WO2020248164A1 - 主动降噪方法、装置、芯片、主动控制系统和存储介质 - Google Patents

主动降噪方法、装置、芯片、主动控制系统和存储介质 Download PDF

Info

Publication number
WO2020248164A1
WO2020248164A1 PCT/CN2019/090912 CN2019090912W WO2020248164A1 WO 2020248164 A1 WO2020248164 A1 WO 2020248164A1 CN 2019090912 W CN2019090912 W CN 2019090912W WO 2020248164 A1 WO2020248164 A1 WO 2020248164A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
frequency domain
error
noise reduction
coefficients
Prior art date
Application number
PCT/CN2019/090912
Other languages
English (en)
French (fr)
Inventor
朱虎
王鑫山
李国梁
郭红敬
韩文凯
Original Assignee
深圳市汇顶科技股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市汇顶科技股份有限公司 filed Critical 深圳市汇顶科技股份有限公司
Priority to CN201980001078.6A priority Critical patent/CN110402540B/zh
Priority to PCT/CN2019/090912 priority patent/WO2020248164A1/zh
Publication of WO2020248164A1 publication Critical patent/WO2020248164A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
    • H03H21/00—Adaptive networks
    • H03H21/0012—Digital adaptive filters
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
    • H03H17/00—Networks using digital techniques
    • H03H2017/0072—Theoretical filter design
    • H03H2017/009—Theoretical filter design of IIR filters
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
    • H03H21/00—Adaptive networks
    • H03H21/0012—Digital adaptive filters
    • H03H2021/0085—Applications
    • H03H2021/0094—Interference Cancelling

Definitions

  • This application relates to the technical field of active noise reduction, and in particular to an active noise reduction method, device, chip, active control system and storage medium.
  • the purpose of some embodiments of this application is to provide an active noise reduction method, device, chip, active control system and storage medium, which can realize adaptive active noise reduction while reducing the computational complexity and hardware in the noise reduction process.
  • the cost of resources is conducive to the realization of hardware.
  • the embodiment of the present application provides an active noise reduction method, including: performing frequency domain adaptive filtering according to the environmental noise signal and the error signal collected at the target noise reduction point, and obtaining the frequency domain coefficients of the adaptive filter corresponding to the main channel
  • the error signal is the signal superimposed on the target noise reduction point after the environmental noise signal passes through the main channel and the secondary channel respectively
  • the frequency domain coefficient of the adaptive filter corresponding to the main channel is converted to infinite Time domain coefficients of the impulse response IIR filter; wherein the order of the IIR filter is smaller than the order of the adaptive filter; updating the time domain coefficients of the IIR filter to the IIR filter
  • the IIR filter is located on the secondary channel and is used to filter the environmental noise signal passing through the secondary channel to output a noise reduction signal.
  • the embodiment of the present application also provides an active noise reduction device, including: a frequency domain adaptive filtering module for performing frequency domain adaptive filtering according to the environmental noise signal and the error signal collected at the target noise reduction point to obtain the main channel The frequency domain coefficient of the corresponding adaptive filter; wherein, the error signal is the signal superimposed at the target noise reduction point after the environmental noise signal passes through the primary channel and the secondary channel respectively; the IIR filter design module and the The frequency domain adaptive filtering module is connected to convert the frequency domain coefficients of the adaptive filter corresponding to the main channel into the time domain coefficients of the infinite impulse response IIR filter, and update the time domain coefficients of the IIR filter To the IIR filter; wherein the order of the IIR filter is smaller than the order of the adaptive filter; the IIR filter is connected to the coefficient conversion module and is located on the secondary channel, It is used to filter the environmental noise signal passing through the secondary channel to output a noise reduction signal.
  • a frequency domain adaptive filtering module for performing frequency domain adaptive filtering according to the environmental noise signal and the error signal collected at the
  • An embodiment of the present application also provides an active noise reduction chip, including: at least one processor; and, a memory communicatively connected with the at least one processor; wherein the memory stores the memory that can be used by the at least one processor.
  • the executed instruction is executed by the at least one processor, so that the at least one processor can execute the above-mentioned active noise reduction method.
  • the embodiment of the present application also provides an active control system, including the above-mentioned active noise reduction chip and infinite impulse response IIR filter; the active noise reduction chip is used to obtain the time domain coefficients of the IIR filter and combine the The time domain coefficients of the IIR filter are updated to the IIR filter; the IIR filter is located on the secondary channel, and is used to filter the environmental noise signal passing through the secondary channel to output a noise reduction signal.
  • An embodiment of the present application also provides a computer-readable storage medium storing a computer program, wherein the computer program is characterized in that the above-mentioned active noise reduction method is implemented when the computer program is executed by a processor.
  • the embodiment of the application performs frequency-domain adaptive filtering according to the environmental noise signal and the error signal collected at the target noise reduction point to obtain the frequency domain coefficients of the adaptive filter corresponding to the main channel, and the main The frequency domain coefficient of the adaptive filter corresponding to the channel is converted to the time domain coefficient of the IIR filter, and the time domain coefficient of the IIR filter is updated to the IIR filter.
  • adaptive filtering in the frequency domain can effectively avoid the update of filter coefficients at each sampling point in the time domain in related technologies, resulting in a large amount of data interaction and high computational complexity, which is conducive to reducing algorithm resource overhead, and Obtaining the frequency domain coefficients of the adaptive filter in the frequency domain can make full use of all the information of the collected signal, which is more conducive to obtaining the time domain coefficients of the IIR filter that converges to the transfer function of the main channel, so that the environmental noise signals pass through the main channel respectively.
  • the signal superimposed on the target noise reduction point after the channel and the secondary channel is closer to 0, which is conducive to more accurate noise reduction.
  • a low-order IIR filter with a lower order than the adaptive filter is used, which can effectively avoid the need for high-order filters and reduce hardware resources while achieving adaptive active noise reduction.
  • the overhead is conducive to hardware implementation.
  • performing frequency domain adaptive filtering according to the environmental noise signal and the error signal collected at the target noise reduction point to obtain the frequency domain coefficient of the adaptive filter corresponding to the main channel includes: converting the environmental noise signal into environmental noise Frequency domain signal, and convert the error signal into an error frequency domain signal; perform an error amplitude control on the error frequency domain signal according to the environmental noise frequency domain signal; according to the error frequency domain signal after the error amplitude control and the error frequency domain signal The frequency domain signal of environmental noise is used to obtain the frequency domain coefficient of the adaptive filter corresponding to the main channel.
  • the error amplitude control is performed on the error frequency domain signal, which can avoid the divergence of the coefficient of the filter when the error signal collected at the target noise reduction point is large due to non-standard wearing. Since different wearing states may cause large or small changes in the error signal collected at the target noise reduction point, the error amplitude control is performed on the error frequency domain signal in the embodiment of the present application to ensure that it can be obtained under different wearing states
  • the time-domain coefficients of the filter that adaptively converges to the transfer function of the main channel are more robust, more suitable for various wearing scenarios, and ensure better noise reduction capabilities.
  • performing error amplitude control on the error frequency domain signal according to the environmental noise frequency domain signal includes: normalizing the error frequency domain signal according to the environmental noise frequency domain signal to obtain the normalization process After the error frequency domain signal; according to the comparison result of the amplitude of the normalized error frequency domain signal and the preset threshold, obtain the error limit coefficient; obtain the error amplitude control after the error limit coefficient Error frequency domain signal.
  • the embodiment of the present application provides a specific error amplitude control method, which is conducive to better error amplitude control on error frequency domain signals.
  • performing normalization processing on the error frequency domain signal according to the environmental noise frequency domain signal to obtain the error frequency domain signal after the normalization processing includes: obtaining the power spectrum and the power spectrum of the environmental noise frequency domain signal respectively The amplitude spectrum of the error frequency domain signal; the ratio of the amplitude spectrum and the power spectrum is used as the amplitude spectrum of the error frequency domain signal after normalization processing.
  • the embodiment of the present application provides a specific normalization processing method, which facilitates obtaining the amplitude spectrum of the error signal after the normalization processing.
  • obtaining the error limiting coefficient according to the comparison result of the amplitude of the normalized error frequency domain signal with a preset threshold value includes: calculating the error limiting coefficient by the following formula:
  • the Ef(k) is the error limiting coefficient
  • is the amplitude of the error frequency domain signal after the normalization processing
  • the ⁇ (k) is the A preset threshold, where k is the frequency point of the error frequency domain signal.
  • converting the frequency domain coefficients of the adaptive filter corresponding to the main channel into the time domain coefficients of the infinite impulse response IIR filter includes: converting the frequency domain coefficients of the adaptive filter corresponding to the main channel into The time domain coefficients of the adaptive filter corresponding to the main channel; the time domain coefficients of the adaptive filter corresponding to the main channel are converted to the time domain coefficients of the IIR filter according to the LMS algorithm, which facilitates the adaptive filtering
  • the frequency domain coefficient of the filter is converted to the time domain coefficient of the IIR filter.
  • the conversion of the time domain coefficients of the adaptive filter corresponding to the main channel into the time domain coefficients of the IIR filter according to the LMS algorithm includes: performing the time domain coefficients of the adaptive filter corresponding to the main channel Flip; generate a random number, and obtain a desired signal according to the time domain coefficient and the random number after the flip; obtain the time domain coefficient of the IIR filter according to the LMS algorithm and the desired signal.
  • the use of random numbers is beneficial to obtain more stable time-domain coefficients of the IIR filter, and can make the time-domain coefficients of the IIR filter more convergent.
  • converting the time domain coefficients of the adaptive filter corresponding to the main channel into the time domain coefficients of the IIR filter according to the LMS algorithm includes converting the time domain coefficients of the adaptive filter corresponding to the main channel according to the LMS algorithm The coefficients are converted into the time domain coefficients of the m-order IIR filter; the time domain coefficients of the m-order IIR filter are converted into the time domain coefficients of a plurality of n-order IIR filters; wherein, the m is greater than the n, so Each of the plurality of n-order IIR filters has respective time-domain coefficients; the updating the time-domain coefficients of the IIR filter to the IIR filter specifically includes: filtering the plurality of n-order IIR filters The time domain coefficients of the filter are updated to the plurality of n-order IIR filters.
  • the time domain coefficients of the m-order IIR filter are converted into the time domain coefficients of multiple n-order IIR filters, that is, the high-order coefficients of the IIR filter are converted into multiple low-order coefficients, and the secondary channel
  • the signal processing effect of the above multiple n-order IIR filters is conducive to better approximating the signal processing effect of the transfer function of the main channel, so that the environmental noise signal passes through the main channel and the secondary channel respectively and is superimposed on the target noise reduction point The signal is closer to 0, which is conducive to achieving a better noise reduction effect.
  • the method further includes: adaptively updating the The coefficients of the adaptive filter corresponding to the secondary channel; update the coefficients of the adaptive filter corresponding to the secondary channel to the adaptive filter corresponding to the secondary channel; said according to the environmental noise signal and the target The environmental noise signal in the frequency domain adaptive filtering of the error signal collected by the noise reduction point is the environmental noise signal passed through the adaptive filter corresponding to the secondary channel.
  • the embodiments of the present application can make the transfer function of the secondary channel change with environmental factors, that is, when the influence of the secondary channel on the environmental noise signal changes with environmental factors, it can also be based on the change of the transfer function of the secondary channel.
  • Change adaptively update the coefficients of the adaptive filter corresponding to the secondary channel, so as to obtain the frequency domain coefficients of the adaptive filter that converges to the transfer function of the main channel, and further obtain the IIR filter that converges to the transfer function of the main channel The time domain coefficient.
  • the adaptively updating the coefficients of the adaptive filter corresponding to the secondary channel specifically includes: adaptively updating the frequency domain coefficients of the adaptive filter corresponding to the secondary channel.
  • the adaptive update of the frequency domain coefficients of the adaptive filter corresponding to the secondary channel in the frequency domain can effectively avoid the update of each sampling point in the time domain in the related technology, resulting in a large amount of data interaction and high computational complexity. The problem is conducive to reducing algorithm resource overhead.
  • adaptively updating the adaptive filter coefficients corresponding to the secondary channel includes: introducing a preset white noise signal on the secondary channel, and according to the white noise signal and the adjusted error signal , Obtaining the coefficients of the adaptive filter corresponding to the secondary channel; wherein the adjusted error signal is a signal obtained by superimposing the white noise signal processed by a preset filter and the error signal;
  • the coefficient of the preset filter is the same as the coefficient of the adaptive filter corresponding to the secondary channel.
  • Fig. 1 is a block diagram of an active noise reduction method according to the first embodiment of the present application
  • Fig. 2 is a flowchart of an active noise reduction method according to the first embodiment of the present application
  • step 201 is a flowchart of the implementation process of step 201 in the first embodiment of the present application.
  • step 201 is a flowchart of the implementation process of step 201 in the second embodiment of the present application.
  • Fig. 5 is a block diagram of an active noise reduction method according to a third embodiment of the present application.
  • Fig. 6 is a flowchart of an active noise reduction method according to a third embodiment of the present application.
  • Fig. 7 is a schematic diagram of an active noise reduction device according to a fourth embodiment of the present application.
  • FIG. 8 is a schematic structural diagram of an active noise reduction chip according to a fifth embodiment of the present application.
  • Fig. 9 is a schematic diagram of an active control system in a sixth embodiment according to the present application.
  • the first embodiment of the present application relates to an active noise reduction method, which can be applied to voice interactive products, including but not limited to electronic devices such as earphones and hearing aids.
  • a technique based on a combination of frequency-domain adaptive filtering and an IIR (Infinite Impulse Response) filter is mainly used to adaptively update the time domain coefficients of the IIR filter on the secondary channel, thereby realizing active reduction. noise.
  • IIR Infinite Impulse Response
  • Noise-cancelling headsets are generally equipped with microphones at two locations.
  • An external microphone is set at the earphone shell.
  • the external microphone can be used to collect ambient noise signals. Put on the headset and set the microphone at the human ear and the microphone at the human ear.
  • the block diagram of the active noise reduction method which is Figure 1, it can be seen that the ambient noise signal collected by the external microphone is x(n), and the error signal collected by the microphone at the human ear is e(n).
  • e(n) d(n)-y(n), where d(n) is the expected signal obtained after the environmental noise signal x(n) passes through the main channel, which can be regarded as
  • the transfer function P(n) is the signal obtained after signal processing of x(n).
  • the main channel can be understood as the transmission path where the environmental noise signal is transmitted to the human ear through the air medium, and the environmental noise signal reaches the human ear through the main channel At this time, the influence of the main channel on the environmental noise signal can be abstracted as the transfer function P(n) of the main channel.
  • P(n) is generally affected by the earphone structure and wearing accuracy.
  • the active control system in Figure 1 is located on the secondary channel.
  • the hardware components on the secondary channel in Figure 1 usually include analog-to-digital converters, speakers, etc. in addition to the IIR filter.
  • the impact of all hardware on the secondary channel on the environmental noise signal will not change with changes in environmental factors such as temperature.
  • the impact of all hardware on the secondary channel on the environmental noise signal can be abstracted as secondary The transfer function of all hardware on the level channel.
  • the transfer functions of all hardware on the secondary channel can be understood as 1 or other fixed values.
  • y(n) is the output signal obtained after the environmental noise signal x(n) passes through the secondary channel, which can be regarded as the signal obtained after the IIR filter on the secondary channel filters x(n).
  • the y(n) output by the IIR filter can stimulate the speaker in the earphone to emit a noise reduction signal y(n).
  • the noise reduction signal y(n) emitted by the speaker and the desired signal d(n) are superimposed in the air.
  • the local microphone can collect the superimposed error signal. If the signal resulting from the superposition of the noise reduction signal y(n) and the desired signal d(n) approaches 0, it can be considered that a better noise reduction effect has been achieved.
  • the specific process of the active noise reduction method in this embodiment may be as shown in FIG. 2, including:
  • Step 201 Perform frequency domain adaptive filtering according to the environmental noise signal and the error signal collected at the target noise reduction point, and obtain the frequency domain coefficient of the adaptive filter corresponding to the main channel.
  • the target noise reduction point can be the position of the microphone in the earphone.
  • the earphone can be provided with an active control system as shown in Figure 1.
  • step 201 can be implemented by the frequency domain adaptive filtering module 102, which performs frequency domain adaptive filtering according to the environmental noise signal and error signal, and obtains the adaptive filter corresponding to the main channel.
  • the specific implementation of frequency domain coefficients can be shown in Figure 3, including:
  • Step 301 Perform frequency domain conversion on the environmental noise signal and the error signal respectively.
  • the environmental noise signal and error signal data can be returned to the frequency domain through Fourier transform FFT, and the frequency domain conversion of the environmental noise signal and error signal can be realized through the following formula:
  • k represents the frequency point
  • x(n) is the environmental noise signal collected by the external microphone
  • e(n) the error signal collected by the microphone at the human ear
  • N represents the number of FFT points.
  • the value of N can be It is 1024
  • U(k) is the environmental noise frequency domain signal after x(n) frequency domain conversion
  • E(k) is the error frequency domain signal after e(n) frequency domain conversion.
  • the environmental noise signal and error signal mentioned in this embodiment and the following embodiments are all time-domain signals
  • the environmental noise frequency-domain signal and error frequency-domain signal are both frequency-domain signals.
  • Step 302 Calculate the frequency domain gradient according to the environmental noise frequency domain signal and the error frequency domain signal.
  • the frequency domain gradient can be calculated by the following formula
  • ⁇ is the iteration step size
  • the selection of the step size is related to the convergence speed. If the step size is selected too large, although the convergence is fast, it may cause the filter coefficients to diverge; if the step size is selected too small, the convergence speed is too slow .
  • U(k) * represents the conjugate signal of the environmental noise frequency domain signal.
  • Step 303 Convert the frequency domain gradient to the time domain gradient.
  • the frequency domain gradient can be converted into a time domain gradient through the inverse Fourier transform IFFT, that is, the frequency domain gradient can be converted into a time domain gradient by the following formula
  • Step 304 Perform zero padding on the time domain gradient.
  • time-domain gradient can be zero-filled by the following formula to obtain the zero-filled time-domain gradient ⁇ (n):
  • Step 305 Perform frequency domain conversion on the time domain gradient after zero padding.
  • the zero-padded time-domain gradient can be converted to the frequency domain by Fourier transform.
  • ⁇ W(k) is the variation of the frequency domain coefficient of the FIR filter.
  • Step 306 Obtain frequency domain coefficients of the adaptive filter corresponding to the main channel according to the time domain gradient after frequency domain conversion.
  • the frequency domain coefficients of the adaptive filter corresponding to the main channel are only obtained from the software level, and the hardware device of the adaptive filter does not necessarily exist.
  • an adaptive filter with frequency domain coefficients belongs to a finite impulse response FIR filter, referred to as FIR filter for short; this embodiment and the following embodiments are described in detail under the term FIR filter.
  • the frequency domain coefficients of the adaptive filter corresponding to the main channel can be referred to as the frequency domain coefficients of the FIR filter, but it is not limited thereto.
  • step 306 the frequency domain coefficients of the FIR filter can be specifically obtained by the following formula:
  • ⁇ W ⁇ +1 (k) represents the variation of the frequency domain coefficient of the FIR filter at the k frequency point obtained by the ⁇ +1 estimation.
  • Step 202 Convert the frequency domain coefficients of the adaptive filter corresponding to the main channel into the time domain coefficients of the IIR filter.
  • the order of the IIR filter is smaller than the order of the adaptive filter corresponding to the main channel, that is, the order of the IIR filter is smaller than the order of the FIR filter.
  • the order of the FIR filter can be 3000 orders, and the order of the IIR filter can be 6 orders.
  • the frequency domain coefficients of the FIR filter with order 3000 can be converted to order 6
  • the time domain coefficients of the IIR filter. This step can be implemented by the IIR filter design module 103 in FIG. 1.
  • the method of converting the frequency domain coefficients of the FIR filter into the time domain coefficients of the IIR filter may be: firstly, the frequency domain coefficients of the FIR filter obtained by the frequency domain adaptive filtering module 102 After doing the inverse Fourier transform IFFT, the time domain coefficient w(n+1) of the FIR filter is obtained, and then the w(n+1) vector is flipped to obtain a new Then generate a set of random numbers x(n+1), according to the random number x(n+1) and Get the desired signal d(n+1), such as combining a random number with Multiply to obtain the desired signal d(n+1), and finally obtain the time domain coefficients of the IIR filter according to the desired signal d(n+1) and the minimum root mean square LMS algorithm.
  • the time domain coefficient of the IIR filter can include the numerator coefficient a(n) and the denominator coefficient b(n), and the time domain coefficient of the filter is
  • the frequency response of the FIR filter is the same as that of the IIR filter with a filter time domain coefficient of a(n)/b(n). It should be noted that those skilled in the art can obtain the time domain coefficients of the IIR filter according to the desired signal d(n+1) and the minimum root mean square LMS algorithm based on the existing related technology. The specific implementation of the time domain coefficient of the IIR filter is not described in detail.
  • the method of converting the frequency domain coefficients of the FIR filter into the time domain coefficients of the IIR filter may be: using the Prony method to realize the conversion of the coefficients.
  • this embodiment only takes the above two coefficient conversion methods as examples.
  • any method that can convert the frequency domain coefficients of the FIR filter into the time domain coefficients of the IIR filter is described in this embodiment. Within the scope of protection.
  • the number of IIR filters in the active control system can be one, as shown in FIG. 1.
  • the manner of converting the frequency domain coefficients of the FIR filter into the time domain coefficients of the IIR filter can refer to the above description.
  • the active control system may include multiple IIR filters, and the multiple IIR filters are in series.
  • the frequency domain coefficients of the FIR filter are converted into the time domain of the IIR filter.
  • the method of coefficients can be as follows: First, according to the LMS algorithm, the time domain coefficients of the FIR filter are converted to the time domain coefficients of the m-order IIR filter, where the time domain coefficients of the FIR filter are converted to the time domain coefficients of the m-order IIR filter.
  • the method of domain coefficients can be as described above: flip the time domain coefficients of the FIR filter corresponding to the main channel to generate random numbers, and obtain the desired signal according to the flipped time domain coefficients and random numbers, according to the LMS algorithm and The desired signal gets the time domain coefficients of the m-order IIR filter. It should be noted that this embodiment only provides an example, and the manner of converting the time domain coefficients of the FIR filter into the time domain coefficients of the m-order IIR filter in specific implementation is not limited to the foregoing manner.
  • the time-domain coefficients of the m-order IIR filter are converted into the time-domain coefficients of multiple n-order IIR filters; where m is greater than n, and the multiple n-order IIR filters have their own time-domain coefficients.
  • m 6
  • n 2
  • the time domain coefficient of the FIR filter is first converted to the time domain coefficient of the 6th order IIR filter, and then the time domain coefficients of the 6th order IIR filter are converted into 3 second order
  • the time domain coefficients of the IIR filter, and finally the numerator and denominator of the time domain coefficients of each second-order IIR filter are obtained.
  • the specific implementation of converting the time domain coefficients of the 6th order IIR filter into the time domain coefficients of three 2nd order IIR filters can refer to the following formula:
  • a ij represents the j-th adjustable coefficient of the numerator of the i-th order 2 IIR filter
  • b ij represents the j-th adjustable coefficient of the denominator of the i-th order 2 filter.
  • Step 203 Update the time domain coefficients of the IIR filter to the IIR filter.
  • the active control system includes an IIR filter
  • the time domain coefficients of the one IIR filter obtained will be written into the IIR filter.
  • the active control system includes multiple IIR filters
  • the The obtained time domain coefficients of the multiple IIR filters are written into the multiple IIR filters respectively.
  • the IIR filter is located on the secondary channel. You can refer to the IIR filter 104 on the secondary channel in FIG. 1. After the IIR filter 104 is written into the filter coefficients, the environmental noise signal passing through the secondary channel can be filtered and output. Noise signal.
  • the IIR filter usually has initial time domain coefficients, and the first time it is updated, it can be considered that the initial time domain coefficients of the IIR filter are updated.
  • the noise reduction signal output by the IIR filter 104 can stimulate the speaker in the headset to emit a noise reduction signal, so that the microphone at the human ear at the target noise reduction point can collect the noise reduction signal y(n) and The superimposed signal of the desired signal d(n) is e(n).
  • the signal d(n) has the same spectrum distribution, the same amplitude, and the opposite phase of the noise reduction signal y(n), so that the error signal e(n) collected by the microphone at the human ear is infinitely close to 0, and is constantly based on the collected
  • the error signal e(n) and the environmental noise signal x(n) are adaptively filtered in the frequency domain, which is beneficial to achieve a good noise reduction effect.
  • this embodiment performs frequency-domain adaptive filtering based on environmental noise signals and error signals, which can effectively avoid updating the filter coefficients at each sampling point in the time domain in related technologies, resulting in a large amount of data interaction .
  • the problem of high computational complexity is conducive to reducing algorithm resource overhead, and obtaining the frequency domain coefficients of the adaptive filter in the frequency domain can make full use of all the information of the collected signal, and it is more conducive to obtaining a transfer function that converges to the main channel.
  • the time domain coefficient of the IIR filter makes the ambient noise signal pass through the main channel and the secondary channel respectively, and the signal superimposed at the target noise reduction point is closer to 0, which is conducive to more accurate noise reduction.
  • a low-order IIR filter with a lower order than the adaptive filter is used, which can effectively avoid the need for a high-order adaptive filter while achieving adaptive active noise reduction.
  • the hardware resource overhead is conducive to hardware implementation.
  • the second embodiment of the present application relates to an active noise reduction method.
  • This embodiment is a further improvement of the first embodiment.
  • the main improvement lies in the frequency domain adaptive filtering according to the environmental noise signal and error signal to obtain adaptive
  • the error amplitude control of the error frequency domain signal is helpful to avoid the divergence of the filter coefficient when the error signal collected at the human ear is large, and it can be adapted to different Wearing state.
  • the frequency domain adaptive filtering is performed according to the environmental noise signal and the error signal, and the frequency domain coefficients of the adaptive filter are obtained, that is, the implementation of step 201 can be the error frequency domain signal error according to the environmental noise frequency domain signal.
  • Amplitude control according to the error frequency domain signal and environmental noise frequency domain signal after the error amplitude control, obtain the frequency domain coefficient of the adaptive filter.
  • the implementation of step 201 in this embodiment may be as shown in FIG. 4, including:
  • Step 401 Perform frequency domain conversion on the environmental noise signal and the error signal respectively.
  • Step 401 is substantially the same as step 301 in the first embodiment, and will not be repeated here to avoid repetition.
  • Step 402 Normalize the error frequency domain signal according to the environmental noise frequency domain signal, and obtain the error frequency domain signal after the normalization process.
  • the power spectrum of the environmental noise frequency domain signal and the amplitude spectrum of the error frequency domain signal can be obtained separately, where the power spectrum of the environmental noise frequency domain signal can be calculated by the following formula:
  • P i (k) i represents the time domain signal power of the ambient noise frequency spectrum
  • a gamma] is in the range 0 ⁇ ⁇ 1 is the forgetting factor
  • U (k ) Is the frequency domain signal of environmental noise.
  • the amplitude spectrum of the error frequency domain signal can be calculated according to the expression of the error frequency domain signal. Then, the ratio of the amplitude spectrum and the power spectrum can be used as the normalized error frequency domain signal.
  • the amplitude spectrum E1(k) of the normalized error frequency domain signal can be obtained by the following formula:
  • E(k) is the amplitude spectrum of the error frequency domain signal
  • P(k) is the power spectrum of the environmental noise frequency domain signal.
  • Step 403 Obtain an error limiting coefficient according to the comparison result of the amplitude of the error frequency domain signal after the normalization process and the preset threshold.
  • the error limiting coefficient Ef(k) can be calculated by the following formula:
  • is the amplitude of the error frequency domain signal after normalization processing
  • the ⁇ (k) is a preset threshold
  • the k is the frequency point of the error frequency domain signal. It can be seen from the above formula that if
  • ⁇ (k) can be preset with multiple values, and different frequency value ranges correspond to different preset thresholds.
  • the value of the preset threshold is:
  • the f is the frequency value of the error frequency domain signal
  • the k is the frequency point of the error frequency domain signal. Since the time domain signal is converted into frequency domain signal based on different forms of Fourier transform under the same frequency, the number of spectrum points will be different. Therefore, the frequency value is used as the reference value for the preset threshold value, which is beneficial to adapt to different forms of Fourier transform.
  • the leaf changes.
  • different frequency value ranges correspond to different specific setting methods of preset thresholds, which can be set by those skilled in the art based on the sensitivity of the human ear to sounds of different frequencies. This is only an example and is not limited in any way.
  • the above-mentioned value method of the preset threshold provided in the embodiment of the present application facilitates the selection of the preset threshold according to different frequency ranges.
  • Step 404 Obtain an error frequency domain signal after error amplitude control according to the error limiting coefficient.
  • the error frequency domain signal after error amplitude control may be the product of the error limiting coefficient and the error frequency domain signal obtained in step 401, that is, the error frequency domain signal after error amplitude control may be: E(k) ⁇ Ef(k).
  • Step 405 Calculate the frequency domain gradient according to the environmental noise frequency domain signal and the error frequency domain signal after the error amplitude control.
  • the frequency domain gradient can be calculated by the following formula:
  • ⁇ is the iteration step size
  • U(k) * represents the conjugate signal of the environmental noise frequency domain signal
  • E(k) ⁇ Ef(k) represents the error frequency domain signal after the error amplitude control.
  • Step 406 Convert the frequency domain gradient to the time domain gradient.
  • Step 407 Perform zero padding on the time domain gradient.
  • Step 408 Perform frequency domain conversion on the time domain gradient after zero padding.
  • Step 409 Obtain the frequency domain coefficients of the FIR filter according to the time domain gradient after frequency domain conversion.
  • Step 406 to step 409 are substantially the same as step 303 to step 306 in the first embodiment. To avoid repetition, they will not be repeated here.
  • the error amplitude control of the error frequency domain signal in this embodiment can avoid the divergence of the filter coefficient when the error signal collected by the target noise reduction point is large. Since different wearing states may cause large or small changes in the error signal collected at the target noise reduction point, the error amplitude control is performed on the error frequency domain signal in the embodiment of the present application to ensure that it can be obtained under different wearing states
  • the time domain coefficient of the filter that adaptively converges to the transfer function of the main channel is more robust and more suitable for various wearing scenarios, ensuring better noise reduction capabilities.
  • the third embodiment of the present application relates to an active noise reduction method.
  • the secondary channel is an ideal channel, that is, the transfer function of all hardware on the secondary channel is 1 or is fixed.
  • the numerical value The third embodiment of the present application mainly introduces the case where the secondary channel is not an ideal channel. Not an ideal channel can be understood as the transfer function of all hardware on the secondary channel will change. For example, the processing capabilities of hardware such as analog-to-digital converters and IIR filters on the secondary channel may change with environmental factors such as temperature. , Resulting in a change in the transfer function of all hardware on the secondary channel.
  • FIG. 5 The block diagram of the active noise reduction method in this embodiment can be as shown in Fig. 5. Compared with Fig. 1, the main difference is that in Fig. 5, a primary secondary channel adaptive filter module 502 is designed.
  • the flow chart of the noise reduction method is shown in Figure 6, including:
  • Step 601 adaptively update the coefficients of the adaptive filter corresponding to the secondary channel, and update the coefficients of the adaptive filter corresponding to the secondary channel to the adaptive filter corresponding to the secondary channel.
  • the secondary channel adaptive filter module 502 can be used to adaptively update the coefficients of the adaptive filter corresponding to the secondary channel.
  • the coefficients of the adaptive filter corresponding to the secondary channel are only obtained from the software level, and the hardware device of the adaptive filter does not necessarily exist on the secondary channel.
  • the filter can be the one located on the secondary channel in Figure 5 505 can be realized by software, that is, the coefficient of x(n) is processed by software, and x(n) is combined with the filter coefficient After being multiplied, it is input into the frequency domain adaptive filtering module 102 in the main channel adaptive filtering module 505.
  • the method of updating the adaptive filter coefficients corresponding to the secondary channel may be: introducing a preset white noise signal on the secondary channel, and performing frequency domain adaptation according to the white noise signal and the adjusted error signal Filtering to obtain the coefficients of the adaptive filter corresponding to the secondary channel; among them, the adjusted error signal is the white noise signal processed by the preset filter and the superimposed signal of the error signal; the coefficients of the preset filter are The coefficients of the adaptive filter corresponding to the secondary channel are the same, and the preset filter can be the one in Figure 5 It can also be understood as the estimated transfer function of all hardware on the secondary channel.
  • the white noise signal can be sent by the white noise module 503 in FIG. 5, and the frequency domain adaptive filtering module 507 in FIG. 5 performs frequency domain adaptive filtering according to the white noise signal and the adjusted error signal to obtain the free channel corresponding to the secondary channel. Adapt the coefficients of the filter, and then write the obtained coefficients of the adaptive filter corresponding to the secondary channel into the with in. It can be considered that when the frequency domain adaptive filtering module 507 is performing frequency domain adaptive filtering at the current moment, with The coefficients written in are the coefficients calculated by the frequency domain adaptive filtering module 507 at the previous time. among them, It is used to offset the influence of the product of the white noise signal and S(z)504 on the error signal after the white noise signal is introduced.
  • S(z)504 is the transfer function of all hardware on the secondary channel. It is understandable that if the estimated transfer function of all hardware on the secondary channel The more convergent the transfer function S(z)504 of all hardware on the secondary channel, the better the cancellation effect. Ideally, the product of the white noise signal and S(z)504 in the adjusted error signal is completely white noise signal versus The product of is cancelled out, that is, the influence of white noise signal does not exist in the adjusted error signal.
  • the frequency domain adaptive filtering is performed according to the white noise signal and the adjusted error signal to obtain the coefficients of the adaptive filter corresponding to the secondary channel in the same manner as in the first or second embodiment.
  • the frequency domain adaptive filtering is performed according to the environmental noise signal and the error signal, and the specific method for obtaining the frequency domain coefficients of the adaptive filter corresponding to the main channel is roughly the same. In order to avoid repetition, it will not be repeated here.
  • adaptively updating the coefficients of the adaptive filter corresponding to the secondary channel may be specifically: adaptively updating the frequency domain coefficients of the adaptive filter corresponding to the secondary channel. That is to say, the adaptive update of the frequency domain coefficients of the adaptive filter corresponding to the secondary channel in the frequency domain can effectively avoid the update of each sampling point in the time domain in the related technology, resulting in a large amount of data interaction and calculation. The problem of high complexity helps to reduce algorithm resource overhead.
  • Step 602 Perform frequency domain adaptive filtering according to the environmental noise signal processed by the filter corresponding to the secondary channel and the adjusted error signal, and obtain the frequency domain coefficient of the adaptive filter corresponding to the main channel.
  • Step 603 Convert the frequency domain coefficients of the adaptive filter corresponding to the main channel into the time domain coefficients of the IIR filter.
  • Step 604 Update the time domain coefficients of the IIR filter to the IIR filter.
  • Step 602 to step 604 are substantially the same as step 201 to step 203 in the first embodiment, and to avoid repetition, they will not be repeated here.
  • the embodiments of the present application can make it possible to obtain the frequency domain coefficients of the FIR filter converging to the transfer function of the secondary channel when the transfer function of all hardware on the secondary channel changes.
  • the time domain coefficients of the IIR filter converging to the transfer function of the secondary channel are obtained, so as to realize the active noise reduction more accurately.
  • the fourth embodiment of the present application relates to an active noise reduction device, as shown in FIG. 7, including: a frequency domain adaptive filtering module 701, which is used to perform frequency domain based on environmental noise signals and error signals collected at target noise reduction points Adaptive filtering to obtain the frequency domain coefficients of the adaptive filter corresponding to the main channel; wherein the error signal is the signal superimposed at the target noise reduction point after the environmental noise signal passes through the main channel and the secondary channel respectively; IIR The filter design module 702 is connected to the frequency domain adaptive filtering module 703, and is used to convert the frequency domain coefficients of the adaptive filter corresponding to the main channel into the time domain coefficients of the infinite impulse response IIR filter, and The time domain coefficients of the IIR filter are updated to the IIR filter; wherein the order of the IIR filter is smaller than the order of the adaptive filter; the IIR filter 703 and the IIR filter The design module 702 is connected to and located on the secondary channel, and is used to filter the environmental noise signal passing through the secondary channel to output a noise reduction signal.
  • this embodiment is a device embodiment corresponding to the first to third embodiments, and this embodiment can be implemented in cooperation with the first embodiment.
  • the related technical details mentioned in the first embodiment are still valid in this embodiment, and in order to reduce repetition, they will not be repeated here.
  • the related technical details mentioned in this embodiment can also be applied to the first embodiment.
  • this embodiment does not introduce units that are not closely related to solving the technical problems proposed by the present invention, but this does not mean that there are no other units in this embodiment. Unit.
  • the fifth embodiment of the present application relates to an active noise reduction chip, as shown in FIG. 8, comprising: at least one processor 801; and a memory 802 communicatively connected with the at least one processor 801; wherein, the memory 802 An instruction that can be executed by the at least one processor 801 is stored, and the instruction is executed by the at least one processor 801, so that the at least one processor 801 can execute the aforementioned active noise reduction method.
  • the memory 802 and the processor 801 are connected in a bus manner.
  • the bus may include any number of interconnected buses and bridges.
  • the bus connects one or more processors and various circuits of the memory together.
  • the bus can also connect various other circuits such as peripheral devices, voltage regulators, power management circuits, etc., which are all well-known in the art, and therefore, no further description will be given here.
  • the bus interface provides an interface between the bus and the transceiver.
  • the transceiver may be one element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices on the transmission medium.
  • the data processed by the processor 801 is transmitted on the wireless medium through the antenna, and further, the antenna also receives the data and transmits the data to the processor 801.
  • the processor 801 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions.
  • the memory 802 may be used to store data used by the processor 801 when performing operations.
  • the sixth embodiment of the present application relates to an active control system, as shown in FIG. 9: including the active noise reduction chip 901 and the infinite impulse response IIR filter 902 as described in the fifth embodiment; the active noise reduction chip 901 Used to obtain the time domain coefficients of the IIR filter 902, and update the time domain coefficients of the IIR filter to the IIR filter 902; the IIR filter 902 is located on the secondary channel and is used for The environmental noise signal of the secondary channel is filtered to output a noise reduction signal.
  • the seventh embodiment of the present invention relates to a computer-readable storage medium storing a computer program.
  • the computer program is executed by the processor, the above method embodiment is realized.
  • the program is stored in a storage medium and includes several instructions to enable a device ( It may be a single-chip microcomputer, a chip, etc.) or a processor (processor) to execute all or part of the steps of the methods described in the embodiments of the present application.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program code .

Landscapes

  • Soundproofing, Sound Blocking, And Sound Damping (AREA)
  • Filters That Use Time-Delay Elements (AREA)

Abstract

本申请部分实施例提供了一种主动降噪方法、装置、芯片、主动控制系统和存储介质。上述主动降噪方法包括：根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数(201)；将主通道对应的自适应滤波器的频域系数转换为IIR滤波器的时域系数(202)；其中，IIR滤波器的阶数小于自适应滤波器的阶数；将IIR滤波器的时域系数更新到IIR滤波器中(203)；其中，所述IIR滤波器位于所述次级通道上，用于对通过所述次级通道的环境噪声信号进行滤波输出降噪信号。采用本申请的实施例，可以在实现自适应主动降噪的同时，降低降噪过程中的计算复杂度、减少硬件资源的开销，有利于硬件的实现。

Description

主动降噪方法、装置、芯片、主动控制系统和存储介质 技术领域
本申请涉及主动降噪技术领域，特别涉及一种主动降噪方法、装置、芯片、主动控制系统和存储介质。
背景技术
现实生活中，由于说话人经常处于各种嘈杂的环境中，环境噪声已严重影响到出行的质量。在诸如免提通信、助听器、耳机和汽车电子之类的许多降噪技术应用中，通常需要降噪技术来降低环境噪声水平。目前，传统自适应主动降噪控制方案都是基于横向结构的滤波器FIR(Finite Impulse Response，有限脉冲响应)与LMS(Least mean square，最小均方根)算法结合来收敛到主通道的传递函数曲线。
然而，发明人发现，相关技术中为了达到更好的收敛效果，通常需要采取高阶的FIR滤波器，这需要足够多的硬件资源。而且，由于高阶滤波器系数是在时域进行迭代，并且要求每一个采样点迭代一次，而主动降噪系统工作的采样率非常高，这就使得算法的计算量很大，不利于硬件实现。
发明内容
本申请部分实施例的目的在于提供一种主动降噪方法、装置、芯片、主 动控制系统和存储介质，可以在实现自适应主动降噪的同时，降低降噪过程中的计算复杂度、减少硬件资源的开销，有利于硬件的实现。
本申请实施例提供了一种主动降噪方法，包括：根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数；其中，所述误差信号为所述环境噪声信号分别通过主通道和次级通道后在目标降噪点叠加后的信号；将所述主通道对应的自适应滤波器的频域系数转换为无限脉冲响应IIR滤波器的时域系数；其中，所述IIR滤波器的阶数小于所述自适应滤波器的阶数；将所述IIR滤波器的时域系数更新到所述IIR滤波器中；其中，所述IIR滤波器位于所述次级通道上，用于对通过所述次级通道的环境噪声信号进行滤波输出降噪信号。
本申请实施例还提供了一种主动降噪装置，包括：频域自适应滤波模块，用于根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数；其中，所述误差信号为所述环境噪声信号分别通过主通道和次级通道后在目标降噪点叠加后的信号；IIR滤波器设计模块与所述频域自适应滤波模块连接，用于将所述主通道对应的自适应滤波器的频域系数转换为无限脉冲响应IIR滤波器的时域系数，并将所述IIR滤波器的时域系数更新到所述IIR滤波器中；其中，所述IIR滤波器的阶数小于所述自适应滤波器的阶数；所述IIR滤波器与所述系数转换模块连接且位于所述次级通道上，用于对通过所述次级通道的环境噪声信号进行滤波输出降噪信号。
本申请实施例还提供了一种主动降噪芯片，包括：至少一个处理器；以及，与所述至少一个处理器通信连接的存储器；其中，所述存储器存储有可被 所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行上述的主动降噪方法。
本申请实施例还提供了一种主动控制系统，包括上述的主动降噪芯片和无限脉冲响应IIR滤波器；所述主动降噪芯片用于获取所述IIR滤波器的时域系数，并将所述IIR滤波器的时域系数更新到所述IIR滤波器中；所述IIR滤波器位于次级通道上，用于对通过所述次级通道的环境噪声信号进行滤波输出降噪信号。
本申请实施例还提供了一种计算机可读存储介质，存储有计算机程序，其特征在于，所述计算机程序被处理器执行时实现上述的主动降噪方法。
本申请实施例现对于现有技术而言，根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数，将主通道对应的自适应滤波器的频域系数转换为IIR滤波器的时域系数，将IIR滤波器的时域系数更新到IIR滤波器中。即在频域进行自适应滤波，可以有效避免相关技术中在时域上每个采样点都更新滤波器系数，导致大量的数据交互、计算复杂度高的问题，有利于降低算法资源开销，并且在频域获取自适应滤波器的频域系数可以充分利用采集到的信号的所有信息，更有利于得到收敛于主通道的传递函数的IIR滤波器时域系数，从而使得环境噪声信号分别通过主通道和次级通道后在目标降噪点叠加后的信号更趋近于0，有利于更准确的降噪。同时，本申请实施例中使用了相对自适应滤波器的阶数更低的低阶IIR滤波器，这样可以在实现自适应主动降噪的同时，有效的避免需要高阶滤波器，降低硬件资源开销，有利于硬件实现。
例如，根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自 适应滤波，获取主通道对应的自适应滤波器的频域系数，包括：将所述环境噪声信号转换为环境噪声频域信号，并将所述误差信号转换为误差频域信号；根据所述环境噪声频域信号对所述误差频域信号进行误差幅度控制；根据误差幅度控制后的误差频域信号和所述环境噪声频域信号，获取主通道对应的自适应滤波器的频域系数。本申请实施例中，对误差频域信号进行误差幅度控制，可以避免由于佩戴不标准导致在目标降噪点采集到的误差信号很大的时候，造成滤波器的系数的发散。由于不同的佩戴状态可能引起在目标降噪点采集到的误差信号发生或大或小的变化，本申请实施例中对误差频域信号进行误差幅度控制，保证在不同的佩戴状态下均可以得到自适应收敛于主通道的传递函数的滤波器的时域系数，更具鲁棒性，更加适用于各种佩戴场景中，保证较好的降噪能力。
例如，根据所述环境噪声频域信号对所述误差频域信号进行误差幅度控制，包括：根据所述环境噪声频域信号对所述误差频域信号进行归一化处理，获取归一化处理后的误差频域信号；根据所述归一化处理后的误差频域信号的幅值与预设阈值的比较结果，获取误差限幅系数；根据所述误差限幅系数获取误差幅度控制后的误差频域信号。本申请实施例提供了一种具体的误差幅度控制方式，有利于更好的对误差频域信号进行误差幅度控制。
例如，根据所述环境噪声频域信号对所述误差频域信号进行归一化处理，获取归一化处理后的误差频域信号，包括：分别获取所述环境噪声频域信号的功率谱和所述误差频域信号的幅度谱；将所述幅度谱和所述功率谱的比值作为所述归一化处理后的误差频域信号的幅度谱。本申请实施例提供了一种具体的归一化处理的方式，方便了获取归一化处理后的误差信号的幅度谱。
例如，根据所述归一化处理后的误差频域信号的幅值与预设阈值的比较结果，获取误差限幅系数，包括：通过以下公式计算所述误差限幅系数：
Figure PCTCN2019090912-appb-000001
其中，所述Ef(k)为所述误差限幅系数，所述|E1(k)|为所述归一化处理后的误差频域信号的幅值，所述τ(k)为所述预设阈值，所述k为所述误差频域信号的频点。本申请实施例提供了一种误差限幅系数的具体计算公式，使得可以很好的避免归一化处理后的误差频域信号的幅值过大。
例如，将所述主通道对应的自适应滤波器的频域系数转换为无限脉冲响应IIR滤波器的时域系数，包括：将所述主通道对应的自适应滤波器的频域系数转换为所述主通道对应的自适应滤波器的时域系数；根据LMS算法将所述主通道对应的自适应滤波器的时域系数转换为所述IIR滤波器的时域系数，方便了将自适应滤波器的频域系数转换为IIR滤波器的时域系数。
例如，所述根据LMS算法将所述主通道对应的自适应滤波器的时域系数转换为所述IIR滤波器的时域系数，包括：将主通道对应的自适应滤波器的时域系数进行翻转；生成随机数，并根据翻转后的所述时域系数和所述随机数获取期望信号；根据LMS算法和所述期望信号得到所述IIR滤波器的时域系数。利用随机数有利于得到更稳定的IIR滤波器的时域系数，且可以使得到的IIR滤波器的时域系数收敛性更好。
例如，根据LMS算法将所述主通道对应的自适应滤波器的时域系数转换为所述IIR滤波器的时域系数，包括根据LMS算法将所述主通道对应的自适应滤波器的时域系数转换为m阶IIR滤波器的时域系数；将所述m阶IIR滤波器的时域系数转换为多个n阶IIR滤波器的时域系数；其中，所述m大于所述n， 所述多个n阶IIR滤波器均对应有各自的时域系数；所述将所述IIR滤波器的时域系数更新到所述IIR滤波器中，具体为：将所述多个n阶IIR滤波器的时域系数更新到所述多个n阶IIR滤波器中。本申请实施例，通过将m阶IIR滤波器的时域系数转换为多个n阶IIR滤波器的时域系数，即将IIR滤波器的高阶系数转变为多个低阶系数，通过次级通道上多个n阶IIR滤波器对信号的处理效果有利于更好的逼近主通道的传递函数对信号的处理效果，使得环境噪声信号分别通过主通道和次级通道后在目标降噪点叠加后的信号更加趋近于0，有利于达到更好的降噪效果。
例如，在所述根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数之前，还包括：自适应更新所述次级通道对应的自适应滤波器的系数；将所述次级通道对应的自适应滤波器的系数更新到所述次级通道对应的自适应滤波器中；所述根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波中的环境噪声信号为通过所述次级通道对应的自适应滤波器的环境噪声信号。本申请实施例可以使得当次级通道的传递函数随环境因素的变化而变化，即次级通道对环境噪声信号的影响随环境因素的变化而变化时，也能根据次级通道的传递函数的变化，自适应更新次级通道对应的自适应滤波器的系数，从而获取到收敛于主通道的传递函数的自适应滤波器的频域系数，进一步得到收敛于主通道的传递函数的IIR滤波器的时域系数。
例如，所述自适应更新所述次级通道对应的自适应滤波器的系数，具体为：自适应更新所述次级通道对应的自适应滤波器的频域系数。在频域进行次级通道对应的自适应滤波器的频域系数的自适应更新，可以有效避免相关技术 中在时域上每个采样点都更新，导致大量的数据交互、计算复杂度高的问题，有利于降低算法资源开销。
例如，自适应更新所述次级通道对应的自适应滤波器系数，包括：在所述次级通道上引入预设的白噪声信号，并根据所述白噪声信号和调整后的所述误差信号，获取所述次级通道对应的自适应滤波器的系数；其中，调整后的所述误差信号为所述白噪声信号经预设滤波器处理后的信号与所述误差信号叠加后的信号；所述预设滤波器的系数与所述次级通道对应的自适应滤波器的系数相同。本申请实施例提供了一种自适应更新所述次级通道对应的滤波器系数的具体方式，可以更方便和准确的自适应更新次级通道对应的自适应滤波器系数。
附图说明
一个或多个实施例通过与之对应的附图中的图片进行示例性说明，这些示例性说明并不构成对实施例的限定，附图中具有相同参考数字标号的元件表示为类似的元件，除非有特别申明，附图中的图不构成比例限制。
图1是根据本申请第一实施例中的主动降噪方法的框图；
图2是根据本申请第一实施例中的主动降噪方法的流程图；
图3是根据本申请第一实施例中的步骤201的实现过程的流程图；
图4是根据本申请第二实施例中的步骤201的实现过程的流程图；
图5是根据本申请第三实施例中的主动降噪方法的框图；
图6是根据本申请第三实施例中的主动降噪方法的流程图；
图7是根据本申请第四实施例中的主动降噪装置的示意图；
图8是根据本申请第五实施例中的主动降噪芯片的结构示意图；
图9是根据本申请第六实施例中的主动控制系统的示意图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请部分实施例进行进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。以下各个实施例的划分是为了描述方便，不应对本申请的具体实现方式构成任何限定，各个实施例在不矛盾的前提下可以相互结合相互引用。
本申请第一实施例涉及一种主动降噪方法，该方法可以适用于语音交互产品中，包括但不限于耳机、助听器等电子设备。本申请实施例中主要采用基于频域自适应滤波和IIR(Infinite Impulse Response，无限脉冲响应)滤波器相结合的技术来自适应更新次级通道上的IIR滤波器的时域系数，进而实现主动降噪。下面对本实施方式的主动降噪方法的实现细节进行具体的说明，以下内容仅为方便理解提供的实现细节，并非实施本方案的必须。
下面以主动降噪方法应用于降噪耳机为例进行具体说明，但在实际应用中并不以此为限。降噪耳机中一般在两个位置处设置有麦克风，耳机外壳处设置有外部麦克风，外部麦克风可用于采集环境噪声信号，戴上耳机贴近人耳处的位置设置有人耳处麦克风，人耳处麦克风可用于采集误差信号。参考主动降噪方法的框图即图1，可知外部麦克风采集的环境噪声信号即为x(n)，人耳处麦克风采集到的误差信号即为e(n)。由图1可以看出，e(n)＝d(n)-y(n)，其中d(n)为环境噪声信号x(n)通过主通道后得到的期望信号，即可认为主通道的传 递函数P(n)对x(n)进行了信号处理后得到的信号，主通道可以理解为环境噪声信号通过空气介质传递到人耳处的传输路径，环境噪声信号通过主通道到达人耳处时，主通道对环境噪声信号的影响可以抽象为主通道的传递函数P(n)，P(n)一般受到耳机结构和佩戴准确度的影响。图1中的主动控制系统位于次级通道上，需要说明的是，在具体实现中图1中的次级通道上的硬件组成除了IIR滤波器通常还包括模数转换器、扬声器等。本实施例中假设次级通道上的所有的硬件对环境噪声信号的影响不会随着温度等环境因素的变化而变化，通常次级通道上所有的硬件对环境噪声信号的影响可以抽象为次级通道上所有硬件的传递函数。本实施例中次级通道上所有硬件的传递函数可以理解为1或是其他固定不变的数值。y(n)为环境噪声信号x(n)通过次级通道后得到的输出信号，即可认为次级通道上的IIR滤波器对x(n)进行滤波处理后得到的信号。IIR滤波器输出的y(n)可以激励耳机中的扬声器发出降噪信号y(n)，该扬声器发出的该降噪信号y(n)与期望信号d(n)在空气中叠加，人耳处麦克风可以采集到叠加后的误差信号。如果降噪信号y(n)与期望信号d(n)叠加后的信号趋近于0，可以认为达到了较好的降噪效果。
本实施方式中的主动降噪方法的具体流程可以如图2所示，包括：
步骤201：根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数。
具体的说，目标降噪点可以为耳机中的人耳处麦克风所在的位置，耳机中可以设置有如图1中的主动控制系统，主动控制系统中包括频域自适应滤波模块102和IIR滤波器设计模块103。假设主动控制系统的输入数据流采样率为Fs＝192KHz，为了降低功耗和硬件资源，主动控制系统可以间接性从存储器 中取1024个点来进行频域自适应滤波。在具体实现中，步骤201可以由频域自适应滤波模块102来实现，频域自适应滤波模块102根据环境噪声信号和误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数的具体实现方式可以如图3所示，包括：
步骤301：对环境噪声信号和误差信号分别进行频域转换。
具体的说，可以将环境噪声信号和误差信号数据通过傅里叶变换FFT边还到频域，通过以下公式可以实现对环境噪声信号和误差信号的频域转换：
Figure PCTCN2019090912-appb-000002
Figure PCTCN2019090912-appb-000003
其中，k代表频点，x(n)为外部麦克风采集到的环境噪声信号，e(n)人耳处麦克风采集到的误差信号，N表示FFT的点数，本实施例中N的取值可以为1024，U(k)为将x(n)频域转换后的环境噪声频域信号，E(k)为将e(n)频域转换后的误差频域信号。需要说明的是，本实施方式以及以下各实施方式中提到的环境噪声信号和误差信号均为时域信号，环境噪声频域信号和误差频域信号均为频域信号。
步骤302：根据环境噪声频域信号和误差频域信号计算频域梯度。
具体的说，可以通过如下公式计算频域梯度
Figure PCTCN2019090912-appb-000004
Figure PCTCN2019090912-appb-000005
其中，μ为迭代步长，步长的选取和收敛速度有关，如果步长选取的太大，虽然收敛很快，但是可能造成滤波器系数发散；如果步长选取太小，收敛速度太慢了。在具体实现中步长的选取有多种方法，有固定步长和变步长，本申请实 施例中步长的选取以μ＝1为示例性说明。U(k) *表示环境噪声频域信号的共轭信号。
步骤303：将频域梯度转换为时域梯度。
具体的说，可以通过傅里叶逆变换IFFT将频域梯度转换为时域梯度，即可以通过如下公式将频域梯度转换为时域梯度
Figure PCTCN2019090912-appb-000006
Figure PCTCN2019090912-appb-000007
步骤304：对时域梯度进行补零。
具体的说，可以通过如下公式对时域梯度进行补零，得到补零后的时域梯度φ(n)：
Figure PCTCN2019090912-appb-000008
步骤305：对补零后的时域梯度进行频域转换。
具体的说，可以通过傅里叶变换对补零后的时域梯度进行频域转换，具体可以参考如下公式：
Figure PCTCN2019090912-appb-000009
其中，ΔW(k)为FIR滤波器的频域系数的变化量。
步骤306：根据频域转换后的时域梯度获取主通道对应的自适应滤波器的频域系数。
需要说明的是，本实施例中只是从软件层面得到主通道对应的自适应滤波器的频域系数，而并不一定存在自适应滤波器这一硬件设备。通常的，本领域中认为存在频域系数的自适应滤波器属于有限脉冲响应FIR滤波器，简称为FIR滤波器；本实施方式及以下各实施方式均以FIR滤波器这一称呼进行具体 说明，也就是说，主通道对应的自适应滤波器的频域系数可以称为FIR滤波器的频域系数，然并不以此为限。
步骤306中，具体可以通过如下公式获取FIR滤波器的频域系数：
Figure PCTCN2019090912-appb-000010
其中，
Figure PCTCN2019090912-appb-000011
表示第λ+1次估计得到的k频点的FIR滤波器的频域系数，
Figure PCTCN2019090912-appb-000012
为第λ次估计得到的k频点的FIR滤波器的频域系数，ΔW λ+1(k)表示第λ+1次估计得到的k频点的FIR滤波器的频域系数的变化量。
步骤202：将主通道对应的自适应滤波器的频域系数转换为IIR滤波器的时域系数。
具体的说，IIR滤波器的阶数小于主通道对应的自适应滤波器的阶数，即IIR滤波器的阶数小于FIR滤波器的阶数。比如说FIR滤波器的阶数可以为3000阶，IIR滤波器的阶数可以为6阶，本实施方式中可以将阶数为3000阶的FIR滤波器的频域系数转换为阶数为6阶的IIR滤波器的时域系数。本步骤可以通过图1中IIR滤波器设计模块103实现。
在一个例子中，将FIR滤波器的频域系数转换为IIR滤波器的时域系数的方式可以为：首先可以对通过频域自适应滤波模块102得到的FIR滤波器频域系数
Figure PCTCN2019090912-appb-000013
作傅里叶逆变换IFFT以后得到FIR滤波器的时域系数w(n+1)，然后将w(n+1)向量翻转得到新的
Figure PCTCN2019090912-appb-000014
随后生成一组随机数x(n+1)，根据随机数x(n+1)和
Figure PCTCN2019090912-appb-000015
得到期望信号d(n+1)，比如将随机数与
Figure PCTCN2019090912-appb-000016
相乘，得到期望信号d(n+1)，最后根据期望信号d(n+1)和最小均方根LMS算法得到IIR滤波器的时域系数。其中，IIR滤波器的时域系数可以包括分子系数a(n)和分母系数b(n)，滤波器时域系数为
Figure PCTCN2019090912-appb-000017
的FIR滤波器的频率响应与滤波器 时域系数为a(n)/b(n)的IIR滤波器的频率响应相同。需要说明的是，本领域技术人员基于已有的相关技术可以实现根据期望信号d(n+1)和最小均方根LMS算法得到IIR滤波器的时域系数，因此本实施方式对具体如何得到IIR滤波器的时域系数的具体实现方式不做具体描述。
在一个例子中，将FIR滤波器的频域系数转换为IIR滤波器的时域系数的方式可以为：利用普罗尼算法Prony Method实现系数的转换。
需要说明的是，本实施例中只是以上述两种系数转换方式为例，在具体实现中任何可以将FIR滤波器的频域系数转换为IIR滤波器的时域系数的方法均在本实施例保护范围之内。
在一个例子中，主动控制系统中的IIR滤波器可以为1个，即如图1中所示的情况。在这种情况下，将FIR滤波器的频域系数转换为IIR滤波器的时域系数的方式可以参考上面的描述。
在另一个例子中，主动控制系统中可以包括多个IIR滤波器，多个IIR滤波器为串联的形式，在这种情况下，将FIR滤波器的频域系数转换为IIR滤波器的时域系数的方式可以为：首先，根据LMS算法将FIR滤波器的时域系数转换为m阶IIR滤波器的时域系数，其中，将FIR滤波器的时域系数转换为m阶IIR滤波器的时域系数的方式可以为上文中描述过的：将主通道对应的FIR滤波器的时域系数进行翻转，生成随机数，并根据翻转后的时域系数和随机数获取期望信号，根据LMS算法和期望信号得到m阶IIR滤波器的时域系数。需要说明的是，本实施例只是提供一个示例，在具体实现中将FIR滤波器的时域系数转换为m阶IIR滤波器的时域系数的方式并不以上述的方式为限。然后，将m阶IIR滤波器的时域系数转换为多个n阶IIR滤波器的时域系数；其中， m大于n，多个n阶IIR滤波器均对应有各自的时域系数。比如说，m为6，n为2，即先将FIR滤波器的时域系数转换为6阶IIR滤波器的时域系数，然后将6阶IIR滤波器的时域系数转换成3个2阶IIR滤波器的时域系数，最后得到每个2阶IIR滤波器的时域系数的分子和分母。具体的将6阶IIR滤波器的时域系数转换成3个2阶IIR滤波器的时域系数的实现方式可以参考如下公式：
Figure PCTCN2019090912-appb-000018
Figure PCTCN2019090912-appb-000019
Figure PCTCN2019090912-appb-000020
Figure PCTCN2019090912-appb-000021
其中，a ij表示第i个2阶IIR滤波器分子的第j个可调系数，b ij表示第i个2阶滤波器分母的第j个可调系数。
步骤203：将IIR滤波器的时域系数更新到IIR滤波器中。
具体的说，如果主动控制系统中包括一个IIR滤波器，则将获取的这一个IIR滤波器的时域系数写入该IIR滤波器中，如果主动控制系统中包括多个IIR滤波器，则将获取的多个IIR滤波器的时域系数分别写入多个IIR滤波器中。IIR滤波器位于次级通道上，可参考图1中的次级通道上的IIR滤波器104，IIR滤波器104被写入滤波器系数后可以对通过次级通道的环境噪声信号进行滤波输出降噪信号。需要说明的是，通常IIR滤波器具备初始时域系数，第一次更新时，可以认为是对IIR滤波器的初始时域系数进行更新。在具体实现中，IIR滤波器104输出的降噪信号可以激励耳机中的扬声器发出降噪信号，使得位于 目标降噪点处的人耳处麦克风可以采集到扬声器发出降噪信号y(n)与期望信号d(n)的叠加信号即e(n)，由于本实施方式中的IIR滤波器的时域系数收敛于主通道的传递函数P(n)，因此有利于使IIR滤波器输出与期望信号d(n)频谱分布相同、幅值大小相同且相位相反的降噪信号y(n)，从而使得人耳处麦克风采集到的误差信号e(n)无限趋近于0，不断根据采集到的误差信号e(n)和环境噪声信号x(n)在频域进行自适应滤波，有利于达到很好的降噪效果。
本实施例相对于现有技术而言，根据环境噪声信号和误差信号进行频域自适应滤波，可以有效避免相关技术中在时域上每个采样点都更新滤波器系数，导致大量的数据交互、计算复杂度高的问题，有利于降低算法资源开销，并且在频域获取自适应滤波器的频域系数可以充分利用采集到的信号的所有信息，更有利于得到收敛于主通道的传递函数的IIR滤波器时域系数，从而使得环境噪声信号分别通过主通道和次级通道后在目标降噪点叠加后的信号更趋近于0，有利于更准确的降噪。同时，本申请实施例中使用了相对自适应滤波器的阶数更低的低阶IIR滤波器，这样可以在实现自适应主动降噪的同时，有效的避免需要高阶自适应滤波器，降低硬件资源开销，有利于硬件实现。
本申请第二实施例涉及一种主动降噪方法，本实施例是对第一实施方式的进一步改进，主要改进之处在于，根据环境噪声信号和误差信号进行频域自适应滤波，获取自适应滤波器的频域系数的过程中对误差频域信号进行了误差幅度控制，有利于避免当人耳处采集到的误差信号很大的时候，造成滤波器系数的发散，并能适应于不同的佩戴状态。
本实施方式中，根据环境噪声信号和误差信号进行频域自适应滤波，获取自适应滤波器的频域系数即步骤201的实现方式，可以为根据环境噪声频域 信号对误差频域信号进行误差幅度控制，根据误差幅度控制后的误差频域信号和环境噪声频域信号，获取自适应滤波器的频域系数。具体的，本实施例步骤201的实现方式可以如图4所示，包括：
步骤401：对环境噪声信号和误差信号分别进行频域转换。
步骤401与第一实施例中步骤301大致相同，为避免重复此处不再赘述。
步骤402：根据环境噪声频域信号对误差频域信号进行归一化处理，获取归一化处理后的误差频域信号。
具体的说，首先可以分别获取环境噪声频域信号的功率谱和误差频域信号的幅度谱，其中，环境噪声频域信号的功率谱可以通过以下公式计算得到：
P i(k)＝γ·P i-1(k)+(1-γ)·|U(k)| 2
其中，P i(k)表示i时刻环境噪声频域信号的功率谱，γ是一个取值范围为0<γ<1的遗忘因子，用来控制上式迭代过程中的有效记忆，U(k)为环境噪声频域信号。误差频域信号的幅度谱可以根据误差频域信号的表达式计算得到。然后，可以将幅度谱和功率谱的比值作为归一化处理后的误差频域信号。具体可以通过以下公式得到归一化处理后的误差频域信号的幅度谱E1(k)：
Figure PCTCN2019090912-appb-000022
其中，E(k)为误差频域信号的幅度谱，P(k)为环境噪声频域信号的功率谱。
步骤403：根据归一化处理后的误差频域信号的幅值与预设阈值的比较结果，获取误差限幅系数。
具体的说，可以通过以下公式计算误差限幅系数Ef(k)：
Figure PCTCN2019090912-appb-000023
其中，所述|E1(k)|为归一化处理后的误差频域信号的幅值，所述τ(k)为预设阈值，所述k为所述误差频域信号的频点。从上述公式可以看出，如果|E1(k)|＞τ(k)，说明归一化处理后的误差频域信号的幅值较大，则
Figure PCTCN2019090912-appb-000024
利用该误差限幅系数对归一化处理后的误差频域信号的幅值进行限幅处理，避免当人耳处采集到的误差信号很大的时候，造成滤波器系数的发散。如果|E1(k)|＜τ(k)，说明归一化处理后的误差频域信号的幅值较小，则
Figure PCTCN2019090912-appb-000025
即无需进行限幅处理，人耳处采集到的误差信号较小。
另外，τ(k)可以预先设置有多个值，不同频率值范围对应不同的预设阈值。在一个例子中，预设阈值的取值方式为：
Figure PCTCN2019090912-appb-000026
其中，所述f为误差频域信号的频率值，所述k为误差频域信号的频点。由于相同频率下基于不同形式的傅里叶变换将时域信号转换为频域信号，频谱点数会有所差异，因此采用频率值作为预设阈值取值的参考量，有利于适应不同形式的傅里叶变换。另外，不同频率值范围对应不同的预设阈值的具体设定方式，本领域技术人员可以基于人耳对不同频率的声音的敏感度而设定，此处仅为一个例子，不作任何限定。本申请实施例中提供的上述的预设阈值的取值方式，方便了根据不同的频率范围选取预设阈值。
步骤404：根据误差限幅系数获取误差幅度控制后的误差频域信号。
具体的说，误差幅度控制后的误差频域信号可以为误差限幅系数与步骤401中得到的误差频域信号的乘积，即误差幅度控制后的误差频域信号可以为： E(k)·Ef(k)。
步骤405：根据环境噪声频域信号和误差幅度控制后的误差频域信号计算频域梯度。
具体的说，可以通过如下公式计算频域梯度：
Figure PCTCN2019090912-appb-000027
其中，μ为迭代步长，U(k) *表示环境噪声频域信号的共轭信号，E(k)·Ef(k)表示误差幅度控制后的误差频域信号。
步骤406：将频域梯度转换为时域梯度。
步骤407：对时域梯度进行补零。
步骤408：对补零后的时域梯度进行频域转换。
步骤409：根据频域转换后的时域梯度获取FIR滤波器的频域系数。
步骤406至步骤409与第一实施方式中步骤303至步骤306大致相同，为避免重复，在此不再一一赘述。
与现有技术相比，本实施方式中对误差频域信号进行误差幅度控制，可以避免在目标降噪点采集到的误差信号很大的时候，造成滤波器的系数的发散。由于不同的佩戴状态可能引起在目标降噪点采集到的误差信号发生或大或小的变化，本申请实施例中对误差频域信号进行误差幅度控制，保证在不同的佩戴状态下均可以得到自适应收敛到主通道的传递函数的滤波器的时域系数，更具鲁棒性，更加适用于各种佩戴场景中，保证较好的降噪能力。
本申请第三实施例涉及一种主动降噪方法，本申请第一、二实施例中均假设次级通道为理想通道，即次级通道上所有硬件的传递函数为1或是其他固定不变的数值。本申请第三实施例中主要介绍次级通道不是理想通道的情况。 不是理想通道可以理解为次级通道上上所有硬件的传递函数会发生变化，比如说次级通道上的模数转换器、IIR滤波器等硬件的处理能力可能会随着温度等环境因素的变化，从而导致次级通道上所有硬件的传递函数会发生变化。
本实施例中的主动降噪方法的框图可以如图5所示：与图1相比，主要区别是图5中多设计了一级次级通道自适应滤波模块502，本实施方式中的主动降噪方法的流程图如图6所示，包括：
步骤601：自适应更新次级通道对应的自适应滤波器的系数，并将次级通道对应的自适应滤波器的系数更新到次级通道对应的自适应滤波器中。
具体的说，可以通过次级通道自适应滤波模块502来自适应更新次级通道对应的自适应滤波器的系数。需要说明的是，本实施例中只是从软件层面得到次级通道对应的自适应滤波器的系数，次级通道上并不一定存在自适应滤波器这一硬件设备，次级通道对应的自适应滤波器可以为图5中位于次级通道上的
Figure PCTCN2019090912-appb-000028
Figure PCTCN2019090912-appb-000029
505可以通过软件实现，即通过软件对x(n)进行系数处理，将x(n)与滤波器系数
Figure PCTCN2019090912-appb-000030
相乘后输入主通道自适应滤波模块505中的频域自适应滤波模块102中。
本实施方式中，更新次级通道对应的自适应滤波器系数的方式可以为：在次级通道上引入预设的白噪声信号，并根据白噪声信号和调整后的误差信号进行频域自适应滤波，获取次级通道对应的自适应滤波器的系数；其中，调整后的误差信号为白噪声信号经预设滤波器处理后的信号与误差信号叠加后的信号；预设滤波器的系数与次级通道对应的自适应滤波器的系数相同，预设滤波器可以为图5中的
Figure PCTCN2019090912-appb-000031
Figure PCTCN2019090912-appb-000032
也可以理解为次级通道上所有硬件的预估传递函数。
其中，白噪声信号可以由图5中的白噪声模块503发出，图5中频域自适应滤波模块507根据白噪声信号和调整后的误差信号进行频域自适应滤波，获取次级通道对应的自适应滤波器的系数，然后将获取的次级通道对应的自适应滤波器的系数写入图5中的
Figure PCTCN2019090912-appb-000033
和
Figure PCTCN2019090912-appb-000034
中。可以认为在当前时刻频域自适应滤波模块507在进行频域自适应滤波时，
Figure PCTCN2019090912-appb-000035
和
Figure PCTCN2019090912-appb-000036
中写入的系数为上一时刻频域自适应滤波模块507计算得到的系数。其中，
Figure PCTCN2019090912-appb-000037
用来抵消引入白噪声信号后白噪声信号与S(z)504的乘积对误差信号的影响，S(z)504为次级通道上所有硬件的传递函数。可以理解的是，如果次级通道上所有硬件的预估传递函数
Figure PCTCN2019090912-appb-000038
越收敛于次级通道上所有硬件的传递函数S(z)504，抵消的效果越好，理想情况下，调整后的误差信号中白噪声信号与S(z)504的乘积完全被白噪声信号与
Figure PCTCN2019090912-appb-000039
的乘积抵消掉，即调整后的误差信号中不存在白噪声信号的影响。
可以理解的是，本实施方式中根据白噪声信号和调整后的误差信号进行频域自适应滤波，获取次级通道对应的自适应滤波器的系数的具体方式与第一或第二实施方式中，根据环境噪声信号和误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数的具体方式大致相同，为避免重复在此不再一一赘述。
另外，在具体实现中还可以在时域基于FXLMS(Filtered-X Least Mean Square，滤波型最小均方根)算法来实现更新次级通道对应的自适应滤波器的系数。
在一个例子中，自适应更新次级通道对应的自适应滤波器的系数，可以具体为：自适应更新次级通道对应的自适应滤波器的频域系数。也就是说，在 频域进行次级通道对应的自适应滤波器的频域系数的自适应更新，可以有效避免相关技术中在时域上每个采样点都更新，导致大量的数据交互、计算复杂度高的问题，有利于降低算法资源开销。
步骤602：根据经次级通道对应的滤波器处理的环境噪声信号和调整后的误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数。
步骤603：将主通道对应的自适应滤波器的频域系数转换为IIR滤波器的时域系数。
步骤604：将IIR滤波器的时域系数更新到IIR滤波器中。
步骤602至步骤604与第一实施方式中步骤201至步骤203大致相同，为避免重复此处不再一一赘述。
与现有技术相比，本申请实施例，可以使得当次级通道上所有硬件的传递函数会发生变化时，也能获取到收敛于次级通道的传递函数的FIR滤波器的频域系数，从而得到收敛于次级通道的传递函数的IIR滤波器的时域系数，以更精确的实现主动降噪。
上面各种方法的步骤划分，只是为了描述清楚，实现时可以合并为一个步骤或者对某些步骤进行拆分，分解为多个步骤，只要包括相同的逻辑关系，都在本专利的保护范围内；对算法中或者流程中添加无关紧要的修改或者引入无关紧要的设计，但不改变其算法和流程的核心设计都在该专利的保护范围内。
本申请第四实施例涉及一种主动降噪装置，如图7所示，包括：频域自适应滤波模块701，用于根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数；其中，所述误差信号为所述环境噪声信号分别通过主通道和次级通道后在目标降噪点叠 加后的信号；IIR滤波器设计模块702与所述频域自适应滤波模块703连接，用于将所述主通道对应的自适应滤波器的频域系数转换为无限脉冲响应IIR滤波器的时域系数，并将所述IIR滤波器的时域系数更新到所述IIR滤波器中；其中，所述IIR滤波器的阶数小于所述自适应滤波器的阶数；所述IIR滤波器703与所述IIR滤波器设计模块702连接且位于所述次级通道上，用于对通过所述次级通道的环境噪声信号进行滤波输出降噪信号。
不难发现，本实施实施例为与第一至三实施例相对应的装置实施例，本实施例可与第一实施例互相配合实施。第一实施例中提到的相关技术细节在本实施例中依然有效，为了减少重复，这里不再赘述。相应地，本实施例中提到的相关技术细节也可应用在第一实施例中。
值得一提的是，为了突出本发明的创新部分，本实施方式中并没有将与解决本发明所提出的技术问题关系不太密切的单元引入，但这并不表明本实施方式中不存在其它的单元。
本申请第五实施例涉及一种主动降噪芯片，如图8所示，包括：至少一个处理器801；以及，与所述至少一个处理器801通信连接的存储器802；其中，所述存储器802存储有可被所述至少一个处理器801执行的指令，所述指令被所述至少一个处理器801执行，以使所述至少一个处理器801能够执行上述主动降噪方法。
其中，存储器802和处理器801采用总线方式连接，总线可以包括任意数量的互联的总线和桥，总线将一个或多个处理器和存储器的各种电路连接在一起。总线还可以将诸如外围设备、稳压器和功率管理电路等之类的各种其他电路连接在一起，这些都是本领域所公知的，因此，本文不再对其进行进一步 描述。总线接口在总线和收发机之间提供接口。收发机可以是一个元件，也可以是多个元件，比如多个接收器和发送器，提供用于在传输介质上与各种其他装置通信的单元。经处理器801处理的数据通过天线在无线介质上进行传输，进一步，天线还接收数据并将数据传送给处理器801。
处理器801负责管理总线和通常的处理，还可以提供各种功能，包括定时，外围接口，电压调节、电源管理以及其他控制功能。而存储器802可以被用于存储处理器801在执行操作时所使用的数据。
本申请第六实施例涉及一种主动控制系统，如图9所示：包括如第五实施方式中所述的主动降噪芯片901和无限脉冲响应IIR滤波器902；所述主动降噪芯片901用于获取所述IIR滤波器902的时域系数，并将所述IIR滤波器的时域系数更新到所述IIR滤波器902中；所述IIR滤波器902位于次级通道上，用于对通过所述次级通道的环境噪声信号进行滤波输出降噪信号。
本发明第七实施方式涉及一种计算机可读存储介质，存储有计算机程序。计算机程序被处理器执行时实现上述方法实施例。
即，本领域技术人员可以理解，实现上述实施例方法中的全部或部分步骤是可以通过程序来指令相关的硬件来完成，该程序存储在一个存储介质中，包括若干指令用以使得一个设备(可以是单片机，芯片等)或处理器(processor)执行本申请各个实施例所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。
本领域的普通技术人员可以理解，上述各实施例是实现本申请的具体 实施例，而在实际应用中，可以在形式上和细节上对其作各种改变，而不偏离本申请的精神和范围。

Claims (17)

  1. 一种主动降噪方法，其特征在于，包括：
    根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数；其中，所述误差信号为所述环境噪声信号分别通过主通道和次级通道后在目标降噪点叠加后的信号；
    将所述主通道对应的自适应滤波器的频域系数转换为无限脉冲响应IIR滤波器的时域系数；其中，所述IIR滤波器的阶数小于所述自适应滤波器的阶数；
    将所述IIR滤波器的时域系数更新到所述IIR滤波器中；其中，所述IIR滤波器位于所述次级通道上，用于对通过所述次级通道的环境噪声信号进行滤波输出降噪信号。
  2. 如权利要求1所述的主动降噪方法，其特征在于，所述根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数，包括：
    将所述环境噪声信号转换为环境噪声频域信号，并将所述误差信号转换为误差频域信号；
    根据所述环境噪声频域信号对所述误差频域信号进行误差幅度控制；
    根据误差幅度控制后的误差频域信号和所述环境噪声频域信号，获取主通道对应的自适应滤波器的频域系数。
  3. 如权利要求2所述的主动降噪方法，其特征在于，所述根据所述环境噪声频域信号对所述误差频域信号进行误差幅度控制，包括：
    根据所述环境噪声频域信号对所述误差频域信号进行归一化处理，获取归一化处理后的误差频域信号；
    根据所述归一化处理后的误差频域信号的幅值与预设阈值的比较结果，获取误差限幅系数；
    根据所述误差限幅系数获取误差幅度控制后的误差频域信号。
  4. 如权利要求3所述的主动降噪方法，其特征在于，所述根据所述环境噪声频域信号对所述误差频域信号进行归一化处理，获取归一化处理后的误差频域信号，包括：
    分别获取所述环境噪声频域信号的功率谱和所述误差频域信号的幅度谱；
    将所述幅度谱和所述功率谱的比值作为所述归一化处理后的误差频域信号的幅度谱。
  5. 如权利要求3或4所述的主动降噪方法，其特征在于，所述根据所述归一化处理后的误差频域信号的幅值与预设阈值的比较结果，获取误差限幅系数，包括：
    通过以下公式计算所述误差限幅系数：
    Figure PCTCN2019090912-appb-100001
    其中，所述Ef(k)为所述误差限幅系数，所述|E1(k)|为所述归一化处理后的误差频域信号的幅值，所述τ(k)为所述预设阈值，所述k为所述误差频域信号的频点。
  6. 如权利要求5所述的主动降噪方法，其特征在于，所述预设阈值设置有多个值，不同频率值范围对应不同的预设阈值；其中，所述不同频率值范围为所述误差频域信号的频率值范围。
  7. 如权利要求6所述的主动降噪方法，其特征在于，所述预设阈值的取值方式为：
    Figure PCTCN2019090912-appb-100002
    其中，所述f为所述误差频域信号的频率值。
  8. 如权利要求1所述的主动降噪方法，其特征在于，所述将所述主通道对应的自适应滤波器的频域系数转换为无限脉冲响应IIR滤波器的时域系数，包括：
    将所述主通道对应的自适应滤波器的频域系数转换为所述主通道对应的自适应滤波器的时域系数；
    根据LMS算法将所述主通道对应的自适应滤波器的时域系数转换为所述IIR滤波器的时域系数。
  9. 如权利要求8所述的主动降噪方法，其特征在于，所述根据LMS算法将所述主通道对应的自适应滤波器的时域系数转换为所述IIR滤波器的时域系数，包括：
    将所述主通道对应的自适应滤波器的时域系数进行翻转；
    生成随机数，并根据翻转后的所述时域系数和所述随机数获取期望信号；
    根据LMS算法和所述期望信号得到所述IIR滤波器的时域系数。
  10. 如权利要求8所述的主动降噪方法，其特征在于，所述根据LMS算法将所述主通道对应的自适应滤波器的时域系数转换为所述IIR滤波器的时域系数，包括：
    根据LMS算法将所述主通道对应的自适应滤波器的时域系数转换为m阶IIR滤波器的时域系数；
    将所述m阶IIR滤波器的时域系数转换为多个n阶IIR滤波器的时域系数；其中，所述m大于所述n，所述多个n阶IIR滤波器均对应有各自的时域系数；
    所述将所述IIR滤波器的时域系数更新到所述IIR滤波器中，具体为：
    将所述多个n阶IIR滤波器的时域系数更新到所述多个n阶IIR滤波器中。
  11. 如权利要求1所述的主动降噪方法，其特征在于，在所述根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数之前，还包括：
    自适应更新所述次级通道对应的自适应滤波器的系数；
    将所述次级通道对应的自适应滤波器的系数更新到所述次级通道对应的自适应滤波器中；
    所述根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波中的环境噪声信号为通过所述次级通道对应的自适应滤波器的环境噪声信号。
  12. 如权利要求11所述的主动降噪方法，其特征在于，所述自适应更新所述次级通道对应的自适应滤波器的系数，具体为：
    自适应更新所述次级通道对应的自适应滤波器的频域系数。
  13. 如权利要求11所述的主动降噪方法，其特征在于，所述自适应更新所述次级通道对应的自适应滤波器的系数，包括：
    在所述次级通道上引入预设的白噪声信号，并根据所述白噪声信号和调整后的所述误差信号进行频域自适应滤波，获取所述次级通道对应的自适应滤波器的系数；其中，所述调整后的所述误差信号为所述白噪声信号经预设滤波器处理后的信号与所述误差信号叠加后的信号；所述预设滤波器的系数与所述次级通道对应的自适应滤波器的系数相同；
    所述根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波中的误差信号为所述调整后的所述误差信号。
  14. 一种主动降噪装置，其特征在于，包括：
    频域自适应滤波模块，用于根据环境噪声信号和在目标降噪点采集到的误差信号进行频域自适应滤波，获取主通道对应的自适应滤波器的频域系数；其中，所述误差信号为所述环境噪声信号分别通过主通道和次级通道后在目标降噪点叠加后的信号；
    IIR滤波器设计模块与所述频域自适应滤波模块连接，用于将所述主通道对应的自适应滤波器的频域系数转换为无限脉冲响应IIR滤波器的时域系数，并将所述IIR滤波器的时域系数更新到所述IIR滤波器中；其中，所述IIR滤波器的阶数小于所述自适应滤波器的阶数；
    所述IIR滤波器与所述IIR滤波器设计模块连接且位于所述次级通道上，用于对通过所述次级通道的环境噪声信号进行滤波输出降噪信号。
  15. 一种主动降噪芯片，其特征在于，包括：
    至少一个处理器；以及，
    与所述至少一个处理器通信连接的存储器；其中，
    所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行如权利要求1至13中任一所述的主动降噪方法。
  16. 一种主动控制系统，其特征在于，包括如权利要求15所述的主动降噪芯片和无限脉冲响应IIR滤波器；
    所述主动降噪芯片用于获取所述IIR滤波器的时域系数，并将所述IIR滤波器的时域系数更新到所述IIR滤波器中；
    所述IIR滤波器位于次级通道上，用于对通过所述次级通道的环境噪声信号进行滤波输出降噪信号。
  17. 一种计算机可读存储介质，存储有计算机程序，其特征在于，所述计算机程序被处理器执行时实现权利要求1至13中任一项所述的主动降噪方法。
PCT/CN2019/090912 2019-06-12 2019-06-12 主动降噪方法、装置、芯片、主动控制系统和存储介质 WO2020248164A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN201980001078.6A CN110402540B (zh) 2019-06-12 2019-06-12 主动降噪方法、装置、芯片、主动控制系统和存储介质
PCT/CN2019/090912 WO2020248164A1 (zh) 2019-06-12 2019-06-12 主动降噪方法、装置、芯片、主动控制系统和存储介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2019/090912 WO2020248164A1 (zh) 2019-06-12 2019-06-12 主动降噪方法、装置、芯片、主动控制系统和存储介质

Publications (1)

Publication Number Publication Date
WO2020248164A1 true WO2020248164A1 (zh) 2020-12-17

Family

ID=68328406

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/090912 WO2020248164A1 (zh) 2019-06-12 2019-06-12 主动降噪方法、装置、芯片、主动控制系统和存储介质

Country Status (2)

Country Link
CN (1) CN110402540B (zh)
WO (1) WO2020248164A1 (zh)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111866666B (zh) * 2020-07-28 2022-07-08 西安讯飞超脑信息科技有限公司 数字降噪滤波器生成方法、相关设备及可读存储介质
CN112037752A (zh) * 2020-09-08 2020-12-04 珠海格力电器股份有限公司 家电设备降噪方法、装置、计算机设备和存储介质
CN112397079A (zh) * 2020-10-20 2021-02-23 浙江大华技术股份有限公司 滤波器及其自适应滤波方法、计算机可读存储介质
CN112735370B (zh) * 2020-12-29 2022-11-01 紫光展锐(重庆)科技有限公司 一种语音信号处理方法、装置、电子设备和存储介质
CN113506556B (zh) * 2021-06-07 2023-08-08 哈尔滨工业大学（深圳） 主动噪声控制方法、装置、存储介质和计算机设备
CN115620738B (zh) * 2022-12-14 2023-08-29 小米汽车科技有限公司 降噪方法、装置、电子设备及介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2487125A (en) * 2011-01-05 2012-07-11 Cambridge Silicon Radio Ltd Active noise cancellation controller with fixed hybrid filters and a third adaptive filter
CN105356861A (zh) * 2015-09-28 2016-02-24 歌尔声学股份有限公司 一种有源降噪的方法和系统
CN107210718A (zh) * 2014-11-20 2017-09-26 迪芬尼香港有限公司 使用多速率fir和全通iir滤波器均衡扬声器系统的声学响应的方法和装置
CN107731217A (zh) * 2017-10-18 2018-02-23 恒玄科技（上海）有限公司 一种实现不同频率响应拟合的主动降噪系统及方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2487125A (en) * 2011-01-05 2012-07-11 Cambridge Silicon Radio Ltd Active noise cancellation controller with fixed hybrid filters and a third adaptive filter
CN107210718A (zh) * 2014-11-20 2017-09-26 迪芬尼香港有限公司 使用多速率fir和全通iir滤波器均衡扬声器系统的声学响应的方法和装置
CN105356861A (zh) * 2015-09-28 2016-02-24 歌尔声学股份有限公司 一种有源降噪的方法和系统
CN107731217A (zh) * 2017-10-18 2018-02-23 恒玄科技（上海）有限公司 一种实现不同频率响应拟合的主动降噪系统及方法

Also Published As

Publication number Publication date
CN110402540B (zh) 2021-03-16
CN110402540A (zh) 2019-11-01

Similar Documents

Publication Publication Date Title
WO2020248164A1 (zh) 主动降噪方法、装置、芯片、主动控制系统和存储介质
US9754605B1 (en) Step-size control for multi-channel acoustic echo canceller
US10056065B2 (en) Adaptive modeling of secondary path in an active noise control system
JP5241921B2 (ja) 電子音響チャンネルの適応制御とイコライゼーションの方法
US8718291B2 (en) ANC for BT headphones
US10657981B1 (en) Acoustic echo cancellation with loudspeaker canceling beamformer
JP2020510240A (ja) リアルタイム音響プロセッサ
CN109727604A (zh) 用于语音识别前端的频域回声消除方法及计算机储存介质
CN108600894A (zh) 一种耳机自适应有源噪声控制系统及方法
EP2519943A2 (en) Adaptive digital noise canceller
CN109379652B (zh) 一种耳机有源噪声控制的次级通道离线辨识方法
Zhou et al. A frequency band constrained filtered–x least mean square algorithm for feedback active control systems
Hasegawa et al. Headrest application of multi-channel feedback active noise control with virtual sensing technique
EP3759708A1 (en) Feedforward active noise control
JPH11168792A (ja) 音場制御装置
WO2021016000A2 (en) Frequency domain adaptation with dynamic step size adjustment based on analysis of statistic of adaptive filter coefficient movement
WO2021016001A1 (en) Input signal-based frequency domain adaptive filter stability control
Xie et al. A feedforward and feedback composite active noise reduction headset based on inverse filter frequency equalization and its DSP system implementation
Liu et al. Active control for vehicle interior noise using the improved iterative variable step-size and variable tap-length LMS algorithms
CN111785243B (zh) 一种耳机有源降噪方法及系统
CN113132848A (zh) 滤波器设计方法、装置以及入耳式主动降噪耳机
TWI777729B (zh) 適應性主動雜訊消除裝置以及使用其之聲音播放系統
WO2023077252A1 (zh) 一种基于FxLMS结构的主动降噪系统、方法及设备
Yin et al. Selective fixed-filter active noise control based on frequency response matching in headphones
US20230131827A1 (en) Method for generating active noise reduction filter, storage medium and earphone

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19932844

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19932844

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 19932844

Country of ref document: EP

Kind code of ref document: A1