the input of the standard target playing sound (such as music) (at this time, x(n) is different from the external noise signal above), err(n) and out(n) are fine-tuned The difference of the target playback sound obtained after the decomposition and synthesis of the band filter bank is used to compare with the err(n) obtained after the decomposition and synthesis of the sub-band filter bank without fine-tuning. The comparison method can be based on the following formula (6):

Among them, the range of w is

In the embodiment of this application, the basic idea of applying the ant colony algorithm to solve the optimization problem is as follows: the walking path of the ants represents the feasible solution of the problem to be optimized, and all the paths of the entire ant colony constitute the solution space of the problem to be optimized, and the path is relatively short The amount of pheromone released by the ants is more, as time goes on, the concentration of pheromone accumulated on the shorter path gradually increases, and the number of ants who choose this path is also increasing. In the end, the entire ants will concentrate on the best path under the action of positive feedback, which corresponds to the optimal solution of the problem to be optimized. The specific algorithm flow can be shown in Figure 12, where one ant is used to represent a feasible solution, and the information contained in one ant includes the values of various variables. First, preset the iteration period, and second, determine the number of ants. The specific process is as follows: ① randomly initialize the ant colony according to the existing excellent initial value, and record the optimal solution in the ant colony; ② enter the ant colony cycle iteration, and divide the initialized ants into two types, as follows:

a. One is the optimal solution in the ant colony last time. Search near the optimal solution. Search up and down according to the following formula. After searching, calculate whether the value of the cost function is reduced, reduce retention, increase discarding, each cycle The post-step coefficient w decreases, which can be expressed by the following formula (7):

X＝{x ₁ ,x ₂ ,...,x _n }

x _i = x _i +w*L (7)

x _i = x _i -w*L

Among them, x _i is the position of ant i, n is the number of ants, w is the update step size, and L is the fixed update length.

b. The other is a non-optimal solution. Random search moves, and there is a certain probability to optimize to the optimal solution. The optimization probability is obtained by calculating the pheromone concentration of the optimal ant and the pheromone concentration of the current ant, which is used to update the pheromone, which can be used The following formula (8) expresses:

Among them, mess _best is the pheromone concentration of the optimal solution ant, and mess _i is the pheromone concentration of the current ant.

Moving to the optimal solution is: x _i = _xi +u*(x _best _-xi );

The random search movement is: x _i = _xi +dx*rand(-1,1);

The update pheromone is: mess _i =(1-p)*mess _i +k*a ^-f(X) .

Among them, p is the pheromone decay factor, f(X) is the cost function, x _i is the position of ant i, dx is the update length, mess _i is the pheromone concentration of the current ant, u is the non-optimal solution step coefficient, x _best is the optimal solution obtained in the previous step, k is the pheromone update coefficient, and a is the update speed.

It should be noted that, in some implementations of the present application, the adjustment method of the filter coefficients of each bandpass filter in the subband filter bank can adopt other methods besides the above-mentioned ant colony algorithm, For example, the gradient update method, which is not limited in this application.

It should also be noted that, in other embodiments of the present application, in addition to the method based on subband decomposition, the whitening of the target playback sound can also be performed by using a whitening filter, or by using statistical The Bayesian method performs white noise, which is not limited in this application.

It should also be noted that, based on the ANC system based on the FxLMS structure described in the above-mentioned embodiments corresponding to FIG. 6 to FIG. 13. FIG. 13 is a schematic flowchart of an active noise reduction method based on the FxLMS structure provided by the embodiment of the present application. The method is applied to an active noise reduction device, and may specifically include the following steps:

1301. Acquire a first signal through a primary path, where the first signal represents ambient sound at a reference microphone of an active noise reduction device.

First, the active noise reduction device obtains a first signal through a primary path, the first signal is used to represent the ambient sound at the reference microphone of the active noise reduction device, and the primary path represents the location where the first signal is transmitted to the speaker of the active noise reduction device The physical path, also known as the air path.

1302. Obtain an estimated signal corresponding to the first signal through a secondary path adaptive filter, the secondary path adaptive filter is used to simulate the secondary path, and the filter coefficients of the secondary path adaptive filter are determined by the subband filter bank Play sound adjustments based on the target.

In addition, the active noise reduction device also needs to obtain the estimated signal corresponding to the first signal through the secondary path adaptive filter, that is, in this case, the first signal is used as the input signal of the secondary path adaptive filter, and the estimated signal As the output signal of the secondary path adaptive filter, the secondary path adaptive filter is used to simulate the secondary path, and the filter coefficients of the secondary path adaptive filter (ie, the secondary path coefficients) are determined by the The band filter bank is adjusted based on the target playing sound (eg, playing music signal).

It should be noted that, in this embodiment of the application, there is no sequential relationship between step 1301 and step 1302. Step 1301 can be executed first, and then step 1302 can be executed, or step 1302 can be executed first, and then step 1301 can be executed. It is also possible to execute step 1301 and step 1302 at the same time, which is not limited in this application.

1303. Obtain the second signal through the secondary path, where the secondary path indicates the circuit path through which the second signal is transmitted to the speaker, and the second signal is the signal obtained by calculating the output signal of the adaptive filter of the primary path and the target playback sound, The primary path adaptive filter is used to simulate the primary path, and the feedback signal of the primary path adaptive filter is obtained from the error signal and the estimated signal.

After the active noise reduction device obtains the first signal and the estimated signal, it will further obtain the second signal through the secondary path, wherein the secondary path represents the circuit path through which the second signal is transmitted to the speaker of the active noise reduction device, and the second The signal is the output signal of the primary path adaptive filter and the signal after the operation of the target playback sound. The primary path adaptive filter is used to simulate the primary path, and the feedback signal of the primary path adaptive filter is obtained from the error signal and the estimated signal.

1304. Acquire a third signal through a loudspeaker, where the third signal is a signal obtained by performing operations on the first signal and the second signal.

After obtaining the first signal and the second signal, the active noise reduction device performs calculations on the first signal and the second signal to obtain a third signal, and finally, the third signal is played through a speaker of the active noise reduction device.

It should be noted that, in some embodiments of the present application, the subband filter bank includes M bandpass filters, M≥1, and the method may further include: filtering each bandpass filter in the subband filter bank The output signal of the filter is down-sampled to obtain a down-sampled signal. Assuming that there are M band-pass filters, a total of M down-sampled signals are obtained, and the obtained M down-sampled signals are used as the input signal of the secondary channel adaptive filter . In the embodiment of the present application, there are two purposes of downsampling: 1. Broaden the specific frequency band passed by each bandpass filter to realize white noise, and the white noise signal is used to adjust the secondary channel self- Coefficients of the adaptive filter; 2. Reduce the calculation amount of the secondary path adaptive filter. And/or, upsampling the output signal of the secondary path adaptive filter to obtain an upsampling signal, one upsampling signal corresponds to one downsampling signal, assuming that there are M bandpass filters, there are also M downsampling signal and M upsampled signals.

It should also be noted that, in some embodiments of the present application, the input signals of the sub-band filter bank may include the following types: the target playback sound, the up-sampled signal output by the sampling module, and the system error signal. The following is an explanation based on different situations:

In the case that the input signal of the subband filter bank is the target playback sound, the output signal of each bandpass filter in the subband filter bank is down-sampled, and the down-sampling signal is obtained as follows: for the sub-band filter The signal output by each bandpass filter in the group (may be referred to as the first output signal) is down-sampled, for example, m times of down-sampling is performed to obtain a down-sampled signal (may be referred to as the first down-sampled signal), wherein, A first output signal corresponds to a first downsampling signal, assuming that the sub-band filter bank includes M bandpass filters, respectively G ₀ (z), G ₁ (z), ..., G _M-1 (z), then there are M first output signals and M first downsampled signals in total. It should be noted that in the embodiment of the present application, the obtained M first downsampled signals are used as input signals of the secondary path adaptive filter, and the output signal obtained by the secondary path adaptive filter based on the input signals may be It is called the second output signal, and M second output signals can be obtained accordingly, and the second output signals are used to adjust the error signal of the entire system.

In the case that the input signal of the sub-band filter bank is the error signal of the whole system, the output signal of each bandpass filter in the sub-band filter bank is down-sampled, and the down-sampled signal is obtained as follows: for the sub-band The signal output by each bandpass filter in the filter bank (may be referred to as the third output signal) is down-sampled, for example, m times of down-sampling is performed to obtain a down-sampled signal (may be called the second down-sampled signal), Wherein, a third output signal corresponds to a second downsampling signal, and the obtained M second downsampling signals are used as input signals of the secondary path adaptive filter, and the secondary path adaptive filter obtains the The output signal can be called the fourth output signal, and M fourth output signals can be obtained accordingly, and the fourth output signal is used to adjust the filter coefficient of the secondary path adaptive filter, that is, to estimate the secondary path access coefficient.

In the case where the input signal of the subband filter bank is the upsampled signal output by the sampling module, the error signal of the whole system is based on the signal output by each bandpass filter in the subband filter bank (which may be referred to as the fifth output signal) and the third signal are calculated, wherein the third signal is the first signal (i.e. the external noise) and the second signal (i.e. the target playback sound and the signal after the output signal of the primary path adaptive filter) The signal obtained by performing the operation.

In the implementation of the above embodiments in this application, the active noise reduction method based on the FxLMS structure can simultaneously adjust the filter coefficients of the primary path adaptive filter and the filter coefficients of the secondary path adaptive filter. In addition, compared with using white Noise is used for secondary channel modeling, and the target playback sound modeling is directly used by the user to listen to, which improves user comfort and is more practical.

It should also be noted that, in some implementations of the present application, the adjustment method of the filter coefficients of each bandpass filter in the subband filter bank may be an ant colony algorithm (for details, please refer to the corresponding embodiment in FIG. 12 above). The process described above), it may also be a gradient update method, which is not limited in this application.

In order to have a more intuitive understanding of the beneficial effects brought by the embodiments of the present application, the technical effects brought by the embodiments of the present application are further compared below. As an example, the sampling frequency of this application is 16KHz, there are 16 band-pass filters in the sub-band filter bank, the frequency range of each band-pass filter is 500Hz, and the sub-band filter is a 128-order low-pass filter . At this time, the primary path coefficient and the secondary path coefficient can be adjusted at the same time. The primary path adopts a step size of 0.001, and the secondary path adopts a step size of 0.1. The simulated real primary path is 41, and the simulated real secondary path is 41. The main adaptive filter used is 128th order and 16th order. Compared with the primary channel and secondary channel of the corresponding FxLMS structure, the results of noise reduction and music retention are as follows:

It can be seen from Figure 14 that the frequency boundary of babble noise and cafe noise is about 4000Hz. After calculation, the babble noise 0-1000Hz sub-band decomposition method can be reduced by 11dB on average, and the FxLMS method can be reduced by 10dB on average. The 1000-2000Hz sub-band decomposition method has an average reduction of 11dB, and the FxLMS method has an average reduction of 6dB. The 2000-3000H sub-band decomposition method has an average drop of 7dB, and the FxLMS method has an average drop of 5dB. The 3000-4000H sub-band decomposition method has an average drop of 7dB, and the FxLMS method has an average drop of 8dB.

Similarly, it can be calculated that the cafe noise 0-1000Hz sub-band decomposition method reduces by 10dB on average, and the FxLMS method reduces by 10dB on average; 8dB, the FxLMS method will drop 4dB on average; the 3000-4000H sub-band decomposition method will drop 9dB on average, and the FxLMS method will drop 3dB on average.

It can be seen that, except for the high frequency part around 3500HZ, the subband decomposition method has a certain noise reduction benefit compared with the traditional FxLMS method.

The original signal adjusted and optimized by the ant colony algorithm, that is, the time-domain waveform of white noise, is shown in Figure 15. The signal obtained after sub-band decomposition and synthesis without adjustment filter is shown in Figure 16, and the signal obtained after sub-band decomposition and synthesis of the filter is shown as Figure 17 shows.

After optimization, the signal similarity between the decomposed and synthesized signal and the original signal is 0.9582 (the maximum similarity is 1, that is, the same waveform). Using white noise for training, it is obvious that the time domain waveform obtained by the adjusted subband filter is closer to original signal.

Calculate the similarity to get: the similarity with the original signal after decomposition and synthesis before optimization is 0.8661, the formula is shown in the following formula (9):

Among them, r _x,out is the similarity between the input signal x and the output signal out, and x _i is the point-by-point value of the original signal

is the mean value of the original signal, out _i is the point-by-point value of the output signal after decomposition and synthesis,

is the average value of the output signal.

The above comparative tests all show that the active noise reduction system provided by the embodiment of the present application has good robustness and good working stability.

FIG. 18 is a schematic diagram of a computer-readable storage medium 900 provided by an embodiment of the present application. The computer-readable storage medium 900 is, for example, the cache memory in the processor 17 in FIG. 3 , the internal memory 121 in FIG. 5 , and the like. The computer-readable storage medium 900 stores one or more programs 902 . . . 906 configured to be executed by one or more processors of the active noise reduction device. One or more programs 902 . . . 906 may individually or collectively include instructions that are executable by processor 17 to implement the methods or processes described herein. It can be understood that the computer-readable storage medium 900 may also include programs for implementing other methods and steps.

FIG. 19 is a schematic structural diagram of an active noise reduction system 1900 based on the FxLMS structure provided by an embodiment of the present application. The active noise reduction system 1900 can be applied to active noise reduction equipment. The active noise reduction system 1900 includes: an acquisition module 1901, used to pass The primary path acquires a first signal, the first signal represents the ambient sound at the reference microphone of the active noise reduction device, and the primary path represents a physical path through which the first signal is transmitted to the speaker of the active noise reduction device ; An acquisition module 1902, configured to obtain an estimated signal corresponding to the first signal through a secondary path adaptive filter, the secondary path adaptive filter is used to simulate a secondary path, and the secondary path adaptive filter The filter coefficients of the filter are adjusted based on the target playback sound by the sub-band filter bank; the electrical path module 1903 is used to obtain the second signal through the secondary path, and the secondary path represents that the second signal is transmitted to the loudspeaker through the circuit path, the second signal is the output signal of the primary path adaptive filter and the signal after the target playback sound calculation, the primary path adaptive filter is used to simulate the primary path, the primary path adaptive filter The feedback signal is obtained according to the error signal and the estimated signal; the operation module 1904 is configured to obtain a third signal through the speaker, and the third signal is a signal obtained by operating the first signal and the second signal .

Although only four modules are shown in FIG. 19 , it is understood that this is illustrative only and does not limit the scope of the present disclosure. The active noise reduction system 1900 may also include corresponding modules for performing the steps in the above-mentioned embodiment corresponding to FIG. 13 .

Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are merely example forms of implementing the claims.

Claims

An active noise reduction system based on the FxLMS structure, applied to active noise reduction equipment, is characterized in that it includes:

a primary path adaptive filter, a secondary path adaptive filter, and a subband filter bank;

The primary path adaptive filter is used to simulate a primary path, and the primary path represents a physical path through which the first signal is transmitted from the reference microphone of the active noise reduction device to the speaker of the active noise reduction device;

The secondary path adaptive filter is used to simulate a secondary path, the secondary path represents a circuit path through which a second signal is transmitted to the loudspeaker, and the second signal is the primary path adaptive filter The output signal of the device and the signal after the target playback sound operation;

The subband filter bank is configured to adjust filter coefficients of the secondary path adaptive filter based on the target playback sound.
The system according to claim 1, wherein the system also includes a sampling module for:

Downsampling the output signal of each bandpass filter in the subband filter bank to obtain a downsampling signal, the downsampling signal is used as the input signal of the secondary path adaptive filter, different bandpass Filters have different passbands;

and / or,

Upsampling is performed on the output signal of the secondary path adaptive filter to obtain an upsampling signal, and one upsampling signal corresponds to one downsampling signal.
The system according to claim 2, wherein the input signal of the subband filter bank comprises:

The target playback sound, or, the upsampled signal, or, an error signal.
The system according to claim 3, wherein the sampling module is specifically used for:

In the case that the input signal of the sub-band filter bank is the target playback sound, the first output signal output by each bandpass filter in the sub-band filter bank is down-sampled to obtain the first lower Sampling signal, a first output signal corresponds to a first down-sampling signal, the first down-sampling signal is used as the input signal of the secondary path adaptive filter, and the obtained output of the secondary path adaptive filter is The second output signal is used to adjust the error signal.
The system according to any one of claims 3-4, wherein the sampling module is specifically further used for:

When the input signal of the subband filter bank is the error signal, downsampling is performed on the third output signal output by each bandpass filter in the subband filter bank to obtain the second downsampling signal, a third output signal corresponds to a second downsampling signal, and the second downsampling signal is used as the input signal of the secondary path adaptive filter, and the obtained second output of the secondary path adaptive filter Four output signals are used to adjust the filter coefficients of the secondary path adaptive filter.
A system according to any one of claims 3-5, characterized in that,

In the case where the input signal of the subband filter bank is the upsampling signal, the error signal is based on the fifth output signal and the third signal output by each bandpass filter in the subband filter bank Obtained by performing calculations, the third signal is a signal obtained by performing calculations on the first signal and the second signal.
The system according to any one of claims 1-6, wherein the adjustment mode of the filter coefficients of each bandpass filter in the subband filter bank includes at least any of the following:

Ant colony algorithm, gradient update method.
A kind of active noise reduction method based on FxLMS structure, is applied to active noise reduction equipment, is characterized in that, comprises:

Obtain a first signal through a primary path, the first signal represents the ambient sound at the reference microphone of the active noise reduction device, and the primary path represents the physical path through which the first signal is transmitted to the speaker of the active noise reduction device path;

The estimated signal corresponding to the first signal is obtained through a secondary path adaptive filter, the secondary path adaptive filter is used to simulate the secondary path, and the filter coefficient of the secondary path adaptive filter is determined by the subband Filter banks are tuned based on the target playback sound;

Obtain the second signal through the secondary path, the secondary path represents the circuit path through which the second signal is transmitted to the loudspeaker, the second signal is the output signal of the primary path adaptive filter and the target playback sound The calculated signal, the primary path adaptive filter is used to simulate the primary path, and the feedback signal of the primary path adaptive filter is obtained according to the error signal and the estimated signal;

Obtaining a third signal through the loudspeaker, where the third signal is a signal obtained by computing the first signal and the second signal.
The method according to claim 8, wherein the subband filter bank comprises M bandpass filters, M≥1, and the method further comprises:

Downsampling the output signal of each bandpass filter in the subband filter bank to obtain a downsampling signal, the downsampling signal is used as the input signal of the secondary path adaptive filter, different bandpass Filters have different passbands;

and / or,

Upsampling is performed on the output signal of the secondary path adaptive filter to obtain an upsampling signal, and one upsampling signal corresponds to one downsampling signal.
The method according to claim 9, wherein the input signal of the subband filter bank comprises:

The target playback sound, or, the upsampled signal, or, the error signal.
The method according to claim 10, wherein said down-sampling the output signal of each band-pass filter in the sub-band filter bank to obtain a down-sampled signal comprises:

In the case that the input signal of the sub-band filter bank is the target playback sound, the first output signal output by each bandpass filter in the sub-band filter bank is down-sampled to obtain the first lower For sampling signals, one first output signal corresponds to one first downsampled signal, and the obtained second output signal output by the secondary path adaptive filter is used to adjust the error signal.
The method according to any one of claims 10-11, wherein said downsampling the output signal of each bandpass filter in said subband filter bank, and obtaining the downsampled signal comprises:

When the input signal of the subband filter bank is the error signal, downsampling is performed on the third output signal output by each bandpass filter in the subband filter bank to obtain the second downsampling signal, a third output signal corresponds to a second downsampled signal, and the obtained fourth output signal output by the secondary path adaptive filter is used to adjust the filter coefficient of the secondary path adaptive filter.
The method according to any one of claims 10-12, characterized in that,

When the input signal of the sub-band filter bank is the up-sampling signal, the error signal is based on the fifth output signal output by each bandpass filter in the sub-band filter bank and the first The three signals are obtained by operation.
The method according to any one of claims 8-13, wherein the adjustment mode of the filter coefficients of each bandpass filter in the subband filter bank includes at least any of the following:

Ant colony algorithm, gradient update method.
An active noise reduction device, the device has the function of implementing the method according to any one of claims 8-14, the function is realized by hardware or by executing corresponding software through hardware, and the hardware or the software includes a or multiple modules corresponding to the functions described.
An active noise reduction device, including a processor and a memory, the processor is coupled to the memory, characterized in that,

The memory is used to store programs;

The processor is configured to execute the program in the memory, so that the device executes the method according to any one of claims 8-14.
A computer-readable storage medium, including a program, which, when run on a computer, causes the computer to execute the method according to any one of claims 8-14.
A computer program product comprising instructions which, when run on a computer, cause the computer to perform the method according to any one of claims 8-14.
A chip, the chip includes a processor and a data interface, the processor reads instructions stored in the memory through the data interface, and executes the method according to any one of claims 8-14.