CN111970612A - Method and system for eliminating crosstalk of bone conduction earphone - Google Patents

Abstract

The invention relates to a method and system for eliminating crosstalk of bone conduction earphones. The method includes: acquiring a bone conduction transfer matrix when a user wears a bone conduction earphone; acquiring a stopband frequency range of a shape factor; and acquiring an approximate crosstalk elimination matrix after filtering ; Use the approximate crosstalk cancellation matrix to eliminate the crosstalk of the binaural signal input to the bone conduction earphone; obtain the sound signal for eliminating crosstalk; play the sound signal for eliminating crosstalk to obtain the sound signal heard by the user's ears; The conduction transfer matrix, the approximate crosstalk cancellation matrix and the sound signal heard by the user's binaurally determine the final crosstalk cancellation matrix; the crosstalk cancellation is performed on the binaural signals input to the bone conduction earphone according to the final crosstalk cancellation matrix. The invention adjusts and obtains a suitable crosstalk elimination matrix according to the bone conduction transmission matrix, the approximate crosstalk elimination matrix and the audio signals heard by the user's ears, and can obtain the best crosstalk elimination effect.

Description

Crosstalk cancellation method and system for bone conduction earphones

technical field

The present invention relates to the technical field of cross-sound elimination, in particular to a cross-sound elimination method and system for bone conduction earphones.

Background technique

Under normal circumstances, sound waves are introduced into the inner ear through air conduction and bone conduction, and then vibrate by the inner and outer lymph fluids of the inner ear. After comprehensive analysis of the cortex, the sound is finally "heared".

According to the different transmission paths of sound waves, it is divided into air conduction earphones and bone conduction earphones. When wearing air conduction headphones, the binaural sound signals played by them can easily ensure the separation of the left and right channel signals, and there is almost no crosstalk problem. However, when wearing bone conduction earphones, there will be serious cross-talk interference due to low binaural isolation, that is, the vibration of any bone conduction earphone will be transmitted to the contralateral cochlea to form cross-talk. The existence of crosstalk will lead to distortion of the spatial information of the sound, which will seriously affect the wearer's sound source localization ability and spatial hearing sense.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a method and system for eliminating cross-sound of bone conduction earphones, and the effect of eliminating the cross-sound is good.

For achieving the above object, the present invention provides the following scheme:

A cross-sound cancellation method for bone conduction earphones, comprising:

Obtain the bone conduction transfer matrix when the user wears the bone conduction headset;

Obtain the stopband frequency range of the shape factor according to the frequency range where the singular values of the personalized bone conduction transfer function on both sides of the user's ears appear;

According to the bone conduction transfer matrix, the stopband frequency range of the shape factor and the filter order during filtering, obtain the filtered approximate crosstalk cancellation matrix;

Use the approximate crosstalk cancellation matrix to eliminate the crosstalk of the binaural signal input to the bone conduction earphone to obtain the sound signal for eliminating the crosstalk; the binaural signal is a signal containing the position information of the sound source;

Playing the sound signal for eliminating crosstalk to obtain the sound signal heard by both ears of the user;

Determine the final crosstalk cancellation matrix according to the bone conduction transfer matrix, the approximate crosstalk cancellation matrix and the sound signals heard by the user's binaural ears;

According to the final cross-talk cancellation matrix, cross-talk cancellation is performed on the binaural signals input to the bone conduction earphone.

Optionally, the acquiring the bone conduction transfer matrix when the user wears the bone conduction headset is specifically:

Use bone conduction swept tones to induce stimulation frequency otoacoustic emission signals;

Calculate the bone conduction transfer function between the bone conduction earphone and the user's bilateral inner ear according to the swept frequency stimulation signal and the stimulation frequency otoacoustic emission signal;

A bone conduction transfer matrix is obtained according to the bone conduction transfer function.

Optionally, obtaining an approximate crosstalk cancellation matrix after filtering according to the bone conduction transfer matrix, the stopband frequency range of the shape factor, and the filter order during filtering, specifically: according to the bone conduction The transfer matrix, the stopband frequency range of the shape factor, and the filter order during filtering are used to obtain an approximate crosstalk cancellation matrix after filtering by using a fast deconvolution method in the frequency domain.

Optionally, determining the final crosstalk cancellation matrix according to the bone conduction transfer matrix, the approximate crosstalk cancellation matrix, and the sound signals heard by the user's binaural ears, specifically:

Calculate the channel separation degree according to the bone conduction transfer matrix and the approximate crosstalk cancellation matrix;

Calculate the performance error according to the sound signal and the expected signal heard by the user's ears;

A final crosstalk cancellation matrix is determined according to the channel separation and the performance error.

Optionally, the binaural signal is obtained by synthesizing a single-channel signal according to a head-related transfer function including sound source position information.

Optionally, the single-channel signal is a pseudo-random sequence signal with a maximum length of 1.5 seconds, and the sampling rate of the single-channel signal is 44.1 kHz.

A cross-sound cancellation system for bone conduction earphones, comprising:

The transfer matrix acquisition module is used to acquire the bone conduction transfer matrix when the user wears the bone conduction earphone;

The shape factor stopband frequency range acquisition module is used to obtain the stopband frequency range of the shape factor according to the frequency range where the singular values of the personalized bone conduction transfer function appear on both sides of the user's ears;

a crosstalk cancellation matrix acquisition module, configured to obtain an approximate crosstalk cancellation matrix after filtering according to the bone conduction transfer matrix, the stopband frequency range of the shape factor, and the filter order during filtering;

The first crosstalk cancellation module is used to eliminate the crosstalk of the binaural signal input to the bone conduction earphone by using the approximate crosstalk cancellation matrix; obtain the sound signal for eliminating the crosstalk; the binaural signal includes the position of the sound source Binaural signals of information;

a sound signal acquisition module, used for playing the sound signal for eliminating crosstalk, and obtaining the sound signal heard by both ears of the user;

a crosstalk cancellation matrix determination module, configured to determine a final crosstalk cancellation matrix according to the bone conduction transfer matrix, the approximate crosstalk cancellation matrix and the sound signals heard by the user's binaural ears;

The second crosstalk cancellation module is configured to perform crosstalk cancellation on the binaural signals input to the bone conduction earphone according to the final crosstalk cancellation matrix.

Optionally, the transfer matrix acquisition module includes:

The transmitting signal unit is used to induce the stimulation frequency otoacoustic emission signal by using the bone conduction sweep tone;

The bone conduction transfer function calculation unit is used to obtain the bone conduction transfer function between the bone conduction earphone and the user's bilateral inner ear according to the frequency sweep stimulation signal and the stimulation frequency otoacoustic emission signal;

A bone conduction transfer matrix obtaining unit, configured to obtain a bone conduction transfer matrix according to the bone conduction transfer function.

Optionally, the crosstalk cancellation matrix determination module includes:

a channel separation degree calculation unit, configured to calculate the channel separation degree according to the bone conduction transfer matrix and the approximate crosstalk cancellation matrix;

a performance error calculation unit, configured to calculate the performance error according to the sound signal and the expected signal heard by the user's ears;

a determining unit, configured to determine a final crosstalk cancellation matrix according to the channel separation degree and the performance error.

Optionally, the first cross-talk cancellation module includes a synthesis unit for synthesizing a single-channel signal according to a head-related transfer function including sound source position information to obtain the binaural signal.

According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

The invention discloses a method and system for eliminating cross-sound of bone conduction earphones. The method includes: acquiring a bone conduction transfer matrix when a user wears a bone conduction earphone; acquiring a stopband frequency range of a shape factor; The frequency range of the stop band and the filter order during filtering are obtained, and the approximate crosstalk cancellation matrix after filtering is obtained; the crosstalk cancellation is performed on the binaural signal input to the bone conduction earphone by the approximate crosstalk cancellation matrix; the crosstalk cancellation matrix is obtained. sound signal; play the sound signal for eliminating crosstalk to obtain the sound signal heard by the user's ears; determine the final crosstalk cancellation matrix according to the bone conduction transfer matrix, the approximate crosstalk cancellation matrix and the sound signal heard by the user's two ears; The final crosstalk cancellation matrix performs crosstalk cancellation on the binaural signals input to the bone conduction earphones. The invention adjusts and obtains a suitable crosstalk elimination matrix according to the bone conduction transmission matrix, the approximate crosstalk elimination matrix and the sound signals heard by the user's ears, and can obtain the best crosstalk elimination effect.

Description of drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings required in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some of the present invention. In the embodiments, for those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative labor.

1 is a flowchart of a method for eliminating crosstalk of a bone conduction earphone according to an embodiment of the present invention;

FIG. 2 is a device connection diagram provided by an embodiment of the present invention;

3 is a process diagram of a fast deconvolution method in the frequency domain provided by an embodiment of the present invention;

4 is a block diagram of a crosstalk cancellation algorithm for binaural bone conduction sound provided by an embodiment of the present invention;

5 is a schematic diagram of a left ear channel separation degree provided by an embodiment of the present invention;

6 is a schematic diagram of the separation degree of the right ear channel provided by an embodiment of the present invention;

7 is a schematic diagram of a performance error provided by an embodiment of the present invention;

FIG. 8 is a structural block diagram of a crosstalk cancellation system for a bone conduction earphone according to an embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

The purpose of the present invention is to provide a method and system for eliminating cross-sound of bone conduction earphones, and the effect of eliminating the cross-sound is good.

In order to make the above objects, features and advantages of the present invention more clearly understood, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

Example 1

FIG. 1 is a flowchart of a method for eliminating crosstalk of a bone conduction earphone according to an embodiment of the present invention. As shown in Figure 1, the method includes:

Step 101: Acquire the bone conduction transfer matrix when the user wears the bone conduction earphone. In this embodiment, the otoacoustic emission signal of the stimulation frequency is induced by using the swept frequency sound of bone conduction, and the bone conduction between the bone conduction earphone and the inner ear of the user is obtained according to the swept frequency stimulation signal and the otoacoustic emission signal of the stimulation frequency. A transfer function, and a bone conduction transfer matrix is obtained according to the bone conduction transfer function.

Step 102: Obtain the stopband frequency range of the shape factor according to the frequency range in which the singular values of the personalized bone conduction transfer function on both sides of the user's ears appear.

Step 103: Obtain a filtered approximate crosstalk cancellation matrix according to the bone conduction transfer matrix, the stopband frequency range of the shape factor, and the filter order during filtering. In this embodiment, it is specifically: according to the bone conduction transfer matrix, the stopband frequency range of the shape factor, and the filter order during filtering, use the frequency domain fast deconvolution method to obtain filtered approximate crosstalk cancellation matrix.

Step 104 : Use the approximate crosstalk cancellation matrix to perform crosstalk cancellation on the binaural signal input to the bone conduction earphone to obtain a crosstalk-eliminated sound signal, where the binaural signal is a signal containing sound source position information. In this embodiment, the binaural signals are obtained by synthesizing single-channel signals according to a head-related transfer function (Head-Related Transfer Function, HRTF) that includes sound source position information. The single-channel signal is a pseudo-random sequence signal with a maximum length of 1.5 seconds, and the sampling rate of the single-channel signal is 44.1 kHz.

Step 105: Play the sound signal for eliminating crosstalk to obtain the sound signal heard by the user's ears. In this embodiment, when performing binaural bone conduction sound reproduction, the wearing position of the two bone conduction earphones should be the same as the head position of the subject when measuring BCTF, and the error should not be too large. The wearing position is the mastoid behind the ears on both sides.

Step 106: Determine a final crosstalk cancellation matrix according to the bone conduction transfer matrix, the approximate crosstalk cancellation matrix, and the sound signals heard by the user's binaurally. In this embodiment, the specific steps include: calculating the channel separation degree according to the bone conduction transfer matrix and the approximate crosstalk cancellation matrix, calculating the performance error according to the sound signal and the expected signal heard by the user's binaural The channel separation and the performance error determine the final crosstalk cancellation matrix.

Step 107: Perform crosstalk cancellation on the binaural signals input to the bone conduction earphone according to the final crosstalk cancellation matrix.

Example 2

A subject (24 years old) with normal hearing and no history of external or middle ear disease was selected. First, the bone conduction swept tone was used to induce the stimulation frequency otoacoustic emission signal, and then the pure swept frequency stimulation frequency otoacoustic emission signal was extracted. , the bone conduction impulse response signal is obtained by cross-correlation operation with the swept frequency stimulation signal, and then the bone conduction transfer function between the bone conduction earphone and the subject's bilateral inner ear can be estimated through Fourier transform, and then the bone conduction transfer function can be estimated. Obtain the corresponding bone conduction transfer matrix. The subjects sat in a soundproof room for the listening experiment, and the equipment connections were shown in Figure 2.

Firstly, according to the frequency range of the singular values of the personalized BCTF (bone conduction transfer function) on both sides of the subject, the stopband frequency range of the shape factor B(z) is set to suppress the pathological phenomenon. The B(z) stopband frequency range of the subject's data was set from 0.55 to 7.5 kHz.

A set of causal finite impulse response filters H(z) are obtained through a fast deconvolution algorithm in the frequency domain, so that the product of H(z) and C(z) is approximately the identity matrix I, which can satisfy the signal reaching both ears is approximately equal to the input binaural signals _XL and _XR . FIG. 3 is a process diagram of a fast deconvolution method in the frequency domain provided by an embodiment of the present invention. As shown in FIG. 3 , X(z) is the input binaural signal, H(z) is the crosstalk cancellation filter, and v(z) ) is the input signal of the bone conduction earphone, w(z) is the estimated signal, d(z) is the expected signal, e(z) is the error signal, A(z) is the objective function, z- ^m represents the target delay m samples point. The design of the crosstalk cancellation filter can be equivalent to the problem of minimizing the cost function, and the cost function is:

J=E+β ₁ V=e ^H (z)e(z)+β ₁ v ^H (z)v(z) (1)

where the superscript H represents the conjugate transpose operator, e ^H (z)e(z) is the performance error term, β ₁ v ^H (z)v(z) is the cost term, where β ₁ is the regularization parameter, Indicates the weight of the frequency weighting function, which is often used to limit the filter gain and matrix invertibility of the crosstalk cancellation system. This parameter can be constant or related to frequency. In order to facilitate the control of the gain of each frequency point, consider adding a frequency-dependent shape factor B(z), that is, decompose β ₁ into a regularized constant gain factor β and a shape factor B(z). According to the condition that J is minimized, the approximate solution of the crosstalk cancellation matrix H can be solved as:

^{H(z)=(CH(z)C(z)+βB H (} z)B(z)) ^{-1 CH} (z)z ^-m (2)

Among them, C represents the bone conduction transfer function of the subject including the transmission characteristics of the bone conduction vibrator, z- ^m is used to ensure the causality of the crosstalk cancellation filter, m in this embodiment is 3.125ms, β is the regularization constant gain factor, 0<β≤1, the shape factor B(z) is the filter gain used by the Z-domain filter to limit the selected frequency band to prevent ill-conditioned phenomena. Digital sampling is performed on the basis of formula (2), and the coefficients of the corresponding discrete frequency points can be obtained, as shown in the following formula:

^{H(z)=[CH(k)C(k)+βB H (} k)B(k)] ^{-1 CH} (k)exp{[-j2π(k-1)m]/N} (3 )

Among them, k=1...N, where N is the order of the filter.

In this embodiment, the regularization constant gain factor β is taken as 10 ^-2 , 10 ^-4 , 10 ^-6 and 10 ^-8 respectively, to observe the influence of different β on the performance of the Crosstalk Cancellation System (CCS), And choose the best β value by channel separation (Channel Separation, CS) and performance error (Performance Error, PE). The filter order is 2048.

The performance of the crosstalk cancellation system is adjusted by CS and PE.

CS represents the amplitude ratio between the crosstalk and the direct sound, and the unit is expressed in dB. The smaller the value, the better the channel separation, and vice versa. The formula for calculating the channel separation of the left and right ears is:

Among them, the bone conduction transmission matrix C and the crosstalk cancellation matrix H are respectively

PE represents the ratio of the output of the system to the desired signal in the frequency domain, and the unit is expressed in dB. The closer the value is to 0dB, the smaller the performance error, and vice versa. The formula for calculating the performance error is:

where X _W is the signal heard by the user's ears, and X _D is the desired signal.

FIG. 4 is a block diagram of a crosstalk cancellation algorithm for binaural bone conduction sound provided by an embodiment of the present invention. After obtaining the crosstalk cancellation matrix H by the fast deconvolution algorithm, the binaural signals _XL and X _R are obtained by filtering the single-channel signal S through HRTF, where S is the MLS signal of 1.5 seconds, and the sampling rate is 44.1 kHz. The input signals Y _L and Y _R of the left and right bone conduction earphones are obtained after filtering by the cross-talk cancellation filter H, and then the subject conducts a subjective listening experiment as shown in Figure 2, and transmits Y _L and Y _R respectively. Give the bone conduction earphones worn on the left and right mastoids. At the same time, according to formulas (4) (5) (8), the CS and PE indicators are calculated, and the shape factor B(z), the regularization constant gain factor β and the filter order N are adjusted according to the calculation results.

FIG. 5 is a schematic diagram of the separation degree of the left ear channel according to an embodiment of the present invention, and FIG. 6 is a schematic diagram of the separation degree of the right ear channel according to an embodiment of the present invention. The curve trends of the two figures are similar. It can be seen from the figure that the smaller the regularized constant gain factor β, the higher the channel separation, that is, the better the crosstalk cancellation effect. This is because the smaller β is, the proportion of the cost term in equation (1) is correspondingly reduced, thereby improving the performance of error approximation and the channel separation degree. When β≤10 ^-4 , the CS value is basically below -50dB. When β=10 ^-2 , the CS value of frequencies above 5kHz is above -50dB, and the four curves all have a large valley near 2kHz. FIG. 7 is a schematic diagram of performance error provided by an embodiment of the present invention. When β≥10 ⁻⁴ , the PE value is close to 0 dB, that is, the performance error is small, and the fluctuation is large in the mid-low frequency range before 4 kHz. As the frequency increases, The performance error becomes larger around 9-10 kHz; when β < 10 ^-4 , the performance error is large in the first 5 kHz, and the PE value is close to 0 dB after this frequency, which may be due to the fact that the stopband range of the shape factor is not set properly. According to the CS and PE indicators of this subject, it can be judged that the optimal regularization constant gain factor suitable for her ranges from β≥10 ^-4 , and the actual most suitable parameter value depends on the shape factor and filter order. The comprehensive influence effect of other parameters can be selected.

Example 3

FIG. 8 is a structural block diagram of a crosstalk cancellation system for a bone conduction earphone according to an embodiment of the present invention. As shown in FIG. 8 , the system includes:

The transfer matrix acquisition module 201 is configured to acquire the bone conduction transfer matrix when the user wears the bone conduction earphone.

In this embodiment, the transfer matrix acquisition module 201 includes: an emission signal unit for inducing a stimulation frequency otoacoustic emission signal by using a bone conduction swept frequency tone; a bone conduction transfer function calculation unit for eliciting a stimulation frequency otoacoustic emission signal according to the swept frequency stimulation signal and the stimulation frequency otoacoustic emission signal to obtain the bone conduction transfer function between the bone conduction earphone and the user's bilateral inner ear; the bone conduction transfer matrix obtaining unit is used for obtaining the bone conduction transfer matrix according to the bone conduction transfer function.

The shape factor stopband frequency range obtaining module 202 is configured to obtain the stopband frequency range of the shape factor according to the frequency range where singular values of the personalized bone conduction transfer function appear on both sides of the user's ears.

The crosstalk cancellation matrix acquisition module 203 is configured to obtain a filtered approximate crosstalk cancellation matrix according to the bone conduction transfer matrix, the stopband frequency range of the shape factor, and the filter order during filtering.

The first crosstalk cancellation module 204 is configured to use the approximate crosstalk cancellation matrix to perform crosstalk cancellation on the binaural signal input to the bone conduction earphone to obtain a crosstalk-eliminated sound signal; the binaural signal includes a sound source location information signal.

In this embodiment, the first cross-talk cancellation module 204 includes a synthesis unit, configured to obtain the binaural signal by synthesizing a single-channel signal according to a head-related transfer function including sound source position information.

The sound signal acquisition module 205 is configured to play the sound signal for eliminating crosstalk to obtain the sound signal heard by both ears of the user.

The crosstalk cancellation matrix determining module 206 is configured to determine the final crosstalk cancellation matrix according to the bone conduction transfer matrix, the approximate crosstalk cancellation matrix and the sound signals heard by the user's binaural ears.

In this embodiment, the crosstalk cancellation matrix determination module 206 includes: a channel separation degree calculation unit, configured to calculate the channel separation degree according to the bone conduction transfer matrix and the approximate crosstalk cancellation matrix; a performance error calculation unit, The performance error is calculated according to the sound signal and the expected signal heard by the user's ears; the determining unit is used for determining the final crosstalk cancellation matrix according to the channel separation degree and the performance error.

The second crosstalk cancellation module 207 is configured to perform crosstalk cancellation on the binaural signals input to the bone conduction earphone according to the final crosstalk cancellation matrix.

According to the embodiments of the present invention, the present invention discloses the following technical effects:

1. The present invention studies crosstalk cancellation in binaural bone conduction sound reproduction according to the bilateral BCTF data measured on the head of different subjects, which can better improve the effect of binaural bone conduction sound reproduction.

2. The present invention adjusts and obtains a suitable crosstalk cancellation matrix according to the bone conduction transmission matrix, the approximate crosstalk cancellation matrix and the audio signals heard by the user's ears, and can obtain the best crosstalk cancellation effect.

3. The applied frequency domain fast deconvolution algorithm has a small amount of calculation, is simple and effective, and has high practicability.

The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the points that are different from other embodiments, and the similar parts between the various embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant part can be referred to the description of the method.

The principles and implementations of the present invention are described herein using specific examples. The descriptions of the above embodiments are only used to help understand the method and the core idea of the present invention; meanwhile, for those skilled in the art, according to the present invention There will be changes in the specific implementation and application scope. In conclusion, the contents of this specification should not be construed as limiting the present invention.

Claims (10)

1. A method for eliminating crosstalk of a bone conduction headset is characterized by comprising the following steps:

acquiring a bone conduction transmission matrix when a user wears a bone conduction earphone;

acquiring a stop band frequency range of the shape factor according to a frequency range of appearance of singular values of personalized bone conduction transfer functions at two sides of ears of a user;

obtaining a filtered approximate cross talk elimination matrix according to the bone conduction transfer matrix, the stop band frequency range of the shape factor and the filter order during filtering;

using the approximate cross-talk elimination matrix to eliminate cross-talk of the binaural signals input to the bone conduction earphone, so as to obtain sound signals eliminating the cross-talk; the binaural signal is a signal containing sound source position information;

playing the sound signals for eliminating the crosstalk to obtain sound signals heard by the two ears of the user;

determining a final cross talk elimination matrix according to the bone conduction transfer matrix, the approximate cross talk elimination matrix and the sound signals heard by the ears of the user;

and carrying out crosstalk elimination on the binaural signals input to the bone conduction earphone according to the final crosstalk elimination matrix.

2. The method for eliminating crosstalk according to claim 1, wherein the obtaining of the bone conduction transmission matrix when the user wears the bone conduction headset specifically includes:

inducing a stimulation frequency otoacoustic emission signal with bone conduction sweep tones;

according to the sweep frequency stimulation signal and the stimulation frequency otoacoustic emission signal, a bone conduction transfer function from the bone conduction earphone to the inner ears on the two sides of the user is worked out;

and obtaining a bone conduction transfer matrix according to the bone conduction transfer function.

3. The method according to claim 1, wherein the filtered approximate cross talk cancellation matrix is obtained according to the bone conduction transfer matrix, the stopband frequency range of the shape factor, and the filter order during filtering, and specifically includes: and obtaining a filtered approximate cross talk elimination matrix by utilizing a frequency domain fast deconvolution method according to the bone conduction transfer matrix, the stopband frequency range of the shape factor and the filter order during filtering.

4. The crosstalk elimination method according to claim 1, wherein the determining a final crosstalk elimination matrix according to the bone conduction transfer matrix, the approximate crosstalk elimination matrix, and the sound signals heard by both ears of the user comprises:

calculating a channel separation degree according to the bone conduction transfer matrix and the approximate crosstalk elimination matrix;

calculating a performance error from the sound signals heard by both ears of the user and the desired signal;

and determining a final crosstalk elimination matrix according to the channel separation degree and the performance error.

5. The method of claim 1, wherein the binaural signal is synthesized from a mono signal according to a head-related transfer function including sound source position information.

6. The method of claim 5, wherein the one-way signal is a maximum length pseudo-random sequence signal of 1.5 seconds, and the sampling rate of the one-way signal is 44.1 kHz.

7. A crosstalk cancellation system for a bone conduction headset, comprising:

the transmission matrix acquisition module is used for acquiring a bone conduction transmission matrix when a user wears the bone conduction earphone;

the shape factor stopband frequency range acquisition module is used for acquiring the stopband frequency range of the shape factor according to the frequency range of the appearance of the singular value of the personalized bone conduction transfer function at two sides of the ears of the user;

the crosstalk elimination matrix obtaining module is used for obtaining a filtered approximate crosstalk elimination matrix according to the bone conduction transfer matrix, the stop band frequency range of the shape factor and the filter order during filtering;

a first crosstalk elimination module for performing crosstalk elimination on a binaural signal input to the bone conduction headset by using the approximate crosstalk elimination matrix; obtaining a sound signal for eliminating crosstalk; the binaural signal is a binaural signal containing sound source position information;

the sound signal acquisition module is used for playing the sound signal for eliminating the crosstalk to obtain sound signals heard by two ears of the user;

a cross talk elimination matrix determination module for determining a final cross talk elimination matrix according to the bone conduction transfer matrix, the approximate cross talk elimination matrix and the sound signals heard by the ears of the user;

and the second crosstalk elimination module is used for performing crosstalk elimination on the binaural signals input to the bone conduction earphone according to the final crosstalk elimination matrix.

8. The crosstalk cancellation system according to claim 7, wherein the transfer matrix acquisition module comprises:

a transmitting signal unit for inducing a stimulation frequency otoacoustic transmitting signal using bone conduction sweep tones;

the bone conduction transfer function calculation unit is used for solving a bone conduction transfer function from the bone conduction earphone to the inner ears on the two sides of the user according to the sweep frequency stimulation signal and the stimulation frequency otoacoustic emission signal;

and the bone conduction transfer matrix acquisition unit is used for obtaining a bone conduction transfer matrix according to the bone conduction transfer function.

9. The crosstalk cancellation system according to claim 7, wherein the crosstalk cancellation matrix determination module comprises:

a channel separation degree calculation unit for calculating a channel separation degree according to the bone conduction transfer matrix and the approximate cross-talk elimination matrix;

a performance error calculation unit for calculating a performance error from the sound signal heard by both ears of the user and the desired signal;

and the determining unit is used for determining a final crosstalk elimination matrix according to the channel separation degree and the performance error.

10. The crosstalk cancellation system according to claim 7, wherein said first crosstalk cancellation module comprises a synthesis unit for synthesizing said binaural signal by a mono signal according to a head-related transfer function containing sound source position information.

Images