Real-time optimization method and system for self-adaptive personal sound area in dynamic environment

Abstract

The invention discloses a real-time optimization method and system for a self-adaptive personal sound area in a dynamic environment, and relates to the field of sound field control, and the method comprises the steps: 1, constructing a sound field cost function, the target is that the error between a B-region reconstruction signal pB [n] and a B-region desired signal xB [n] and the error between a D-region reconstruction signal pD [n] and a D-region desired signal xD [n] are both smaller than a preset threshold, pB [n] = RBW, and pD [n] = RDW; step 2, updating a control filter W to obtain a driving signal u of the loudspeaker array; and step 3, playing the driving signal u through a loudspeaker array and spreading the driving signal u to a B-area microphone array and a D-area microphone array through air, and taking a difference between a B-area reconstruction signal and an actual signal obtained by the B-area microphone array and a difference between a D-area reconstruction signal and an actual signal obtained by the D-area microphone array as error signals, and continuously repeating the step 2 and the step 3 by using the RRLS method, updating the RB and the RD in real time, and realizing the tracking and compensation of the acoustic environment change.

Description

Technical Field

[0001] The present invention relates to the field of sound field control, and in particular to a real-time optimization method and system for adaptively optimizing personal sound zones in a dynamic environment. Background Art

[0002] Personal sound zone systems can generate independent listening areas for multiple users within the same physical space, and have broad application prospects in areas such as automotive cockpits, mobile devices, and public spaces. Currently, most related research focuses on achieving optimal performance in fixed acoustic environments and assumes that the precise transfer function from the speaker to the microphone is known in advance. However, in real-world applications, the acoustic environment changes over time. For example, furniture movement, people moving around, and temperature fluctuations will cause the transfer function to vary, thus limiting the performance and practicality of personal sound zone systems.

[0003] Therefore, those skilled in the art are committed to developing a real-time optimization system that can adapt to individual vocal zones in a dynamic acoustic environment to solve the above-mentioned defects in the prior art. Summary of the Invention

[0004] In view of the above-mentioned defects of the prior art, the technical problem to be solved by the present invention is how to solve the problem of performance degradation of the personal vocal zone system in a dynamic acoustic environment due to the unknown or changing acoustic environment, so as to enable the personal vocal zone system to provide users with independent high-quality listening areas stably and efficiently in different environments.

[0005] To achieve the above object, the present invention provides a real-time optimization method for adaptively optimizing personal vocal zones in a dynamic environment, the method comprising the following steps:

[0006] Step 1: Construct a sound field cost function. The goal of the sound field cost function is to reconstruct the signal p in area B. B [n] and the expected signal x in area B B [n] and the D-region reconstructed signal p D [n] and the expected signal x in area D D The errors between [n] are all less than the pre-set threshold;

[0007] Among them, p B [n] = R B W, p D [n] = R D W, R B is the acoustic transmission matrix from the loudspeaker array to the microphone array in area B, R D is the acoustic transmission matrix from the loudspeaker array to the microphone array in zone D, and W is the control filter;

[0008] Step 2: Update the control filter W to obtain the driving signal u of the speaker array;

[0009] Step 3: The driving signal u is played through the speaker array and propagated through the air to the microphone array in area B and the microphone array in area D. The difference between the reconstructed signal in area B and the actual signal obtained by the microphone array in area B and the difference between the reconstructed signal in area D and the actual signal obtained by the microphone array in area D are used as error signals to perform RRLS online system identification. The RRLS method is used to continuously repeat steps 2 and 3 to update the acoustic transfer matrix R in real time. B and R D , to achieve tracking and compensation of changes in acoustic environment.

[0010] Furthermore, in step 1, the sound field cost function is obtained by minimizing the reconstructed signal p in region B. B [n] and the expected signal x in area B B The square of the error between [n] and the D-region reconstructed signal p D [n] and the desired signal x in the D region D The sum of the squares of the errors between [n] is used to achieve the goal of the sound field cost function.

[0011] Furthermore, the weighted objective of the sound field cost function is to minimize the sum of squares of the weighted error vector, and find the control filter W that minimizes the weighted objective. The weighted objective J[n] is specifically:

[0012] J[n]=E{β||p B [n]-x B [n]|| 2 +(1-β)||p D [n]-x D [n]|| 2}

[0013] Where β∈[0,1] is the weight of the error in area B, and (1-β) is the weight of the error in area D. When β=1, only the sound pressure matching in area B is focused on, and the sound pressure in area D is not controlled. When β=0.5, the sound pressures in both areas B and D are optimized simultaneously. E represents the mathematical expectation.

[0014] Furthermore, let x D [n] = 0, so that the D area can not hear the sound, and the weighted target J[n] becomes:

[0015] J[n]=E{β||p B [n]-x B [n]|| 2 +(1-β)||pD [n]|| 2}

[0016] Then substitute the signal model into the weighted target J[n]:

[0017] J[n]=W T [βz B +(1-β)z D ]W-β(W T Q B +Q B T W)+βE{X B T X B}

[0018] Among them, z B =E{R B T R B}, z D =E{R D T R D}, Q B =E{R B T X B};

[0019] Finally, add the regularization term ρW T W, transforms the weighted target J[n] into:

[0020] J[n]=W T [βz B +(1-β)z D ]W-β(W T Q B +Q B T W)+βE{X B T X B}+pW T W.

[0021] Furthermore, in step 2, the control filter W is updated using a stochastic gradient descent method.

[0022] Furthermore, in the stochastic gradient descent method, the gradient of the weighted target J[n] is calculated

[0023]

[0024] Update W(n+1) again:

[0025]

[0026] Where μ is the step size, which must satisfy:

[0027]

[0028] To ensure convergence, H B is the matrix representation of the transfer function from the loudspeaker array to the microphone array in zone B, H D is a matrix representation of the transfer function from the loudspeaker array to the D-zone microphone array.

[0029] Furthermore, a momentum variable v is introduced to record the accumulated information of the previous gradient;

[0030] The update rules of the momentum variable v and the control filter W are as follows:

[0031]

[0032] Among them, v n is the momentum at the nth iteration, α is the momentum coefficient, which controls the degree of retention of the previous momentum information; η is the learning rate;

[0033] According to the new momentum v n+1 Update W(n+1) to move in the direction of negative momentum:

[0034] W(n+1)=W(n)-v n+1

[0035] When the direction of the gradient remains consistent over multiple iterations, momentum accumulates, causing the step size μ of the parameter update to gradually increase, thereby accelerating convergence.

[0036] Furthermore, in step 2, the control filter W is updated using Newton's method.

[0037] Furthermore, in the RRLS online system identification of step 3, a regularization term is introduced, and according to the recursive least squares RLS algorithm, the cost function ξ of the object model from the speaker array to the m-th microphone in area B is m [n] is defined as:

[0038]

[0039] Among them, λ represents the forgetting factor, δ represents the regularization parameter, and p B,m [n] represents the sound pressure at the mth microphone, represents the real-time estimated transfer function from the loudspeaker array to the mth microphone;

[0040] Calculate the prior estimation error ξ B[n]:

[0041]

[0042] The driving matrix of the transfer function from the loudspeaker array to the mth microphone in area B is The update is:

[0043]

[0044] Calculate the gain vector k[n]:

[0045]

[0046] Update the covariance matrix

[0047]

[0048] in,

[0049] Similarly, the driving matrix of area D can be obtained Update:

[0050]

[0051] The present invention also provides a real-time optimization system for adaptive personal vocal zones in a dynamic environment, comprising a speaker array consisting of L speakers and two microphone arrays divided into zones B and D, wherein the microphone arrays respectively comprise M B and M D A microphone, characterized in that it also includes the real-time optimization method for adaptive personal vocal zone in a dynamic environment as described in any of the above items.

[0052] The present invention provides a method and system for real-time optimization of adaptive personal vocal zones in a dynamic environment, which at least have the following technical effects:

[0053] 1. The technical solution provided by this invention has dynamic adaptability, including real-time response to environmental changes and dynamic adjustment of control filters. The real-time response to environmental changes is achieved through online RRLS modeling. The system can track changes in the acoustic path (such as personnel movement and temperature fluctuations) in real time and update the transfer function matrix, avoiding the performance degradation caused by fixed models in traditional methods. In the dynamic adjustment of the control filter, the LMS algorithm quickly adjusts the control weights based on the current environmental model to maintain the quality of the sound field reconstruction.

[0054] 2. The technical solution provided by the present invention is robust and stable. Regularization anti-overfitting includes the regularization term in RRLS to suppress model overfitting under noise interference and ensure stable parameter estimation.

[0055] 3. The technical solution provided by the present invention has multiple independent sound zones for users. By expanding the multi-channel control architecture, independent sound fields can be generated for different users, such as isolating the driver's and passenger's listening areas.

[0056] The concept, specific structure and technical effects of the present invention will be further described below in conjunction with the accompanying drawings to fully understand the purpose, characteristics and effects of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a flow chart of a real-time optimization method according to a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0058] The following describes several preferred embodiments of the present invention with reference to the accompanying drawings to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms of embodiments, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0059] Embodiments of the present invention provide an adaptive personal sound zone method and system suitable for dynamic environments with online system identification. The method uses the least mean squares (LMS) algorithm to control the update of the control filter and the regularized recursive least squares (RRLS) algorithm to enable the system model to adapt to changes in the acoustic environment, creating independent listening areas for multiple users in the same space.

[0060] Example 1

[0061] like Figure 1 As shown, the embodiment of the present invention provides a real-time optimization method for adaptively optimizing personal vocal zones in a dynamic environment, comprising the following steps:

[0062] Step 1: Construct the sound field cost function. The goal of the sound field cost function is to reconstruct the signal p in area B. B [n] and the expected signal x in area B B [n] and the D-region reconstructed signal p D [n] and the expected signal x in area D D The errors between [n] are all less than the pre-set threshold;

[0063] Among them, p B [n] = R B W, p D [n] = R D W, R B is the acoustic transmission matrix from the loudspeaker array to the microphone array in area B, R D is the acoustic transmission matrix from the loudspeaker array to the microphone array in zone D, and W is the control filter;

[0064] Step 2: Update the control filter W to obtain the driving signal u of the speaker array;

[0065] Step 3: The driving signal u is played through the speaker array and propagated through the air to the microphone arrays in area B and area D. The difference between the reconstructed signal in area B and the actual signal obtained by the microphone array in area B and the difference between the reconstructed signal in area D and the actual signal obtained by the microphone array in area D are used as error signals to perform RRLS online system identification. Steps 2 and 3 are continuously repeated using the RRLS method to update the acoustic transfer matrix R in real time. B and R D , to achieve tracking and compensation of changes in acoustic environment.

[0066] Example 2

[0067] Based on Example 1, in step 1, the sound field cost function is to make the reconstructed signal as close as possible to the desired signal, which can usually be achieved by minimizing the sum of squared errors between them, and the importance of the errors in the two regions is measured according to different weights.

[0068] In particular, the sound field cost function is calculated by minimizing the reconstruction signal p in region B. B [n] and the expected signal x in area B B The square of the error between [n] and the D-region reconstructed signal p D [n] and the expected signal x in area D D The sum of the squares of the errors between [n] is used to achieve the goal of the sound field cost function.

[0069] In particular, the weighted objective of the sound field cost function is to minimize the sum of squares of the weighted error vector, and find the control filter W that minimizes the weighted objective. The weighted objective J[n] is specifically:

[0070] J[n]=E{β||p B [n]-x B [n]|| 2 +(1-β)||p D [n]-x D [n]|| 2}

[0071] Here, β∈[0,1] is the weight of the error in region B, and (1-β) is the weight of the error in region D. When β=1, the focus is solely on sound pressure matching in region B, with no control over the sound pressure in region D. When β=0.5, the sound pressure in both regions is optimized, balancing the importance of the errors in regions B and D through the weight β, ensuring that the reconstructed signal is optimal in terms of weighted mean square error. E represents the mathematical expectation, meaning that the cost is considered in the sense of statistical average. In practical applications, acoustic signals may be affected by random factors such as noise and environmental changes. By calculating the expectation, the objective function can be optimized for the average performance across a wide range of possible scenarios, not just for a specific situation.

[0072] In particular, let x D [n] = 0, so that area D cannot hear the sound, and the weighted target J[n] becomes:

[0073] J[n]=E{β||p B [n]-x B [n]|| 2 +(1-β)||p D [n]|| 2}

[0074] Then substitute the signal model into the weighted target J[n]:

[0075] J[n]=W T [βz B +(1-β)z D ]W-β(W T Q B +Q B T W)+βE{X B T X B}

[0076] Among them, z B =E{R B T R B}, z D =E{R D T R D}, Q B =E{R B T X B};

[0077] Finally, add the regularization term pW T W, transforms the weighted target J[n] into:

[0078] J[n]=W T [βz B +(1-β)zD ]W-β(W T Q B +Q B T W)+βE{X B T X B}+pW T W.

[0079] Example 3

[0080] Based on embodiment 1 or 2, in step 2, the control filter W is updated using the stochastic gradient descent method.

[0081] In stochastic gradient descent, the gradient of the weighted target J[n] is calculated

[0082]

[0083] Update W(n+1) again:

[0084]

[0085] Where μ is the step size, which must satisfy:

[0086]

[0087] To ensure convergence, H B is the matrix representation of the transfer function from the loudspeaker array to the microphone array in area B, H D It is the matrix representation of the transfer function from the loudspeaker array to the microphone array in zone D.

[0088] The step size controls the magnitude of the filter parameter update during each iteration. A smaller step size can make the algorithm converge more stably but may result in slower convergence. A larger step size can accelerate convergence but may also cause the algorithm to oscillate or even fail to converge. The step size setting is determined after careful consideration to meet the simulation requirements for algorithm stability and convergence speed.

[0089] In particular, the stochastic gradient descent method with momentum comprehensively considers the previous gradient information, which overcomes the shortcomings of the traditional stochastic gradient descent method to a certain extent, can more effectively optimize the objective function, accelerate convergence and reduce oscillations. Specifically, a momentum variable v is introduced to record the accumulated information of the previous gradient;

[0090] The update rules for the momentum variable v and the control filter W are as follows:

[0091]

[0092] Among them, vn is the momentum at the nth iteration, α is the momentum coefficient, which controls the degree of retention of the previous momentum information; η is the learning rate, α is usually around 0.9s;

[0093] According to the new momentum v n+1 Update W(n+1) to move in the direction of negative momentum:

[0094] W(n+1)=W(n)-v n+1

[0095] When the direction of the gradient remains consistent across multiple iterations, momentum accumulates, gradually increasing the step size μ of the parameter update, thereby accelerating convergence. When the direction of the gradient changes, momentum acts as a buffer, preventing wild oscillations in the parameter update. Because momentum incorporates information about previous gradients, it prevents the parameters from drastically changing direction due to the current gradient.

[0096] In particular, in step 2, the control filter W may also be updated using the Newton method.

[0097] Example 3

[0098] In real-world dynamic environments, acoustic characteristics constantly change from the moment a speaker emits sound to the moment a microphone receives it. Accurate sound field control must account for this variation to avoid errors caused by changes in the acoustic transfer function. To address this issue, real-time system identification based on the RLS algorithm is introduced. This method continuously updates model parameters based on real-world data, thereby tracking the system's dynamic changes in real time.

[0099] Based on Examples 1, 2, or 3, step 3 updates the transfer function matrices of areas B and D through RRLS online system identification. Specifically, the drive signal obtained in step 2 is played through the speaker array, propagates through areas B and D through the air, and is recorded by two microphone arrays for monitoring sound placed in areas B and D. The difference between the expected signal and the actual signal recorded by the microphone array is used as an error signal. By continuously comparing the difference between the two, the acoustic transfer matrix is ​​updated so that the model can better fit the actual situation.

[0100] In particular, in the RRLS online system identification in step 3, a regularization term is introduced, and according to the recursive least squares RLS algorithm, the cost function ξ of the object model from the loudspeaker array to the m-th microphone in zone B is m [n] is defined as:

[0101]

[0102] Among them, λ represents the forgetting factor, δ represents the regularization parameter, and p B,m[n] represents the sound pressure at the mth microphone, represents the real-time estimated transfer function from the loudspeaker array to the mth microphone;

[0103] Calculate the prior estimation error ξ B [n]:

[0104]

[0105] The driving matrix of the transfer function from the loudspeaker array to the mth microphone in area B The update is:

[0106]

[0107] Calculate the gain vector k[n]:

[0108]

[0109] Update the covariance matrix

[0110]

[0111] in,

[0112] Similarly, the driving matrix of area D can be obtained Update:

[0113]

[0114] Through repeated cycles of steps 2 and 3, the acoustic transfer matrix is ​​updated and the control filter is adjusted in real time to reconstruct the desired signal through the speaker array.

[0115] Example 4

[0116] The embodiment of the present invention further provides a real-time optimization system for adaptively adjusting personal vocal zones in a dynamic environment, comprising a speaker array consisting of L speakers and two microphone arrays divided into zones B and D, wherein the microphone arrays respectively include M B and M D A microphone, characterized in that it also includes the real-time optimization method for adaptive personal vocal zone in a dynamic environment according to any one of embodiments 1 to 3.

[0117] The preferred embodiments of the present invention have been described in detail above. It should be understood that numerous modifications and variations based on the concepts of the present invention are possible without inventive effort by those skilled in the art. Therefore, any technical solution that can be derived by one skilled in the art through logical analysis, reasoning, or limited experimentation based on the concepts of the present invention and the prior art should be within the scope of protection defined by the claims.

Claims

1. A real-time optimization method for adaptive personal vocal zones in a dynamic environment, characterized in that: The method comprises the following steps: Step 1: Construct a sound field cost function. The goal of the sound field cost function is to reconstruct the signal p in area B. B [n] and the expected signal x in area B B [n] and the D-region reconstructed signal p D [n] and the expected signal x in area D D The errors between [n] are all less than the pre-set threshold; Among them, p B [n] = R B W, p D [n] = r D w, R B is the acoustic transmission matrix from the loudspeaker array to the microphone array in area B, R d is the acoustic transmission matrix from the loudspeaker array to the microphone array in zone D, and W is the control filter; Step 2: Update the control filter W to obtain the driving signal u of the speaker array; Step 3: The driving signal u is played through the speaker array and propagated through the air to the microphone array in area B and the microphone array in area D. The difference between the reconstructed signal in area B and the actual signal obtained by the microphone array in area B and the difference between the reconstructed signal in area D and the actual signal obtained by the microphone array in area D are used as error signals to perform RRLS online system identification. The RRLS method is used to continuously repeat steps 2 and 3 to update the acoustic transfer matrix R in real time. B and R D , to achieve tracking and compensation of changes in acoustic environment.

2. The method for real-time optimization of adaptive personal vocal zones in a dynamic environment according to claim 1, characterized in that: In step 1, the sound field cost function is obtained by minimizing the reconstructed signal p in region B. B [n] and the expected signal x in area B B The square of the error between [n] and the D-region reconstructed signal p D [n] and the desired signal x in the D region D The sum of the squares of the errors between [n] is used to achieve the goal of the sound field cost function.

3. The real-time optimization method for adaptively optimizing personal vocal zones in a dynamic environment according to claim 2, wherein: The weighted target of the sound field cost function is to minimize the sum of squares of the weighted error vector and find the control filter W that minimizes the weighted target. The weighted target J[n] is specifically: J[n]=E{β‖p B [n]-x B [n]‖ 2 +(1−β)‖p D [n]-x D [n]‖ 2 } Where β∈[0,1] is the weight of the error in area B, and (1-β) is the weight of the error in area D. When β=1, only the sound pressure matching in area B is focused on, and the sound pressure in area D is not controlled. When β=0.5, the sound pressures in both areas B and D are optimized simultaneously. E represents the mathematical expectation.

4. The method for real-time optimization of adaptive personal vocal zones in a dynamic environment according to claim 3, wherein: Let x D [n] = 0, so that the D area can not hear the sound, and the weighted target J[n] becomes: J[n]=E{β‖p B [n]-x B [n]‖ 2 +(1−β)‖p D [n]‖ 2 } Then substitute the signal model into the weighted target J[n]: J[n]=W T [βz B +(1-β)z D ]W-β(W T Q B +Q B T W)+βE{X B T X B } where z B = E{R B T R B}, z D = E{R D T R D}, Q B = E{R B T X B}; Finally, add the regularization term ρW T W, transforms the weighted target J[n] into: J[n]=W T [βz B +(1-β)z D ]W-β(W T Q B +Q B T W)+βE{X B T X B }+ρW T W。 5. The real-time optimization method for adaptively optimizing personal vocal zones in a dynamic environment according to claim 4, characterized in that: In step 2, the control filter W is updated using the stochastic gradient descent method.

6. The method for real-time optimization of adaptive personal vocal zones in a dynamic environment according to claim 5, characterized in that: In the stochastic gradient descent method, the gradient of the weighted target J[n] is calculated Update W(n+1) again: Where μ is the step size, which must satisfy: To ensure convergence, H B is the matrix representation of the transfer function from the loudspeaker array to the microphone array in zone B, H D is a matrix representation of the transfer function from the loudspeaker array to the D-zone microphone array.

7. The real-time optimization method for adaptively optimizing personal vocal zones in a dynamic environment according to claim 6, characterized in that: By introducing a momentum variable v to record the accumulated information of the previous gradient; The update rules of the momentum variable c and the control filter W are as follows: Among them, v m is the momentum at the nth iteration, α is the momentum coefficient, which controls the degree of retention of the previous momentum information; η is the learning rate; According to the new momentum v n+1 Update W(n+1) to move in the direction of negative momentum: W(n+1)=W(n)-v m+1 When the direction of the gradient remains consistent over multiple iterations, momentum accumulates, causing the step size μ of the parameter update to gradually increase, thereby accelerating convergence.

8. The method for real-time optimization of adaptive personal vocal zones in a dynamic environment according to claim 4, characterized in that: In step 2, the control filter W is updated using Newton's method.

9. The method for real-time optimization of adaptive personal vocal zones in a dynamic environment according to claim 1, wherein: In the RRLS online system identification of step 3, a regularization term is introduced, and according to the recursive least squares RLS algorithm, the cost function ξ of the object model from the speaker array to the mth microphone in area B is m [n] is defined as: Among them, λ represents the forgetting factor, δ represents the regularization parameter, and p B,m [n] represents the sound pressure at the mth microphone, represents the real-time estimated transfer function from the loudspeaker array to the mth microphone; Calculate the prior estimation error ξ B [n]: The driving matrix of the transfer function from the loudspeaker array to the mth microphone in area B is The update is: Calculate the gain vector k[n]: Update the covariance matrix in, Similarly, the driving matrix of area D can be obtained Update:

10. A real-time optimization system for adaptive personal vocal zones in a dynamic environment, comprising a speaker array consisting of L speakers and two microphone arrays divided into zones B and D, wherein: The microphone arrays include M B and M D A microphone, characterized in that it also includes the real-time optimization method for adaptive personal vocal zone in a dynamic environment according to any one of claims 1 to 9.

Images