CN111193977B

CN111193977B - Noise reduction method of earphone, self-adaptive FIR filter, noise removal filter bank and earphone

Info

Publication number: CN111193977B
Application number: CN201911292251.0A
Authority: CN
Inventors: 李倩
Original assignee: Bestechnic Shanghai Co Ltd
Current assignee: Bestechnic Shanghai Co Ltd
Priority date: 2019-12-13
Filing date: 2019-12-13
Publication date: 2021-10-01
Anticipated expiration: 2039-12-13
Also published as: CN111193977A

Abstract

The disclosure relates to a noise reduction method of an earphone, an adaptive FIR filter, a denoising filter bank and the earphone. The noise reduction method comprises the following steps: collecting an external environment noise signal; filtering with a first Infinite Impulse Response (IIR) filter based on the external ambient noise signal to obtain a first noise reference signal to counteract effects of the out-of-ear microphone and the speaker; acquiring a residual noise signal and adjusting coefficients of a first Finite Impulse Response (FIR) filter based on the residual noise signal and a first noise reference signal; the first noise reference signal is filtered with a first FIR filter and the in-ear inverse noise is output via the speaker to cancel in-air with the noise entering the ear. The method can adaptively adjust the filter coefficient of the FIR filter according to any one of the wearing posture of the earphone, the ear structure and the noise scene change, and achieves good suppression effect on the noise under different conditions under the conditions of adaptive rapid convergence and low order of the FIR filter.

Description

Noise reduction method of earphone, self-adaptive FIR filter, noise removal filter bank and earphone

Technical Field

The present disclosure relates to a headphone and a noise reduction method of the headphone, and more particularly, to a noise reduction method of a headphone, an adaptive FIR filter, a noise removal filter bank, and a headphone.

Background

With the social progress and the improvement of the living standard of people, the earphone becomes an indispensable living article for people. Increasingly, earphones with active noise suppression functions are widely accepted by markets and customers, and users can enjoy comfortable noise reduction experience in various noisy environments such as airports, subways, airplanes and restaurants.

However, different noise conditions, different headphone structures can affect the active noise suppression function of existing headphones. First, most of the active noise reduction schemes provided by the current headphones are that a user selects a filter coefficient according to a noise scene, for example, the noise scene may include: airplanes, restaurants, subways, streets, and the like. The user sets different noise reduction coefficients of the headset by selecting different noise scenes. When a user switches among a plurality of scenes, the scenes need to be selected for multiple times to adjust the noise reduction coefficient, and the user experience is greatly influenced by the method. Even in the same scene, the noise conditions are not consistent, for example, subways in rush hours and late-night subways on duty have completely different noise intensities, and it is obviously not appropriate to use the same noise reduction coefficient for subway scenes in different periods. Secondly, earphones on the market have multiple structures, for example, in-ear earphones, half-in-ear earphones and the like, different earphones have different wearing modes, different users also have different ear structures, and the mode that the user wears the earphones has certain influence on noise reduction.

Therefore, earphones with adaptive active noise reduction function have been produced. The current adaptive active noise reduction earphone mostly depends on a Finite Impulse Response (FIR) filter to realize the adaptive noise reduction function. However, the order of the adaptive active noise suppression system implemented by the FIR filter is high, and the computational complexity and power consumption are also high.

Disclosure of Invention

The present disclosure is provided to solve the above-mentioned problems occurring in the prior art.

The present disclosure needs an adaptive active noise reduction method for an earphone, an adaptive FIR filter, a noise removal filter group, and an earphone, which can adaptively adjust a filter coefficient of the FIR filter according to any one of an earphone wearing posture, a human ear structure, and a noise scene change, so that a good suppression effect can be achieved for noise under different conditions under the conditions of fast adaptive convergence and a low order of the required FIR filter.

According to a first aspect of the present disclosure, there is provided a noise reduction method of a headphone including an out-of-ear microphone, a filter bank, a speaker, and an in-ear microphone. The noise reduction method comprises the following steps: collecting an external ambient noise signal via the out-of-ear microphone; filtering with a first Infinite Impulse Response (IIR) filter based on the acquired external environment noise signal to obtain a first noise reference signal, thereby cancelling the effects of the out-of-ear microphone and the speaker; acquiring a residual noise signal via the in-ear microphone and adjusting coefficients of a first Finite Impulse Response (FIR) filter based on the residual noise signal and the first noise reference signal; and filtering the first noise reference signal by using the first FIR filter, and outputting in-ear reverse noise through the loudspeaker to perform air cancellation with noise of the external environment noise entering into the ear.

According to a second aspect of the present disclosure, an adaptive FIR filter is provided. The adaptive FIR filter includes: a first FIR filtering module configured to FIR filter based on the length and the filter coefficient; an acquisition module configured to acquire a residual noise signal and acquire a noise reference signal after external environment noise is cancelled by the action of the out-of-ear microphone and the loudspeaker; an adjustment module configured to: adjusting filter coefficients of the first FIR filtering block based on a noise reference signal and a residual noise signal.

According to a third aspect of the present disclosure, a denoising filter bank is provided. The denoising filter bank includes: an adaptive FIR filter according to various embodiments of the present disclosure; a first IIR filter whose coefficients are derived in advance by measurement to cancel the effects of the out-of-ear microphone and the loudspeaker. The first IIR filter is configured to: and generating the noise reference signal based on the external environment noise signal collected by the microphone outside the ear, and feeding the noise reference signal to an acquisition module in the self-adaptive FIR filter.

According to a fourth aspect of the present disclosure, there is provided a headset, which is an in-ear headset or a semi-in-ear headset. The earphone includes: an ear microphone configured to acquire an external ambient noise signal; an in-ear microphone configured to acquire a residual noise signal; the denoising filter bank according to various embodiments of the present disclosure is configured to: generating an in-ear inverse noise signal based on the acquired external environment noise signal; and a speaker configured to output the in-ear inverse noise based on the in-ear inverse noise signal for air cancellation with noise of the external environment noise entering into the ear.

By utilizing the noise reduction method, the self-adaptive FIR filter, the noise removal filter group and the earphone, the filter coefficient of the FIR filter can be self-adaptively adjusted according to any one of the wearing posture of the earphone, the structure of the human ear and the noise scene change, so that the noise under different conditions can be well inhibited under the conditions of realizing self-adaptive rapid convergence and lower order of the needed FIR filter.

Drawings

In the drawings, which are not necessarily drawn to scale, like reference numerals may describe similar components in different views. Like reference numerals having letter suffixes or different letter suffixes may represent different instances of similar components. The drawings illustrate various embodiments generally by way of example and not by way of limitation, and together with the description and claims serve to explain the disclosed embodiments. Such embodiments are illustrative, and are not intended to be exhaustive or exclusive embodiments of the present apparatus or method.

Fig. 1 shows a schematic diagram of a noise reduction method of a headphone according to an embodiment of the present disclosure;

fig. 2 shows a flow chart of a noise reduction method of the headset shown in fig. 1;

fig. 3 shows a schematic diagram of a noise reduction method of a headphone according to an embodiment of the present disclosure;

fig. 4 shows a flow chart of a sub-flow for adjusting and deriving coefficients of a first FIR filter in the noise reduction method of the headphone shown in fig. 3;

FIG. 5 shows a block diagram of an adaptive FIR filter according to an embodiment of the present disclosure;

FIG. 6 shows a block diagram of a denoising filter bank according to an embodiment of the disclosure; and

fig. 7 shows a schematic structural diagram of a headset according to an embodiment of the present disclosure.

Detailed Description

For a better understanding of the technical aspects of the present disclosure, reference is made to the following detailed description taken in conjunction with the accompanying drawings. Embodiments of the present disclosure are described in further detail below with reference to the figures and the detailed description, but the present disclosure is not limited thereto. The order in which the various steps described herein are described as examples should not be construed as a limitation if there is no requirement for a context relationship between each other, and one skilled in the art would know that sequential adjustments may be made without destroying the logical relationship between each other, rendering the overall process impractical.

Fig. 1 shows a schematic diagram of a noise reduction method 100 of a headphone according to an embodiment of the present disclosure. As shown in fig. 1, there is an environmental noise 101a in the environment where the user is located, which is noise generated in the environment where the user is located. When the user wears the headphone, the ambient noise 101a passes through the headphone and enters behind the human ear to form in-ear noise 101b, and the in-ear noise 101b has a relatively low noise intensity with respect to the ambient noise 101 a.

An earphone according to an embodiment of the present disclosure may include an out-of-ear microphone 102,

filter banks

107a and 107b, a speaker 108, and an in-ear microphone 103. The ear microphone 102 and the ear microphone 103 may be digital microphones or analog microphones, and the analog microphones are exemplified as analog microphones that are usually used together with an analog-to-digital converter. The headset may further comprise a first analog-to-digital converter 104, a second analog-to-digital converter 105 and a digital-to-analog converter 106. First, an external environment noise signal 101a is collected via the ear microphone 102 and input to the first analog-to-digital converter 104, which performs an analog-to-digital conversion process on the collected environment noise 101a to obtain a digital signal x1(n) of the external environment noise (where n represents the current sampling time). Based on the digital signal of the external ambient noise, filtering is performed using a first Infinite Impulse Response (IIR) filter 107a to obtain a first noise reference signal u1(n), thereby canceling the effects of the ear microphone 102 and the speaker 108. The first noise reference signal u1(n) is input to the first FIR filter 107b to be filtered, and the in-ear inverse noise (i.e., synthesized in-ear inverse noise) 101c is output via the digital-to-analog converter 106 and the speaker 108. The ambient noise signal 101a is fit-synthesized into an inverse noise signal having an approximate signal intensity and a direction opposite to that of the in-ear noise 101b after passing through the filtering action of the first IIR filter 107a and the first FIR filter 107 b. The inverse noise signal is subjected to digital-to-analog conversion by the digital-to-analog converter 106 and is then played by the speaker 108, forming a synthesized inverse noise 101c in the ear. The inverse noise 101c is similar to the in-ear noise 101b in signal strength and opposite in direction to the in-ear noise 101b, thereby generating an air cancellation effect in the user's ear to achieve an effect of reducing the in-ear noise 101 b.

The noise transmission path from the external environment noise 101a to the in-ear microphone 103 may be divided into a first path on which the signal is applied by the out-of-ear microphone 102 and the earphone, respectively, and a second path on which the signal is applied by the speaker 108 and the ear canal from the speaker 108 to the in-ear microphone 103, respectively. The noise reduction method according to the present disclosure divides the filter bank into a first IIR filter 107a, whose transfer function is substantially invariant, and an adaptive first FIR filter 107 b. The effect of the off-ear microphone 102 and the loudspeaker 108, in which the transfer functions of the first path and the second path remain relatively stable, is simulated by using the first IIR filter 107a, for example, so that the transfer function of the first IIR filter 107a is an inverse function of the product of the transfer functions of the off-ear microphone 102 and the loudspeaker 108, thereby eliminating adaptive calculation of at least part of the filter bank (the first IIR filter 107a), reducing the order of the adaptive filter (e.g., the first FIR filter 107b), reducing the computational complexity and power consumption, and also increasing the speed of adaptive convergence. In some embodiments, the coefficients of the first IIR filter 107a may be obtained in advance by measurement. The transfer functions of the ear microphone 102 and the loudspeaker 108 are usually relatively stable, and the coefficients of the first IIR filter 107a can be fixed after being measured and set in advance, so that the effects of the ear microphone 102 and the loudspeaker 108 can still be removed relatively accurately. The coefficients of the first IIR filter 107a in the filter bank may be fixed, while only the coefficients of the first FIR filter 107b need to be adaptively adjusted. The residual noise signal may be acquired by the in-ear microphone 103 and converted into a digital signal e1(n) of the residual noise by the second analog-to-digital converter 105, and the coefficient of the first FIR filter 107b may be adjusted based on the digital signal e1(n) of the residual noise and the first noise reference signal u1(n), so that the filter coefficient of the first FIR filter 107b may be adaptively adjusted for a change of a transfer function on the first path and the second path due to any one of a wearing manner, an individual human ear structure, and a noise scene change, and a good noise reduction effect may be ensured in each case. According to the combined design of the first IIR filter 107a and the first self-adaptive FIR filter 107b with fixed coefficients, the self-adaptive calculation of the IIR filter which can cause instability is avoided, the order and the calculation complexity of the FIR filter are reduced, the self-adaptive convergence speed is higher, the response speed of noise reduction processing is higher, a feedback loop is more stable, and meanwhile, the noise under different conditions can be well suppressed.

The coefficients of the first FIR filter 107b may be adjusted based on the residual noise signal e1(n) and the first noise reference signal x1(n) in various ways. The method for adjusting the coefficients of the second FIR filter 107d will be described in detail with reference to fig. 3, and the method for adjusting the coefficients of the first FIR filter 107b can also be used for adjusting the coefficients of the first FIR filter 107b, which is not described herein again. Furthermore, the coefficients of the first FIR filter 107b may also be adjusted based on the residual noise signal e1(n) and the first noise reference signal x1(n) using a subband adaptive filtering algorithm and a frequency-domain adaptive filtering algorithm.

According to the embodiment of the present disclosure, since the synthesized in-ear inverse noise 101c is similar to the in-ear noise 101b in signal strength, there is a small part of residual noise after cancellation effect, that is, the residual noise 101d exists in the ear, and the in-ear microphone 103 acquires the residual noise 101d and inputs it to the second analog-to-digital converter 105 to obtain a digital signal of the residual noise. To this end, the headphone noise reduction process 100 acquires the ambient noise 101a outside the ear, fits the first IIR filter 107a and the first FIR filter 107b to obtain the inverse noise 101c with the intensity close to and the opposite direction to the in-ear noise 101b, thereby generating air noise cancellation to achieve the noise reduction effect, and the cancelled residual noise 101d is transmitted to the adaptive filter, i.e., the first FIR filter 107b, through the feedback loop. The first FIR filter 107b can adjust the noise reduction coefficient of the filter according to the change of the environmental noise in different scenes, and improve the noise reduction effect of the earphone through a stable feedback loop.

Fig. 2 shows a flow chart of a noise reduction method of the headset shown in fig. 1. The noise reduction method 200 starts in step 201: an external ambient noise signal is collected via an out-of-ear microphone. In step 202, filtering is performed by using a first IIR filter based on the collected external environment noise signal to obtain a first noise reference signal, so as to cancel the effects of the out-of-ear microphone and the loudspeaker. In some embodiments, the coefficients of the first IIR filter may be obtained by pre-measuring the out-of-ear microphone and the speaker, respectively. For example, the voice acquisition circuit may be constructed by an ear microphone and an analog-to-digital converter, which perform fourier transform on a voice digital signal output from the analog-to-digital converter and an input voice signal, respectively, and calculate a transfer function of the ear microphone based thereon. It is also possible to calculate the transfer function of the loudspeaker in a similar manner, multiply the transfer function of the extra-aural microphone with the transfer function of the loudspeaker and take the inverse of the multiplication, and design and construct the first IIR filter using the obtained inverse as the transfer function. In step 203, a residual noise signal may be acquired via the in-ear microphone and coefficients of a first FIR filter may be adjusted based on the residual noise signal and the first noise reference signal. In this way, the coefficients of the first FIR filter can be adaptively adjusted for a specific situation, so as to minimize or control the residual noise signal in the current situation below an acceptable threshold. By adjusting the coefficients of the first FIR filter based on the residual noise signal and the first noise reference signal that cancels the effects of the extra-aural microphone and the loudspeaker, adaptive adjustment of the portion that remains substantially stable with respect to the transfer function on the transmission path can be avoided, thereby reducing the order of the FIR filter, reducing the computational complexity and power consumption, and accelerating convergence thereof.

In step 204, the first noise reference signal is filtered by the first FIR filter, and the in-ear inverse noise is output via the speaker to perform air cancellation with the noise of the external environment noise entering into the ear. Note that the first FIR filter that performs filtering may employ filter coefficients after adaptive adjustment. Note that the arrows between the steps in fig. 2 do not define the order of precedence between the steps, but rather the steps may be implemented in a different order without destroying the logical relationship of the overall process. For example, steps 201 and 202 may be performed in parallel with step 203 or before step 203, as long as the input signals required for adaptive adjustment, i.e. the residual noise signal and the first noise reference signal, may be provided for step 203. Step 204 may be performed with the first FIR filter having updated filter coefficients after step 203 is completed, i.e. after the adaptive updating of the coefficients of the first FIR filter is completed. Step 203 is executed to obtain the updated filter coefficients after adaptive convergence, and the first FIR filter that switches the updated filter coefficients is executed to step 204, so that seamless docking of the denoising process can be maintained and interruption of the denoising process can be avoided.

Fig. 3 shows a schematic diagram of a noise reduction method 300 of a headphone according to an embodiment of the present disclosure, wherein the same components as those in fig. 1 are respectively denoted by the same reference numerals, and a repeated description of the same components is omitted here. The difference from the noise reduction method 100 shown in fig. 1 is that the adaptive adjustment of the filter coefficients of the second FIR filter 107d is performed at a low sampling rate, thereby further reducing the amount of operation of the adaptive adjustment of the filter coefficients and ensuring the real-time performance of the operation; and converts the resulting second FIR filter 107d at a low sample rate to a high sample rate to obtain the first FIR filter 107 b. Therefore, the self-adaptive process under the low sampling rate is easier to converge to the optimal solution, and the self-adaptive process is better in the aspects of device maintenance and noise reduction effect; the first IIR filter 107a and the first FIR filter 107b at a high sampling rate are used for active noise reduction, the system delay is low, and the synthesized reverse noise and the in-ear noise can obtain a better cancellation effect. Further, the faster and easier convergence process of the filter coefficients at low sampling rates, combined with the lower latency and faster response speed of the filter at high sampling rates, further increases the speed of the overall noise reduction process.

The method 300 is described in detail below with reference to fig. 3.

The acquired external ambient noise signal x1(n) may be downsampled by the first downsampling module 112 to obtain a low sample rate external ambient noise signal x (n) and fed to the second IIR filter 107 c. The first IIR filter 107a at high sampling rates may determine the filter coefficients in a variety of ways, such as by measurement in accordance with various embodiments of the present disclosure. The first IIR filter 107a is converted to a low sample rate by a second conversion module 111 to obtain said second IIR filter 107 c. The downsampled external ambient noise signal x (n) is filtered by a second IIR filter 107c at a low sampling rate to obtain a second noise reference signal u (n). This second noise reference signal u (n) is in fact down-sampled with respect to the first noise reference signal u1(n) and fed to the second FIR filter 107 d. The residual noise signal e1(n) is downsampled by a second downsampling block 109 to obtain a downsampled residual noise signal e (n) and fed to a second FIR filter 107 d. The coefficients of the second FIR filter 107d may be adjusted based on said second noise reference signal u (n) at low sampling rate and the residual noise signal e (n) at low sampling rate.

Specifically, the filter coefficients of the second FIR filter 107d may be adjusted according to the following formula (1):

where w is the filter coefficient of the second FIR filter 107d and μ is the iteration step coefficient. In some embodiments, μmay range between 0.1 and 0.5 as desired.

w can be expressed as formula (2):

w(n)＝[w₀(n),w₁(n),w₂(n),...,w_M-1(n)]^Tequation (2)

u (n) can be expressed as formula (3):

u(n)＝[u(n),u(n-1),...,u(n-M+1)]^Tequation (3)

Where M is the length (number of taps) of the second FIR filter 107d, and n represents the current sampling time of the second FIR filter 107 d.

e (n) can be expressed as formula (4):

e(n)＝d(n)-w^T(n) u (n), formula (4)

Where d (n) represents the noise entering the ear through the headphone at time n. Equation (4) only shows the relationship between the residual noise signal e (n) and other signals, and during the adaptive coefficient adjustment, the residual noise signal is collected by the in-ear microphone 103, analog-to-digital converted by the second analog-to-digital converter 105, and then down-sampled by the second down-sampling module 109, without using equation (4) for calculation.

The filter coefficients w of the second FIR filter 107d may be iteratively calculated according to equation (1) until the residual noise signal e (n) at the low sampling rate reaches a minimum or is below a threshold value.

The second FIR filter 107d with adaptively converged coefficients may be converted to a high sampling rate by the first conversion module 110 to obtain the first FIR filter 107b for active noise reduction. Algorithms for the FIR filter to convert between different sampling rates are frequency sampling, resampling, etc.

In some embodiments, the sampling rate of the signal before downsampling is the same as the high sampling rate and greater than or equal to 192KHz, and the sampling rate of the signal after downsampling is the same as the low sampling rate and less than or equal to 48 KHz. With equation (1), adaptive system identification can be achieved at a sampling rate of 48kHz, resulting in filter coefficients with a time resolution of 20 μ s.

Fig. 4 shows a flow chart of a sub-flow 400 for adjusting and deriving coefficients of a first FIR filter in the noise reduction method of the headphone shown in fig. 3. As shown in FIG. 4, the sub-process 400 begins at step 401: the acquired external ambient noise signal is down sampled. At step 402, the first IIR filter is converted to a low sample rate to obtain a second IIR filter. Here, the filter coefficient of the first IIR filter at a high sampling rate may be obtained according to the pre-measurement means of the various embodiments of the present disclosure. In step 403, the downsampled external environment noise is filtered by using a second IIR filter to obtain a second noise reference signal. At step 404, the residual noise signal is down sampled. Subsequently, the coefficients of the second FIR filter may be adjusted based on the second noise reference signal and the downsampled residual noise signal (step 405). In step 406, the second FIR filter with the adjusted coefficients is converted to a high sampling rate to obtain the first FIR filter at the high sampling rate. The first FIR filter at high sampling rate may be connected in series with the first IIR filter at high sampling rate for noise reduction. Note that the arrows in fig. 4 do not limit the order relationship between these steps, and steps 401 and 406 may be performed in an order different from that shown in fig. 4 as long as the implementation of each step in the sub-flow 400 is not hindered. In particular, the steps 401 and 403 for obtaining the second noise reference signal and the step 404 for obtaining the down-sampled residual noise signal may be performed in any order with respect to each other as long as it is ensured that the step 405 is provided with the second noise reference signal on which the adjustment is based and the down-sampled residual noise signal. In some embodiments, the coefficients of the second FIR filter may be iteratively adjusted in step 405 until convergence. Accordingly, in step 406, the second FIR filter with the adaptive convergence coefficients is converted to a high sampling rate to obtain the first FIR filter that can be used for actual noise reduction. In this way, the conversion step 406 can be ensured to work only for the second FIR filter with a low sampling rate of the adaptive convergence coefficient, and the first FIR filter with a high sampling rate converted can also generate the in-ear inverse noise with a good cancellation effect, in view of the fact that the adaptive convergence coefficient is obtained in a case where the generated in-ear inverse noise and the residual noise signal after the in-ear noise cancellation are stabilized below the threshold value.

As shown in fig. 4, all the adaptive calculation processes of the filter coefficients are implemented at a low sampling rate, and by performing FIR filter calculation and conversion at the low sampling rate, the order and the calculation complexity of the noise suppression system of the FIR filter can be reduced, and simultaneously, the power consumption caused by calculation can be reduced.

Fig. 5 shows a block diagram of an adaptive FIR filter 500 according to an embodiment of the disclosure. The adaptive FIR filter 500 may include: a first FIR filtering module 503 configured to perform FIR filtering based on the length and the filter coefficient; an obtaining module 501 configured to obtain a residual noise signal and obtain a noise reference signal after external environment noise is cancelled by the external microphone and the speaker; an adjustment module 502 configured to: adjusting filter coefficients of the first FIR filtering block based on a noise reference signal and a residual noise signal.

In some embodiments, the first FIR filtering module 503 may be implemented as programmable (e.g., at least filter coefficient writable) hardware, such as any one of an FPGA (field programmable gate array), an ASIC (application specific integrated circuit), an SOC (system on a chip), a DSP (digital signal processor) chip. In some embodiments, the first FIR filtering module 503 may be implemented based on logic circuits and a memory table, and the filter coefficients may be stored in corresponding memory locations in the memory table; by writing the filter coefficients adaptively adjusted in the respective storage units, the adaptive FIR filter can be realized. Further, the obtaining module 501 and the adjusting module 502 may be implemented as executable computer instructions stored on a memory and executable by a micro-processing unit, such as but not limited to a DSP, a single chip, a SOC, an ARM (advanced reduced instruction set computer) processor, an Intel processor, a microprocessor without internal interlocking pipeline stages (MIPS processor), and the like. In some embodiments, first FIR filtering module 503 may also be implemented as executable computer instructions stored on a memory and executable by a micro-processing unit.

In some embodiments, the adjustment module 502 is further configured to: the filter coefficients are adjusted based on the downsampled noise reference signal and the downsampled residual noise signal to obtain the coefficients of the FIR filter at the low sampling rate. Also, the adaptive FIR filter 500 may further include a first conversion module 504 configured to: the coefficients of the FIR filter of low sample rate are converted into the coefficients of the FIR filter of high sample rate and fed to the first FIR filtering block. In some embodiments, the obtaining module 501, the adjusting module 502 and the first converting module 504 may be implemented as logic circuits, that is, implemented in a hardware form; however, in view of the reduction of the order of the adaptive FIR filter (fixing the coefficient of the IIR filter) and the low sampling rate of the calculation of the adaptive filter coefficient, the implementation via computer executable instructions (i.e. software implementation) can still ensure a fast convergence speed and a fast noise reduction processing speed, and the cost and the manufacturing difficulty can be significantly reduced by the software implementation.

In some embodiments, the adjustment module 502 may adjust the filter coefficients based on the downsampled noise reference signal and the downsampled residual noise signal according to equation (1) described above.

Fig. 6 shows a block diagram of a denoising filter bank 600 according to an embodiment of the present disclosure. The noise canceling filter bank 600 may include an adaptive FIR filter 500 (e.g., as shown in fig. 5) according to various embodiments of the present disclosure, a first IIR filter 601 whose coefficients are derived in advance by measurement to cancel the effects of the out-of-ear microphone and the speaker. Wherein the first IIR filter 601 may be configured to: the noise reference signal is generated based on the external ambient noise signal acquired by the out-of-ear microphone and fed to the acquisition module 502, so that the adjustment module 502 acquires the noise reference signal and the residual noise signal via the acquisition module 502 to adjust the coefficients of the first FIR filtering module 503.

In some embodiments, the adjustment module 502 may be further configured to: the filter coefficients are adjusted based on the downsampled noise reference signal and the downsampled residual noise signal to obtain the coefficients of the FIR filter at the low sampling rate. Also, the adaptive FIR filter 500 may further include a first conversion module 504 configured to: the coefficients of the FIR filter of low sample rate are converted to the coefficients of the FIR filter of high sample rate and fed to said first FIR filtering block 503. The denoising filter bank 600 further includes a second conversion module 602 configured to: the first IIR filter 601 is converted to a low sample rate, wherein the low sample rate noise reference signal is generated with the IIR filter at the low sample rate based on the external ambient noise signal and fed to the adjustment module 502.

In some embodiments, the first IIR filter 601 may be implemented as an FPGA or an ASIC. In view of the fact that the filter coefficient is relatively fixed, and the IIR filter has a common logic circuit design of an FPGA or an ASIC, the filter processing speed can be faster when the IIR filter is implemented as the FPGA or the ASIC, the design difficulty is low, and the cost is also low. The second conversion module 602, like the obtaining module 501, the adjusting module 502 and the first conversion module 504, can be implemented as a software module, thereby significantly reducing the cost and the manufacturing difficulty. In some embodiments, the adjustment module 502 may adjust the filter coefficients based on the downsampled noise reference signal and the downsampled residual noise signal according to equation (1) described above.

Fig. 7 shows a schematic structural diagram of an earphone 700 according to an embodiment of the present disclosure. The headset 700 may be an in-ear headset or a semi-in-ear headset. As shown in fig. 7, the headset 700 may include: an out-of-ear microphone 701 configured to collect an external ambient noise signal; an in-ear microphone 702 configured to acquire a residual noise signal; the denoising filter bank 703 according to various embodiments of the present disclosure is configured to: generating an in-ear inverse noise signal based on the acquired external environment noise signal; and a speaker 704 configured to output the in-ear inverse noise based on the in-ear inverse noise signal for air cancellation with noise of the external environment noise entering into the ear.

In some embodiments, the denoising filter bank 703 includes an IIR filter whose filter coefficients are relatively fixed and an FIR filter whose filter coefficients can be adaptively adjusted for each case. For an in-ear headphone, noise in different scenes and different individual in-ear structures affect noise reduction, and for a half-in-ear headphone, the headphone wearing posture also has a large effect on noise reduction. By introducing the FIR filter of which the filter coefficient can be adjusted adaptively according to various situations, good noise reduction effect can be ensured in various situations. And the IIR filter with relatively fixed filter coefficient is used for fitting the part with relatively unchanged transfer function on the transmission path, so that the order of the self-adaptive FIR filter is reduced, the calculation complexity and the power consumption are reduced, and the self-adaptive convergence speed is accelerated. The combined design of the IIR filter with fixed coefficient and the self-adaptive FIR filter avoids the self-adaptive calculation of the IIR filter which can cause instability, reduces the order and the calculation complexity of the FIR filter, has faster self-adaptive convergence speed, leads the response speed of noise reduction processing to be faster, leads a feedback loop to be more stable, and can achieve good suppression effect on noise of an in-ear earphone and a half-in-ear earphone under different conditions.

In some embodiments, the out-of-ear microphone 701 and the in-ear microphone 702 are each digital microphones or have an analog-to-digital converter electrically connected downstream and a digital-to-analog converter connected upstream of the speaker 704.

Moreover, although exemplary embodiments have been described herein, the scope thereof includes any and all embodiments based on the disclosure with equivalent elements, modifications, omissions, combinations (e.g., of various embodiments across), adaptations or alterations. The elements of the claims are to be interpreted broadly based on the language employed in the claims and not limited to examples described in the present specification or during the prosecution of the application, which examples are to be construed as non-exclusive. It is intended, therefore, that the specification and examples be considered as exemplary only, with a true scope and spirit being indicated by the following claims and their full scope of equivalents.

The above description is intended to be illustrative and not restrictive. For example, the above-described examples (or one or more versions thereof) may be used in combination with each other. For example, other embodiments may be used by those of ordinary skill in the art upon reading the above description. In addition, in the foregoing detailed description, various features may be grouped together to streamline the disclosure. This should not be interpreted as an intention that a disclosed feature not claimed is essential to any claim. Rather, inventive subject matter may lie in less than all features of a particular disclosed embodiment. Thus, the following claims are hereby incorporated into the detailed description as examples or embodiments, with each claim standing on its own as a separate embodiment, and it is contemplated that these embodiments may be combined with each other in various combinations or permutations. The scope of the invention should be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.

Claims

1. A noise reduction method for a headset comprising an out-of-ear microphone, a filter bank, a speaker, and an in-ear microphone, the noise reduction method comprising:

collecting an external ambient noise signal via the out-of-ear microphone;

filtering with a first Infinite Impulse Response (IIR) filter based on the acquired external environment noise signal to obtain a first noise reference signal, thereby cancelling the effects of the out-of-ear microphone and the speaker;

acquiring a residual noise signal via the in-ear microphone;

the external environment noise signal collected by the microphone outside the ear is downsampled, the first IIR filter is converted into a low sampling rate to obtain a second IIR filter, and the downsampled external environment noise signal is filtered by the second IIR filter to obtain a second noise reference signal;

down-sampling the residual noise signal;

adjusting the coefficient of a second FIR filter based on the second noise reference signal and the downsampled residual noise signal, and converting the coefficient-adjusted second FIR filter to a high sampling rate to obtain a first FIR filter;

and filtering the first noise reference signal by using the first FIR filter, and outputting in-ear reverse noise through the loudspeaker to perform air cancellation with noise of the external environment noise entering into the ear.

2. The noise reduction method according to claim 1, wherein a sampling rate of the signal before the down-sampling is the same as the high sampling rate and is greater than or equal to 192KHZ, and a sampling rate of the signal after the down-sampling is the same as the low sampling rate and is less than or equal to 48 KHZ.

3. The noise reduction method according to claim 1, wherein adjusting the coefficients of the second FIR filter based on the second noise reference signal and the downsampled residual noise signal is according to the following equation (1):

wherein w (n) ═ w₀(n)，w₁(n)，w₂(n)，...，w_M-1(n)]^TM is the length of the second FIR filter and n is the secondCurrent sampling instant of the second FIR filter, w (n) being the filter coefficients of said second FIR filter at instant n, which are comprised from w₀(n) to w_M-1(n) a total of M components, w (n +1) being the filter coefficients of said second FIR filter at time instant n +1, including from w₀(n +1) to w_M-1(n +1) M components, u (n) being the second noise reference signal at time n, e (n) being the residual noise signal after down-sampling at time n, μ being an iteration step size coefficient.

4. The noise reduction method according to claim 1, wherein coefficients of the first IIR filter are obtained in advance by measurement.

5. An adaptive FIR filter, characterized in that the adaptive FIR filter comprises:

a first FIR filtering module configured to FIR filter based on the length and the filter coefficient;

an acquisition module configured to acquire a residual noise signal and acquire a noise reference signal after external environment noise is cancelled by the action of the out-of-ear microphone and the loudspeaker;

an adjustment module configured to: adjusting filter coefficients of the first FIR filtering module based on a noise reference signal and a residual noise signal; adjusting filter coefficients based on the downsampled noise reference signal and the downsampled residual noise signal to obtain coefficients of the FIR filter with a low sampling rate; and

a conversion module configured to: the coefficients of the low sample rate FIR filter are converted to the coefficients of the high sample rate FIR filter and fed to the first FIR filtering block.

6. The adaptive FIR filter according to claim 5, characterized in that the adjustment of the filter coefficients based on the downsampled noise reference signal and the downsampled residual noise signal is according to the following formula (1):

wherein w (n) ═ w₀(n)，w₁(n)，w₂(n)，...，w_M-1(n)]^TM is the length of the low-sampling-rate FIR filter, n is the current sampling instant of the low-sampling-rate FIR filter, w (n) is the coefficient of the low-sampling-rate FIR filter at instant n, which includes from w₀(n) to w_M-1(n) a total of M components, w (n +1) being the filter coefficients of said low sample rate FIR filter at time instant n +1, including from w₀(n +1) to w_M-1(n +1) M components in total, u (n) being the downsampled noise reference signal at time n, e (n) being the downsampled residual noise signal at time n, μ being an iteration step-size coefficient.

7. A denoising filter bank, comprising:

the adaptive FIR filter according to claim 5;

a first IIR filter whose coefficients are derived in advance by measurement to cancel the effects of an out-of-ear microphone and the speaker, wherein the first IIR filter is configured to: and generating the noise reference signal based on the external environment noise signal collected by the out-of-ear microphone, and feeding the noise reference signal to the acquisition module.

8. The denoising filter bank of claim 7, wherein adjusting filter coefficients based on the downsampled noise reference signal and the downsampled residual noise signal is according to the following equation (1):

wherein w (n) ═ w₀(n)，w₁(n)，w₂(n)，...，w_M-1(n)]^TM is the length of the low sampling rate FIR filter and n is the low sampling rate FIR filterThe current sampling instant of the FIR filter of rate, w (n) are the coefficients of the FIR filter of low sampling rate, which are comprised from w₀(n) to w_M-1(n) a total of M components, w (n +1) being the filter coefficients of said low sample rate FIR filter at time instant n +1, including from w₀(n +1) to w_M-1(n +1) M components in total, u (n) being the downsampled noise reference signal at time n, e (n) being the downsampled residual noise signal at time n, μ being an iteration step-size coefficient.

9. A headset that is an in-ear headset or a semi-in-ear headset, the headset comprising:

an ear microphone configured to acquire an external ambient noise signal;

an in-ear microphone configured to acquire a residual noise signal;

the denoising filter bank of claim 7 or 8, configured to: generating an in-ear inverse noise signal based on the acquired external environment noise signal; and

a speaker configured to output in-ear inverse noise based on the in-ear inverse noise signal for air cancellation with noise of the external ambient noise entering an ear.

10. The headset of claim 9, wherein the out-of-ear microphone and the in-ear microphone are each a digital microphone or an analog microphone electrically connected downstream to an analog-to-digital converter, and wherein the speaker has a digital-to-analog converter connected upstream.