Using voice accelerometer signals to reduce noise in headsets

Abstract

The present invention relates to the field of speech and audio signal processing, namely noise reduction in headsets. In particular, the present invention proposes a device and a method for reducing noise in headsets. The device is used to: obtain a microphone signal; obtain a speech accelerometer signal; adjust the speech accelerometer signal according to the microphone signal; in a first operating mode, the device is also used to: modify the microphone signal by attenuating low frequencies; merge the modified microphone signal with the adjusted speech accelerometer signal; and output the merged signal.

Description

Technical Field

[0001] The present invention relates to the field of speech and audio signal processing, and in particular to noise reduction in a headset. The present invention provides a device and a method for reducing noise in a headset, respectively. The device and the method solve the following problem, that is, when using a headset, for example, when making a phone call in a noisy environment, the speech obtained by the microphone is mixed with noise. Background Art

[0002] The traditional approach to solving the above noise problem is to use the same noise reduction methods used in the telephone service. For example, spectral subtraction (which can be traced back to the late 1970s, see, for example, "Suppression of acoustic noise in speech using spectral subtraction by SF Boll (IEEE Trans. Acoust. Signal Proc., Vol. ASSP-27, April 1979")). In addition, voice activity detection (VAD) and beamformers have also been studied for many years.

[0003] A newer approach is to use a voice accelerometer to generate a voice accelerometer signal. The advantage of this voice accelerometer signal is that its signal-to-noise ratio (SNR) is inherently better suited to the user's voice, but only within a narrow frequency range (usually below 1kHz).

[0004] US 20140093093 A1 (relating to Apple Inc.'s "AirPods") describes that VAD is calculated from a speech accelerometer and used to steer a beamformer and update noise estimates in noise suppression. Figure 8 compares the spectrum of the output signal of the "AirPods" (a) with the spectrum of the output signal produced by a device based on the above conventional approach (b).

[0005] US20180367882 A1 (see Fig. 9 ) describes the use of speech accelerometer signals for estimating speech in order to achieve more accurate speech estimation in noise suppression (and other speech enhancement tasks).

[0006] US20170337933 A1 (see Fig.10) describes that noise suppression is performed in two stages: first, in a "first equalizer", and secondly after a filter bank (FB) using a gain. The gain is adjusted based on the voice signature (VS), which is derived from the voice accelerometer signal. The filter bank is used to provide frequency division.

[0007] With the above method, if the SNR of the microphone signal is too low, that is, in the presence of noise, the user voice in the microphone signal cannot be restored. Summary of the invention

[0008] In view of the above problems and shortcomings, embodiments of the present invention are intended to improve conventional methods. The purpose of the present invention is to provide a device and a method for reducing noise in a headset, respectively, which can effectively reduce the noise in a microphone signal without losing information about the speech component in the microphone signal. Specifically, in the presence of noise (i.e., even if the SNR of the microphone signal is low), it should be possible to recover the user's voice (i.e., the sound / speech picked up by the microphone) in the microphone signal.

[0009] This object is achieved by the embodiments of the invention described in the attached independent claim. Advantageous implementations of the embodiments of the invention are further defined in the dependent claims.

[0010] Specifically, embodiments of the present invention are based on the recognition that if the SNR of the microphone signal is too low, it is not possible to recover the user voice component in the microphone signal using only the microphone signal itself. Accordingly, embodiments of the present invention propose to additionally use a voice accelerometer signal to improve the SNR in the microphone signal. Noise reduction can then be continued using conventional methods that provide adequate noise reduction at moderate noise levels. Specifically, by utilizing the voice accelerometer signal, the SNR of the microphone signal can be improved in low frequencies.

[0011] A first aspect of the present invention provides a device for reducing noise in a headset, the device being used to: obtain a microphone signal; obtain a speech accelerometer signal; adjust the speech accelerometer signal according to the microphone signal; in a first operating mode, the device is also used to: modify the microphone signal by attenuating low frequencies; merge the modified microphone signal with the adjusted speech accelerometer signal; and output the merged signal.

[0012] Specifically, adjusting the voice accelerometer signal according to the microphone signal may mean modifying the voice accelerometer signal so that its amplitude and phase are the same or at least similar to the amplitude and phase of the microphone signal. This may be achieved by adaptive filtering using the microphone signal and the voice accelerometer signal as input.

[0013] The "low frequencies" of the microphone signal attenuated in the first mode may include frequencies below 1 kHz, may include frequencies below 2 kHz, and may include frequencies below 3 kHz. The attenuated frequency range may be selected based on the noise level in the environment of the device.

[0014] In the first mode, the device according to the first aspect is capable of using the speech accelerometer signal to improve the SNR of the microphone signal, especially at the low frequencies. Thus, the noise in the microphone signal can be reduced without losing any information, especially information about the speech component in the microphone signal. Conventional noise reduction methods can then be applied to the combined signal by the first device or another device, etc.

[0015] In an implementation of the first aspect, the device comprises: a filter bank comprising a plurality of filters; the filter bank is configured to combine the modified microphone signal with the adjusted speech accelerometer signal in the first mode.

[0016] For example, depending on the SNR of the microphone signal, different filters in a filter bank or different filter banks may be used, especially at the low frequencies. These can be easily implemented, making the device according to the first aspect very flexible to different environmental (noise) conditions.

[0017] In an implementation of the first aspect, the filter group includes a first filter and a second filter, the first filter is a high-pass filter, and the second filter is a low-pass filter; in the first mode, the device is also used to: use the first filter to filter the microphone signal to obtain the modified microphone signal; and use the second filter to filter the adjusted speech accelerometer signal before merging the adjusted speech accelerometer signal with the modified microphone signal.

[0018] In other words, the high pass (first) filter is used to attenuate the low frequencies of the microphone signal. The low pass (second) filter is used to attenuate the high frequencies of the voice accelerometer signal. A typical voice accelerometer signal has content only in lower frequencies, for example below 3 kHz, or even just below 1 kHz. Accordingly, the low pass filter may be used to pass frequencies below 3 kHz, or even just below 1 kHz, of the voice accelerometer signal.

[0019] In an implementation of the first aspect, the speech accelerometer signal is adjusted according to the microphone signal at a sampling rate lower than the sampling rate of the microphone signal; and the second filter is used to interpolate the adjusted speech accelerometer signal to the sampling rate of the microphone signal.

[0020] In an implementation of the first aspect, in the second operation mode, the device is further configured to: delay the microphone signal; and output the delayed microphone signal.

[0021] Accordingly, the device is used to generate a delay of the microphone signal. The delay may be caused by filtering the microphone signal. Specifically, when the delay is generated, the microphone signal does not change, that is, the content of the microphone signal remains unchanged, but is shifted in time. Delaying the microphone signal enables smooth switching between the first operating mode and the second operating mode, and vice versa.

[0022] If noise reduction is not required, the second operation mode is preferably used. In this case, the voice accelerometer signal may not be merged with the microphone signal and the low frequencies of the microphone signal may not be attenuated. The different operation modes make the device very flexible and efficient in different noise conditions.

[0023] In an implementation of the first aspect, the device is further configured to: apply noise suppression to the combined signal in the first mode and / or apply noise suppression to the delayed microphone signal in the second mode.

[0024] That is, the signal output by the device may be further subjected to noise suppression or reduction, in particular, based on conventional methods.

[0025] In an implementation of the first aspect, the filter bank is configured to delay the microphone signal in the second mode.

[0026] In an implementation manner of the first aspect, in the second mode, the device is further configured to: filter the microphone signal using the first filter and the second filter to obtain the delayed microphone signal.

[0027] The first filter and the second filter may be configured such that when the microphone signal is filtered using both filters, the content of the microphone signal is not changed. The microphone signal is simply delayed by the filtering process.

[0028] In an implementation of the first aspect, the device is configured to adjust the filters in the filter bank when changing between the first mode and the second mode.

[0029] That is, in the first mode and the second mode, the filter bank can use different filters respectively. For example, in the first mode, a high-pass filter is needed to suppress the low frequencies, and the suppression depends on the noise level. In the second mode, a filter is needed that delays but does not change the microphone signal. The filters in the filter bank can also be adjusted for different noise conditions in the first mode, that is, different filters are used. For example, under normal noise conditions, a moderate attenuation of the low frequencies of the microphone signal may be sufficient. Under strong / wind noise conditions, a stronger attenuation of the low frequencies may be required. Under strong wind noise conditions, the higher frequencies of the microphone signal (for example, above 1kHz, or even above 3kHz) may also be additionally attenuated.

[0030] Therefore, different first filters can be used according to the noise.

[0031] In an implementation of the first aspect, each of the filters in the filter bank includes less than 10 taps, specifically each includes 7 taps.

[0032] Accordingly, the filters in the filter bank are relatively short and can therefore be realized particularly efficiently.

[0033] In an implementation manner of the first aspect, the device is used to: determine whether an environment is quiet or noisy; if the environment is noisy, select the first mode; if the environment is quiet, select the second mode.

[0034] If the environment is quiet, the device may also select a third mode in which the microphone signal is output without any delay.

[0035] A "quiet" or "noisy" environment (the environment of the device, i.e., the surroundings of the device) can be distinguished based on the noise level or some more complex method. That is, if the noise level is below a threshold, the environment can be considered "quiet", and if the noise level is above the threshold, the environment can be considered "noisy". Of course, different thresholds can be used to more accurately distinguish "noisy", "very noisy", etc., or to distinguish "ordinary noise" (e.g., cafeteria, car, street, station, campus, etc.) from "wind noise". The noise level can be considered (only) within a certain frequency range. Exemplary thresholds for distinguishing between "quiet" and "noisy" environments can be noise sound pressure levels of 30dB, 50dB, or 60dB.

[0036] In an implementation of the first aspect, the device is further configured to, when determining that the environment is noisy: determine whether the environment is windy; and if the environment is windy, enhance the attenuation of the low frequency of the microphone signal in the first mode.

[0037] A "windy" environment may be determined based on the presence of wind noise in the microphone signal. Wind noise may be detected and, in particular, distinguished from normal noise. For example, wind noise may be highly non-stationary and may vary, for example including gusts of wind. Normal noise may be more stationary and uniform.

[0038] In an implementation of the first aspect, the device is configured to gradually change between the first mode and the second mode.

[0039] This helps to avoid noise pumping effects and other discontinuities. For example, the filters in the filter bank can be adjusted so that they are between the filters used in the first mode and the filters used in the second mode. For example, when gradually changing from the second mode to the first mode, the attenuation of the low frequencies can be gradually increased.

[0040] In an implementation of the first aspect, the device includes an adaptive filter, which is used to adjust the voice accelerometer signal according to the microphone signal, specifically, to adjust the amplitude and phase of the voice accelerometer signal to the amplitude and phase of the microphone signal.

[0041] Specifically, with respect to the speech component, the amplitude and phase are adjusted to be the same or at least similar. That is, the speech component in the adjusted speech accelerometer signal can match the speech component in the microphone signal.

[0042] In an implementation manner of the first aspect, the device (specifically, a headset) includes: a microphone, used to generate the microphone signal; and a voice accelerometer, used to generate the voice accelerometer signal.

[0043] The device may further comprise the adaptive filter and / or the filter bank and / or the processing circuit.

[0044] A second aspect of the present invention provides a method for reducing noise in a headset, the method comprising: obtaining a microphone signal; obtaining a speech accelerometer signal; adjusting the speech accelerometer signal according to the microphone signal; in a first operating mode, the method further comprises: modifying the microphone signal by attenuating low frequencies; merging the modified microphone signal with the adjusted speech accelerometer signal; and outputting the merged signal.

[0045] In an implementation of the second aspect, the method comprises combining the modified microphone signal with the adjusted speech accelerometer signal in the first mode using a filter bank, wherein the filter bank comprises a plurality of filters.

[0046] In an implementation of the second aspect, the filter group includes a first filter and a second filter, the first filter is a high-pass filter, and the second filter is a low-pass filter; in the first mode, the method also includes: filtering the microphone signal using the first filter to obtain the modified microphone signal; and filtering the adjusted speech accelerometer signal using the second filter before merging the adjusted speech accelerometer signal with the modified microphone signal.

[0047] In an implementation of the second aspect, the speech accelerometer signal is adjusted according to the microphone signal at a sampling rate lower than a sampling rate of the microphone signal; and the second filter interpolates the adjusted speech accelerometer signal to the sampling rate of the microphone signal.

[0048] In an implementation manner of the second aspect, in a second operation mode, the method further includes: delaying the microphone signal; and outputting the delayed microphone signal.

[0049] In an implementation of the second aspect, the method further includes: applying noise suppression to the combined signal in the first mode and / or applying noise suppression to the delayed microphone signal in the second mode.

[0050] In an implementation of the second aspect, the filter bank delays the microphone signal in the second mode.

[0051] In an implementation manner of the second aspect, in the second mode, the method further includes: filtering the microphone signal using the first filter and the second filter to obtain the delayed microphone signal.

[0052] In an implementation of the second aspect, the method includes adjusting the filters in the filter bank when changing between the first mode and the second mode.

[0053] In an implementation of the second aspect, each of the filters in the filter bank includes less than 10 taps, specifically each includes 7 taps.

[0054] In an implementation manner of the second aspect, the method includes: determining whether an environment is quiet or noisy; if the environment is noisy, selecting the first mode, and if the environment is quiet, selecting the second mode.

[0055] In an implementation manner of the second aspect, the method further includes, when determining that the environment is noisy: determining whether the environment is windy; and if the environment is windy, enhancing the attenuation of the low frequency of the microphone signal in the first mode.

[0056] In an implementation of the second aspect, the method includes: gradually changing between the first mode and the second mode.

[0057] In an implementation of the second aspect, the method includes: using an adaptive filter to adjust the voice accelerometer signal according to the microphone signal, specifically, adjusting the amplitude and phase of the voice accelerometer signal to the amplitude and phase of the microphone signal.

[0058] In an implementation of the second aspect, the method (specifically performed in a headset) includes: generating the microphone signal using a microphone; and generating the voice accelerometer signal using a voice accelerometer.

[0059] The method according to the second aspect and its implementations achieves all the above advantages and effects of the device according to the first aspect and its implementations. The above definitions and explanations of the device according to the first aspect also apply to the method according to the second aspect.

[0060] It should be noted that all devices, elements, units and modules described in this application can be implemented in software or hardware elements or any type of combination thereof. All steps performed by various entities described in this application and the functions described to be performed by various entities are intended to indicate that the corresponding entities are suitable for or used to perform the corresponding steps and functions.

[0061] Although in the description of the specific embodiments below, specific functions or steps performed by an external entity are not reflected in the description of the specific detailed elements of the entity that performs the specific steps or functions, the technician should be aware that these methods and functions can be implemented in corresponding hardware or software elements or any combination thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In conjunction with the accompanying drawings, the following description of specific embodiments illustrates various aspects and implementations of the present invention, wherein:

[0063] Figure 1 The device provided by the embodiment of the present invention is shown;

[0064] Figure 2 The device provided by the embodiment of the present invention is shown, especially in the first operation mode;

[0065] Figure 3 The device provided by the embodiment of the present invention is shown, especially in the second operation mode;

[0066] Figure 4(a) to Figure 4(c) An example of a filter bank of a device provided by an embodiment of the present invention is shown;

[0067] Figure 5 The mode selection process performed by the device provided in the embodiment of the present invention is schematically shown;

[0068] FIG6(a) and FIG6(b) compare the frequency spectrum of the output signal of the device provided by the embodiment of the present invention with the frequency spectrum of the output signal obtained by a traditional method;

[0069] Figure 7 The method provided by the embodiment of the present invention is shown;

[0070] FIG8( a ) and FIG8( b ) compare the spectrum of the output signal of Apple's “AirPods” with the spectrum of the output signal obtained by another conventional method;

[0071] Fig. 9 Another conventional noise reduction method is schematically shown;

[0072] Fig.10 Another conventional noise reduction method is schematically shown. DETAILED DESCRIPTION

[0073] Figure 1 The device 100 provided by an embodiment of the present invention is shown. The device 100 is suitable for reducing noise in a headset, or at least suitable for supporting the reduction of noise in a headset. The device 100 may be included in the headset. The device 100 may also be the headset, for example, it may be a headset suitable for a mobile phone or other mobile device. However, although the device 100 is suitable for a headset, it may also be suitable for any other microphone-based device or application.

[0074] The device 100 may include a processing circuit (not shown) for performing, carrying out or starting various operations of the device 100 described herein. The processing circuit may include hardware and software. The hardware may include analog circuits or digital circuits, or both analog circuits and digital circuits. The digital circuit may include components such as an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a digital signal processor (DSP) or a multi-purpose processor. In one embodiment, the processing circuit includes one or more processors and a non-transient memory connected to the one or more processors. The non-transient memory may carry an executable program code, which, when executed by one or more processors, causes the device 100 to perform, carry out or start the operations or methods described herein.

[0075] The device 100 is used to obtain a microphone signal 101, which is specifically provided by one or more microphones or a microphone array. That is, the microphone signal 101 can be provided by one microphone, or it can be a signal generated by combining multiple microphones. The device 100 may include the one or more microphones to generate the microphone signal 101 itself. However, the device 100 may also receive the microphone signal 101 from the one or more (external) microphones or from another device.

[0076] The device 100 is also used to obtain a voice accelerometer signal 102, which is specifically provided by one or more voice accelerometers or a voice accelerometer array. That is, the voice accelerometer signal 102 can be provided by a voice accelerometer, or it can be a signal generated by combining multiple voice accelerometers. The device 100 can include the one or more voice accelerometers to generate the voice accelerometer signal 102 itself. However, the device 100 can also receive the voice accelerometer signal 102 from the one or more voice accelerometers or from another device. It is worth noting that the voice accelerometer can be used to measure vibrations (e.g., in the x, y and z directions), such as vibrations caused at / in the user's head. Specifically, the vibration can be caused by the user's voice, that is, when the user speaks to the one or more microphones. Therefore, the voice accelerometer signal 102 can reflect the user's voice, especially at low frequencies. The voice accelerometer can also be implemented by a microphone, that is, the voice accelerometer signal 102 can be a microphone signal, wherein the microphone can be placed in the user's ear.

[0077] The device 100 is further configured to adjust the voice accelerometer signal 102 according to the microphone signal 101 to obtain an adjusted voice accelerometer signal 103, for example by taking the microphone signal 101 as an input to the adjustment process. For example, the adjustment may be performed by adaptive filtering of the voice accelerometer signal 102 and the microphone signal 101. Specifically, the adjustment may align the amplitude and / or phase of the microphone signal 101 with the amplitude and / or phase of the voice accelerometer signal 102 with respect to the voice signal component (but not the noise component caused by the noise leaking into the accelerometer signal 102).

[0078] The device 100 can operate in at least a first operating mode (eg, Figure 1 ), specifically, it can also be operated in the second operating mode ( Figure 1 Accordingly, the device 100 may be configured to gradually change between the first mode and the second mode.

[0079] In the first operation mode, the device 100 is further configured to modify the microphone signal 101 by attenuating low frequencies (of the microphone signal 101) to obtain a modified microphone signal 104. Therefore, noise in the microphone signal 101, which is mainly reflected at low frequencies, can be reduced. In addition, the device 100 is further configured to combine the modified microphone signal 104 with the adjusted voice accelerometer signal 103 and then output a combined signal 105. Therefore, the (adjusted) voice accelerometer signal 102 can be used to improve the SNR of the (modified) microphone signal 101. The first operation mode is most suitable for noisy environments.

[0080] In the second operation mode, the device 100 is further configured to delay the microphone signal 101 to obtain a delayed microphone signal 204, and then output the delayed microphone signal 204 ( Figure 1 Not shown, but see e.g. Figure 3 ). In particular, apart from the delay, the microphone signal 101 may not be modified in the second mode. For example, the microphone signal 101 is not merged with the (adjusted) voice accelerometer signal 102. The second operation mode is best suited for quiet environments.

[0081] In the first mode, the device 100 generates a combined signal 105, and in the second mode, the device 100 generates a delayed microphone signal 204. In the intermediate mode, the generated signal may be defined as:

[0082] Resulting signal = f*combined signal + (1-f)*delayed microphone signal

[0083] If the first mode and the second mode are gradually changed, f gradually changes from zero (0) to one (1), and vice versa.

[0084] Figure 2 A device 100 provided by an embodiment of the present invention is shown. Figure 2 The embodiment of the device 100 shown is based on Figure 1 The embodiment of the device 100 shown in FIG. Figure 2 is shown as having further optional features. Accordingly, Figure 2 The device 100 includes Figure 1 Therefore, like features are labeled with like reference numerals and have like functions.

[0085] Figure 2 In particular, a first operating mode of the device 100 is shown. Figure 3 The embodiment of the present invention provides Figure 2 The same device 100 is shown, but specifically showing the second mode of operation of the device 100.

[0086] like Figure 2 and Figure 3As shown respectively, the device 100 may include an adaptive filter 203, which may be used to adjust the voice accelerometer signal 102 according to the microphone signal 101. Specifically, the adaptive filter 203 may be used to adjust the amplitude and / or phase of the voice accelerometer signal 102 to the amplitude and / or phase of the microphone signal 101, especially with respect to the sound / voice component in the signal. To this end, the adaptive filter 203 may receive the voice accelerometer signal 102 as an input, and may also receive the adjusted voice accelerometer signal 103 and the microphone signal 101 as another input (specifically, the input may be the sum of the signal 103 and the signal 101).

[0087] like Figure 2 and Figure 3 As further shown, the device 100 may include a filter bank 200, the filter bank 200 including a plurality of filters (201, 202), specifically including a first filter 201 and a second filter 202. In the first mode ( Figure 2 ), the filter bank 200 may be used to perform said merging of the modified microphone signal 104 with the adjusted speech accelerometer signal 103. In said second mode ( Figure 3 ), the filter bank 200 can be used to cause the delay of the microphone signal 101. The filters (201, 202) in the filter bank 200 can be adjustable, i.e. can be changed. For example, their filtering behavior with respect to the passing amplitude and / or frequency can be changed.

[0088] The first filter 201 may be a high-pass filter, and the second filter 202 may be a low-pass filter. Figure 2 ), the microphone signal 101 may be filtered using the first filter 201 to obtain the modified microphone signal 104, and the adjusted speech accelerometer signal 103 may be filtered using the second filter 202 before being combined with the modified microphone signal 104. In the second mode ( Figure 3 ), the microphone signal 101 can be filtered using the first filter 201 and the second filter 202 simultaneously to obtain a delayed microphone signal 204. It is worth noting that passing any type of signal through the (analysis-synthesis) filter bank will produce a delayed version of the signal.

[0089] exist Figure 2In the device 100 shown, i.e. in the first operating mode of the device 100, the SNR of the microphone signal 101 can be improved, especially in the low frequency band, by first modifying the voice accelerometer signal 102 to have an amplitude and / or phase similar to the microphone signal 101 of the user's voice. This can be achieved by an adaptive filter 203 (wherein the adaptation can only be performed when the user speaks, i.e. when the pitch is detected). In addition, by secondly using the filter bank 200 to merge the signal 101 and the signal 102, the noise reduction is then continued as in the conventional method (based on the merged signal 105).

[0090] The filter bank 200 used in the first mode or in the second mode may depend on the SNR of the microphone signal 101, especially at low frequencies. In a quiet environment, the SNR of the microphone signal 101 is usually better than the SNR of the speech accelerometer signal 102, because movement will induce low-level noise in the speech accelerometer signal 102. Then, preferably, the speech accelerometer signal 102 is not used, but the microphone signal 101 itself and the filter bank 200 are (only) used to generate the delay (i.e. to generate the delayed microphone signal 204). This is as Figure 3 The second mode is shown, and is therefore best suited for quiet environments. In noisy environments (e.g., cafeterias, cars, streets, stations, campuses, etc.) or in breezy environments, it may be sufficient to reduce low-frequency noise by 10-20 dB, etc. (by attenuating the microphone signal 101 at low frequencies). However, wind noise may also be very strong at low frequencies, in which case additional attenuation may be required. This is as Figure 2 The first mode is shown, and is therefore best suited for noisy environments, even windy environments. The low frequency attenuation of the microphone signal 101 can be achieved by a high pass filter as the first filter 201 .

[0091] Typically, the speech accelerometer signal 102 has content only in its lower frequencies. Therefore, the adaptive filtering using the adaptive filter 203 can be performed at a lower sampling rate, and the filter bank 200 can include interpolating the adaptively filtered speech accelerometer signal 103 to 16kHz or the like (i.e., after inserting enough zeros between samples, the attenuation of the second filter 202 in the filter bank 200 must be deeper). Therefore, the speech accelerometer signal 102 can be adjusted according to the microphone signal 101 at a sampling rate lower than the sampling rate of the microphone signal 101. Then, the second filter 202 can be used to interpolate the adjusted speech accelerometer signal 103 to the sampling rate of the microphone signal 101.

[0092] For the filters (201, 202) in the filter bank 200, very short filters may be selected, in which case they can be implemented particularly efficiently. For example, in the exemplary filter bank 200, each of the filters (201, 202) may have less than 10 taps, for example each may have only 7 taps. Due to the interpolation of the adjusted speech accelerometer signal 104, the length of the combined second filter 202 may be longer, for example 32.

[0093] Figure 4(a) to Figure 4(c) The device 100 (specifically, Figure 2 or Figure 3 201 , 202 in a filter bank 200 of the device 100 shown.

[0094] In FIG4( a ), the filters ( 201 , 202 ) are best suited for normal noise (not wind noise). The first filter 201 is a high pass filter for filtering the microphone signal 101 ; the second filter 202 is a low pass filter for filtering the voice accelerometer signal 102 . The first filter 101 attenuates the microphone signal 101 below 3 kHz by up to 20 dB at very low frequencies close to 0 kHz. The second filter 202 passes the voice accelerometer signal up to approximately 3-4 kHz.

[0095] In FIG. 4( b ), the filters ( 201 , 202 ) are best suited for wind noise. The first filter 201 is a high-pass filter for filtering the microphone signal 101 ; the second filter 202 is a low-pass filter for filtering the voice accelerometer signal 102 . At very low frequencies close to 0 kHz, the first filter 101 attenuates the microphone signal 101 below 3 kHz by up to 80 dB (even infinitely attenuated at about 0 kHz). That is, compared to FIG. 4( a ), the low-frequency attenuation of the microphone signal 101 is significantly enhanced. The second filter 202 passes the voice accelerometer signal up to about 3-4 kHz.

[0096] In FIG4( c ), the filters ( 201 , 202 ) are best suited for normal noise, as in FIG4( a ), but without any interpolation. The first filter 201 is a high pass filter for filtering the microphone signal 101 ; the second filter 202 is a low pass filter for filtering the voice accelerometer signal 102 . As in FIG4( a ), the first filter 101 attenuates the microphone signal 101 below 3 kHz by up to 20 dB at very low frequencies close to 0 kHz. The second filter 202 may be shorter, for example, may have 7 taps.

[0097] Figure 5The device 100 (specifically, Figure 2 or Figure 3 1. The mode selection process performed by the device 100 shown in FIG. 1. In general, the device 100 can determine whether the environment (its surroundings) is quiet or noisy, and can select the first mode when the environment is noisy, and select the second mode when the environment is quiet. In addition, when the device 100 determines that the environment is noisy, the device 100 can also determine whether the environment is windy. If the environment is windy, the low-frequency attenuation of the microphone signal 101 in the first mode can be enhanced. The mode can be changed gradually to avoid noise pumping effects and other discontinuities. In a windy environment, the higher frequencies (upper frequency band) of the microphone signal 101 can also be additionally attenuated, for example, by 18 dB.

[0098] Specifically, Figure 5 As shown, the device 100 may further include an analysis filter bank 500, a high pass filter 501, a first mixer 503, and a second mixer 502. The first mixer 503 may receive the speech accelerometer signal 101 and the microphone signal 101 as the output of the analysis filter bank 500, and may determine whether the environment is quiet. The second mixer 502 may receive the microphone signal 101 as the output of the analysis filter bank 500 and the same microphone signal 101 after passing through the high pass filter 501, and may determine whether the environment is windy. The filter bank 200 may be selected separately, i.e., its filters (201, 202) may be selected according to the determined noise conditions, and the microphone signal 101 or the microphone signal 101 and the speech accelerometer signal 102 may be operated separately.

[0099] Figures 6(a) and 6(b) compare the spectrum of the output signal of the device 100 provided by an embodiment of the present invention with the spectrum of the output signal of a device that performs a traditional method (e.g., the spectral subtraction described in the background section). Specifically, spectrogram 6(a) is the result of using the device 100 in a first operating mode (i.e., generating a combined signal 105) and then applying a traditional noise suppression method to the combined signal 105. Specifically, spectrogram 6(b) is the result of applying the same traditional noise suppression method directly to the microphone signal (without using a voice accelerometer signal). Compared with the traditional device, the noise of spectrogram 6(a) of the output signal of the device 100 is significantly less. It is worth noting that in strong wind conditions, the device 100 can use the traditional method.

[0100] Figure 7 1 shows a method 700 provided by an embodiment of the present invention. The method 700 is applicable to reducing noise in a headset. The method 700 may be executed by the device 100, specifically by Figure 1 , Figure 2 or Figure 3 The device 100 shown executes the method 700. The details of the method 700 may be implemented as described above with respect to the device 100.

[0101] The method 700 comprises: step 701 , obtaining 701 a microphone signal 101 ; step 702 , obtaining a speech accelerometer signal 102 ; and step 703 , adjusting the speech accelerometer signal 102 according to the microphone signal 101 .

[0102] In the first operating mode, method 700 further includes: step 704, modifying the microphone signal by attenuating low frequencies (of microphone signal 101); step 705, combining the modified microphone signal 104 with the adjusted voice accelerometer signal 103; step 706, outputting the combined signal 105.

[0103] In the second operating mode ( Figure 7 (not shown), the method 700 further includes the following steps: delaying the microphone signal 101; and outputting the delayed microphone signal 204.

[0104] The invention has been described in conjunction with different embodiments and implementations as examples. However, from a study of the drawings, the invention and the independent claims, a person skilled in the art will be able to understand and implement other variations when implementing the claimed invention. In the claims and in the specification, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" does not exclude a plurality. A single element or other unit may fulfil the functions of several entities or items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not mean that a combination of these measures cannot be effectively used.

Claims

1. A device (100) for reducing noise in a headset, characterized in that The device (100) is used for: Get microphone signal (101); Obtaining a voice accelerometer signal (102); adjusting the speech accelerometer signal (102) according to the microphone signal (101); In the first operation mode, the device (100) is further configured to: modifying the microphone signal (101) by attenuating low frequencies; combining the modified microphone signal (104) with the adjusted voice accelerometer signal (103); Output the combined signal (105); In the second operation mode, the device (100) is further configured to: Delaying the microphone signal (101); Outputting a delayed microphone signal (204); The device (100) is also used for: Gradually change between the first mode of operation and the second mode of operation.

2. The device (100) according to claim 1, characterized in that include: A filter bank (200) comprising a plurality of filters (201, 202); The filter bank (200) is used to combine the modified microphone signal (104) with the adjusted speech accelerometer signal (103) in the first operating mode.

3. The device (100) according to claim 2, characterized in that: The filter bank (200) comprises a first filter (201) and a second filter (202), wherein the first filter (201) is a high-pass filter and the second filter (202) is a low-pass filter; In the first operation mode, the device (100) is further configured to: filtering the microphone signal (101) using the first filter (201) to obtain the modified microphone signal (104); The adjusted speech accelerometer signal (103) is filtered using the second filter (202) before being combined with the modified microphone signal (104).

4. The device (100) according to claim 3, characterized in that: adjusting the speech accelerometer signal (102) according to the microphone signal (101) at a sampling rate lower than the sampling rate of the microphone signal (101); The second filter (202) is used to interpolate the adjusted speech accelerometer signal (103) to the sampling rate of the microphone signal (101).

5. The device (100) according to claim 1, characterized in that Also used for: Noise suppression is applied to the combined signal (105) in the first operating mode and / or to the delayed microphone signal (204) in the second operating mode.

6. The device (100) according to claim 3 or 4, characterized in that: The filter bank (200) is used to delay the microphone signal (101) in the second operation mode.

7. The device (100) according to claim 6, characterized in that In the second operation mode, the device (100) is further configured to: The microphone signal (101) is filtered using the first filter (201) and the second filter (202) to obtain the delayed microphone signal (204).

8. The device (100) according to claim 7, characterized in that Used for: The filters (201, 202) in the filter bank (200) are adjusted when changing between the first mode of operation and the second mode of operation.

9. The device (100) according to claim 8, characterized in that: Each of the filters (201, 202) in the filter bank (200) comprises less than 10 taps, in particular each comprises 7 taps.

10. The device (100) according to claim 5, characterized in that Used for: Determine whether the environment is quiet or noisy; If the environment is noisy, the first operation mode is selected, and if the environment is quiet, the second operation mode is selected.

11. The device (100) according to claim 10, characterized in that Also used when determining that the environment is noisy: determining whether the environment is windy; The attenuation of the low frequencies of the microphone signal (101) in the first operating mode is enhanced if the environment is windy.

12. The device (100) according to claim 1, characterized in that include: An adaptive filter (203) is used to adjust the voice accelerometer signal (102) according to the microphone signal (101), specifically, to adjust the amplitude and phase of the voice accelerometer signal (102) to the amplitude and phase of the microphone signal (101).

13. The device (100) according to claim 1, characterized in that Specifically, it is a headset, the device (100) include: A microphone, configured to generate the microphone signal (101); A speech accelerometer is used to generate the speech accelerometer signal (102).

14. A method for reducing noise in a headset, characterized in that The method comprises: Get microphone signal (101); Obtaining a voice accelerometer signal (102); adjusting the speech accelerometer signal (102) according to the microphone signal (101); In the first operation mode, the method further comprises: modifying the microphone signal (101) by attenuating low frequencies; combining the modified microphone signal (104) with the adjusted voice accelerometer signal (103); Output the combined signal (105); In the second operation mode, the method further comprises: Delaying the microphone signal (101); Outputting a delayed microphone signal (204); Gradually change between the first mode of operation and the second mode of operation.

Images