Adaptive sound zone control method based on low-rank decomposition

Abstract

This invention provides an adaptive acoustic zone control method based on low-rank decomposition. Addressing the problems of existing adaptive acoustic zone control methods, such as excessively high-dimensional control filters, high computational complexity in adaptive updates, and difficulty in achieving rapid adjustment during sudden changes in the sound field environment, this method proposes to decompose the high-dimensional control filter into several shorter sub-filters using low-rank decomposition. An adaptive acoustic zone control model with these sub-filters as variables is established and solved. The resulting sub-filters are then integrated into a final control filter to drive the speaker array and achieve acoustic zone control. This method can rapidly complete the adaptive update of the control filter when the sound field environment changes abruptly, significantly improving the response speed while maintaining good acoustic contrast and bright-zone signal fidelity.

Description

Technical Field

[0001] This invention relates to the field of sound field control technology, and in particular to an adaptive sound zone control method based on low-rank decomposition. Background Technology

[0002] Sound zone control technology is a key technology that enables the creation of multiple independent and non-interfering acoustic zones within the same shared space, with broad application prospects in smart homes, car audio systems, and open-plan offices. Its basic principle is to generate a specific sound field by spatially filtering the speaker array signal, allowing listeners in the bright zone to receive clear target sound while suppressing sound pressure energy in the dark zone to a minimum, thereby achieving spatial sound field isolation.

[0003] Adaptive acoustic zone control methods can adjust control filters in real time according to dynamic changes in the sound field environment, ensuring a good listening experience for listeners even with environmental changes, thus becoming a research hotspot in recent years. However, existing adaptive acoustic zone control methods have significant shortcomings and defects. In existing adaptive acoustic zone control methods, the dimensionality of the control filters is usually very high, often reaching thousands or even tens of thousands of coefficients, resulting in high computational complexity when the adaptive algorithm updates the filters at each time sampling. This problem leads to slow convergence speed of existing methods, making it difficult to quickly adjust the control filters when the sound field environment changes abruptly. Consequently, this results in increased signal distortion in bright areas and a significant decrease in the contrast between bright and dark areas, severely affecting the listening experience.

[0004] Therefore, developing an adaptive acoustic zone control method based on low-rank decomposition is of great significance. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive acoustic zone control method based on low-rank decomposition to solve the problems existing in the prior art.

[0006] The technical solution adopted to achieve the purpose of this invention is as follows: an adaptive sound zone control method based on low-rank decomposition, applied to a sound zone control system. The sound zone control system includes a loudspeaker array, several microphones respectively disposed in the bright and dark zones, and a signal processing unit. The method includes the following steps:

[0007] S1) The microphone collects the sound pressure signals at each measurement point in the bright and dark areas respectively, and transmits the sound pressure signals to the signal processing unit;

[0008] S2) Based on acoustic theory and the layout of the sound zone control system, and based on the room impulse response between the loudspeaker array and each microphone, a mathematical model for measuring sound pressure in the bright and dark zones is established.

[0009] S3) Decompose the control filter used to drive the loudspeaker array into several shorter sub-filters, and substitute the low-rank decomposition form into the measurement sound pressure model in step S2) to obtain the measurement sound pressure expression with the sub-filters as variables.

[0010] S4) Construct an adaptive sound zone control model: Using the sub-filters obtained in step S3) as variables, construct an adaptive sound zone control mathematical model with the objectives of minimizing the sound pressure error in the bright zone and minimizing the sound pressure energy in the dark zone, which includes a forgetting factor and regularization constraints.

[0011] S5) The signal processing unit, based on the sound pressure signal collected by the microphone in step S1), and by minimizing the mathematical model constructed in step S4), recursively solves for each optimal sub-filter.

[0012] S6) The signal processing unit integrates the optimal sub-filters obtained in step S5) into a final control filter according to the low-rank decomposition structure, and outputs it to the loudspeaker array to drive the loudspeaker array to implement sound field control in the bright and dark areas.

[0013] Furthermore, the loudspeaker array is a linear array. The number of microphones is the same in both the bright and dark areas. The loudspeaker array and the microphones are arranged on the same horizontal plane, and the geometric centers of the bright and dark areas are symmetrically distributed about the loudspeaker array.

[0014] Furthermore, in step S2), the measured sound pressure at the m-th microphone in the bright or dark area is modeled as follows:

[0015]

[0016] This represents the sound pressure level collected at the m-th microphone within the bright / dark area at the n-th time sampling point. For the input signal sampled at the nth time, T For transpose, K and J are the room impulse response length and the control filter length of a single channel, respectively. It is a matrix composed of the room impulse responses between the m-th microphone in the bright / dark zone and each loudspeaker. For length is The control filter vector, This represents the number of channels.

[0017] Furthermore, in step S3), the control filter of the original length JL is represented as a combination of short sub-filters using the principle of the nearest Kronecker integral solution, so as to reduce the computational complexity of adaptive update.

[0018] Perform low-rank decomposition on the control filter:

[0019]

[0020] in, , , and They are respectively of length , and Sub-filters, For low-rank estimation parameters, This is the Kronecker product. Based on... ,get:

[0021]

[0022] in, , and For dimension , and The identity matrix, , , , , , , , , , , , , , .

[0023] Furthermore, in step S4), the adaptive acoustic zone control mathematical model is based on the forgetting factor. The weighted recursive least squares form has an objective function comprising: the weighted cumulative sum of squares of the differences between the measured sound pressure and the desired sound pressure at each microphone in the bright area; the weighted cumulative sum of squares of the measured sound pressure energy at each microphone in the dark area; and a regularization penalty term for the control filter. The desired sound pressure in the bright area is obtained by convolving the desired room impulse response with the input signal.

[0024] Furthermore, in step S4), the mathematical model for adaptive sound zone control is as follows:

[0025]

[0026] in, and This refers to the number of microphones in the bright and dark areas. Forgetting factor, and For regularization parameters, Let m be the desired sound pressure level at the m-th microphone within the bright area. It is a matrix composed of the desired room impulse response.

[0027] Substituting the microphone measurement signals from the bright / dark areas, we obtain a mathematical model with three sub-filters as variables:

[0028] in, , , , , , , , , , , , , , , , yes The element rearrangement form.

[0029] Furthermore, in step S5), the recursive solution formulas for the three optimal sub-filters are as follows:

[0030]

[0031] in, , , , , , .

[0032] Furthermore, in step S6), the three types of sub-filters are integrated into the final control filter according to the following structure:

[0033]

[0034] in, yes The j2th block of length J2 in the middle, yes The Middle A length of J 12 The block, yes The Middle A length of J 11 L's block.

[0035] Furthermore, the room impulse response matrix is ​​obtained through actual measurement or simulation using the mirror source method. When the sound field environment undergoes abrupt changes, a forgetting factor is used... By controlling the weight decay of historical data, rapid tracking and adaptive adjustment of sound field changes can be achieved.

[0036] Furthermore, the sound zone control system also includes a virtual sound source. The virtual sound source is used to generate the desired sound field signal. The desired sound pressure level in the bright zone is obtained by convolving the output signal of the virtual sound source with the desired room impulse response, and serves as a reference signal for calculating the sound pressure level error in the bright zone.

[0037] The technical effects of this invention are beyond doubt:

[0038] A. By performing low-rank decomposition on the control filter, the high-dimensional control filter vector is decomposed into three types of shorter sub-filters, thus transforming the high-dimensional adaptive update problem into several low-dimensional sub-optimization problems. Since the dimension of the sub-filters is much lower than that of the original control filter, the matrix operation scale required by the adaptive algorithm at each time sampling is greatly reduced, significantly reducing the computational load and substantially improving the adaptive update speed of the control filter;

[0039] B. In the event of abrupt changes in the sound field environment, it can quickly complete the recursive update of sub-filters and integrate the updated sub-filters into the final control filter, driving the speaker array to respond rapidly to changes in the sound field. After abrupt changes in the sound zone position, it can restore the sound contrast and signal distortion in the bright zone to a good level in a shorter time, and the adaptive adjustment response speed is significantly improved;

[0040] C. While achieving rapid adaptive adjustment, it maintains effective control over the sound pressure error in the bright zone and the sound pressure energy in the dark zone. The sound zone control performance is not significantly degraded due to low-rank decomposition, thus balancing computational efficiency and control accuracy. Attached Figure Description

[0041] Figure 1 Layout of the sound zone control system;

[0042] Figure 2 The results of the control of sound contrast and signal distortion in the bright zone are shown in the example when the sound zone position changes abruptly at 3.5 seconds after the white noise signal is played. Detailed Implementation

[0043] The present invention will be further described below with reference to embodiments, but it should not be construed that the scope of the present invention is limited to the following embodiments. Various substitutions and modifications made based on ordinary technical knowledge and common practices in the art without departing from the above-described technical concept of the present invention should be included within the scope of protection of the present invention.

[0044] Example 1:

[0045] See Figure 1 This embodiment provides an adaptive acoustic zone control method based on low-rank decomposition, which is applied to an acoustic zone control system. The acoustic zone control system includes a loudspeaker array, a plurality of microphones respectively disposed in the bright and dark zones, and a signal processing unit. The method includes the following steps:

[0046] S1) The microphone collects the sound pressure signals at each measurement point in the bright and dark areas respectively, and transmits the sound pressure signals to the signal processing unit;

[0047] S2) Based on acoustic theory and the layout of the sound zone control system, and based on the room impulse response between the loudspeaker array and each microphone, a mathematical model for measuring sound pressure in the bright and dark zones is established.

[0048] S3) Decompose the control filter used to drive the loudspeaker array into several shorter sub-filters, and substitute the low-rank decomposition form into the measurement sound pressure model in step S2) to obtain the measurement sound pressure expression with the sub-filters as variables.

[0049] S4) Construct an adaptive sound zone control model: Using the sub-filters obtained in step S3) as variables, construct an adaptive sound zone control mathematical model with the objectives of minimizing the sound pressure error in the bright zone and minimizing the sound pressure energy in the dark zone, which includes a forgetting factor and regularization constraints.

[0050] S5) The signal processing unit, based on the sound pressure signal collected by the microphone in step S1), and by minimizing the mathematical model constructed in step S4), recursively solves for each optimal sub-filter.

[0051] S6) The signal processing unit integrates the optimal sub-filters obtained in step S5) into a final control filter according to the low-rank decomposition structure, and outputs it to the loudspeaker array to drive the loudspeaker array to implement sound field control in the bright and dark areas.

[0052] Example 2:

[0053] The main content of this embodiment is the same as that of Embodiment 1, except that the loudspeaker array is a linear array. The number of microphones in the bright area and the dark area is the same. The loudspeaker array and the microphones are arranged on the same horizontal plane, and the geometric centers of the bright area and the dark area are symmetrically distributed about the loudspeaker array.

[0054] Example 3:

[0055] The main content of this embodiment is the same as that of embodiment 1 or 2, wherein, in step S2), the measured sound pressure at the m-th microphone in the bright or dark area is modeled as follows:

[0056]

[0057] Let m be the sound pressure level collected at the m-th microphone in the bright / dark area at the n-th time sampling point (the number of microphones is the same in both the bright and dark areas). For the input signal sampled at the nth time, T For transpose, K and J are the room impulse response length and the control filter length of a single channel, respectively. It is a matrix composed of the room impulse responses between the m-th microphone in the bright / dark zone and each loudspeaker. For length is The control filter vector, This represents the number of channels.

[0058] Example 4:

[0059] The main content of this embodiment is the same as any one of embodiments 1 to 3. In step S3), the control filter of the original length JL is represented as a combination of short sub-filters by using the nearest Kronecker integral solution principle, so as to reduce the computational complexity of adaptive update.

[0060] Perform low-rank decomposition on the control filter:

[0061]

[0062] in, , , and They are respectively of length , and Sub-filters, For low-rank estimation parameters, This is the Kronecker product. Based on... ,get:

[0063]

[0064] in, , and For dimension , and The identity matrix, , , , , , , , , , , , , , .

[0065] Example 5:

[0066] The main content of this embodiment is the same as any one of embodiments 1 to 4, wherein, in step S4), the adaptive sound zone control mathematical model is based on the forgetting factor. The weighted recursive least squares form has an objective function comprising: the weighted cumulative sum of squares of the differences between the measured sound pressure and the desired sound pressure at each microphone in the bright area; the weighted cumulative sum of squares of the measured sound pressure energy at each microphone in the dark area; and a regularization penalty term for the control filter. The desired sound pressure in the bright area is obtained by convolving the desired room impulse response with the input signal.

[0067] The mathematical model for adaptive sound zone control is as follows:

[0068]

[0069] in, and This refers to the number of microphones in the bright and dark areas. Forgetting factor, and For regularization parameters, Let m be the desired sound pressure level at the m-th microphone within the bright area. It is a matrix composed of the desired room impulse response.

[0070] Substituting the microphone measurement signals from the bright / dark areas, we obtain a mathematical model with three sub-filters as variables:

[0071] in, , , , , , , , , , , , , , , , yes The element rearrangement form.

[0072] Example 6:

[0073] The main content of this embodiment is the same as any one of embodiments 1 to 5, wherein, in step S5), the recursive solution formulas for the three optimal sub-filters are as follows:

[0074]

[0075] in, , , , , , .

[0076] Example 7:

[0077] The main content of this embodiment is the same as any one of embodiments 1 to 6, wherein, in step S6), the three types of sub-filters are integrated into the final control filter according to the following structure:

[0078]

[0079] in, yes The j2th block of length J2 in the middle, yes The Middle A length of J 12 The block, yes The Middle A length of J 11 L's block.

[0080] Example 8:

[0081] The main content of this embodiment is the same as any one of embodiments 1 to 7, wherein the room impulse response matrix is ​​obtained through actual measurement or simulation using the mirror source method. When the sound field environment undergoes abrupt changes, a forgetting factor is used. By controlling the weight decay of historical data, rapid tracking and adaptive adjustment of sound field changes can be achieved.

[0082] Example 9:

[0083] The main content of this embodiment is the same as any one of embodiments 1 to 8, wherein the sound zone control system further includes a virtual sound source. The virtual sound source is used to generate the desired sound field signal. The desired sound pressure in the bright zone is obtained by convolving the output signal of the virtual sound source with the desired room impulse response, and is used as a reference signal for calculating the sound pressure error in the bright zone.

[0084] Example 10:

[0085] The main content of this embodiment is the same as any one of embodiments 1 to 9, wherein the present invention includes the following steps:

[0086] Step 1: Construct a sound pressure measurement system within the sound range and perform low-rank decomposition on the control filter.

[0087] Step 101: Construct a sound zone for measuring sound pressure.

[0088] Based on acoustic theory and sound zone control system layout, model the sound pressure measurement of microphones in bright / dark zones:

[0089]

[0090] Let m be the sound pressure level collected at the m-th microphone in the bright / dark area at the n-th time sampling point (the number of microphones is the same in both the bright and dark areas). For the input signal sampled at the nth time, T "This is the transpose, where K and J are the room impulse response length and the control filter length for a single channel, respectively." It is a matrix composed of the room impulse responses between the m-th microphone in the bright / dark zone and each loudspeaker. For length is The control filter vector, This represents the number of channels.

[0091] Step 102: Perform low-rank decomposition on the control filter.

[0092]

[0093] in, , , and They are respectively of length , and Sub-filters, For low-rank estimation parameters, "For the Kronecker product. Based on..." ,get:

[0094]

[0095] in, , and For dimension , and The identity matrix, , , , , , , , , , , , , , .

[0096] Substituting the low-rank decomposition form of the filter back into the sound pressure measurement model, we get:

[0097]

[0098] in, yes The form of rearranging the elements:

[0099]

[0100] , for The first in List.

[0101] Step 2: Construct an adaptive sound zone control model

[0102] The following adaptive sound zone control mathematical model is constructed:

[0103]

[0104] in, and This refers to the number of microphones in the bright and dark areas. Forgetting factor, and For regularization parameters, Let m be the desired sound pressure level at the m-th microphone within the bright area. It is a matrix composed of the desired room impulse response.

[0105] Substituting the microphone measurement signals from the bright / dark areas, we obtain a mathematical model with three sub-filters as variables:

[0106]

[0107] in, , , , , , , , , , , , , , , .

[0108] Step 3: Solve for the sub-filters and integrate them into the final control filter.

[0109] Step 301: Solve for the sub-filter

[0110] By minimizing the mathematical model established in step 2 with the sub-filter as the variable, the optimal sub-filter can be obtained as follows:

[0111]

[0112] in, , , , , , .

[0113] Step 302: Integrate the final control filter

[0114] Integrate the sub-filters into the final control filter:

[0115]

[0116] in, yes The j2th block of length J2 in the middle, yes The Middle A length of J 12 The block, yes The Middle A length of J 11 L's block.

[0117] Example 11:

[0118] The main content of this embodiment is the same as any one of embodiments 1 to 10. However, to verify the accuracy of the invention, an adaptive sound zone control simulation is performed. The specific process is as follows:

[0119] 1. Determine the positions of the loudspeaker array and microphones in the bright and dark sound zones, and determine the room impulse response;

[0120] 2. Based on step 1, model the microphone, measure the sound pressure, and perform low-rank decomposition on the control filter;

[0121] 3. Establish an adaptive acoustic zone control model based on step 2;

[0122] 4. Calculate the optimal sub-filter based on step 3 and integrate them into the final control filter.

[0123] The simulation settings are as follows: The layout of the sound zone control system is as follows. Figure 1 As shown, a linear loudspeaker array with 16 elements is placed in a size of In the room, the plane at a height of z = 1.5 m. The bright area, dark area, and speaker array are on the same plane, with their geometric centers located at... and The interior contains 37 microphones evenly distributed. The distance between each microphone and speaker is 0.09 m. The virtual sound source is located at... This is used to simulate the desired sound field. The room impulse response is simulated using the Mirror Source Toolbox.

[0124] Simulation results:

[0125] Figure 2 The results demonstrate the control of acoustic contrast and signal distortion in the bright zone when the sound zone position changes abruptly after 3.5 seconds of playing a white noise signal. As shown in the figure, compared with the adaptive sound zone control method without low-rank decomposition, this invention can quickly adjust the acoustic contrast and signal distortion in the bright zone back to a good level after a sudden change in the sound field environment.

Claims

1. An adaptive acoustic zone control method based on low-rank decomposition, characterized in that, The method is applied to a sound zone control system; the sound zone control system includes a loudspeaker array, a plurality of microphones respectively disposed in the bright zone and the dark zone, and a signal processing unit; the method includes the following steps: S1) The microphone collects the sound pressure signals at each measurement point in the bright and dark areas respectively, and transmits the sound pressure signals to the signal processing unit; S2) Based on acoustic theory and the layout of the sound zone control system, and based on the room impulse response between the loudspeaker array and each microphone, a mathematical model for measuring sound pressure in the bright and dark zones is established. S3) Decompose the control filter used to drive the loudspeaker array into several shorter sub-filters, and substitute the low-rank decomposition form into the measurement sound pressure model in step S2) to obtain the measurement sound pressure expression with the sub-filters as variables. S4) Construct an adaptive sound zone control model: Using the sub-filters obtained in step S3) as variables, construct an adaptive sound zone control mathematical model with the objectives of minimizing the sound pressure error in the bright zone and minimizing the sound pressure energy in the dark zone, which includes a forgetting factor and regularization constraints. S5) The signal processing unit, based on the sound pressure signal collected by the microphone in step S1), and by minimizing the mathematical model constructed in step S4), recursively solves for each optimal sub-filter. S6) The signal processing unit integrates the optimal sub-filters obtained in step S5) into a final control filter according to the low-rank decomposition structure and outputs it to the loudspeaker array; drives the loudspeaker array to implement sound field control in the bright and dark areas.

2. The adaptive acoustic zone control method based on low-rank decomposition according to claim 1, characterized in that: The loudspeaker array is a linear array; the number of microphones in the bright and dark areas is the same; the loudspeaker array and microphones are arranged on the same horizontal plane, and the geometric centers of the bright and dark areas are symmetrically distributed about the loudspeaker array.

3. The adaptive acoustic zone control method based on low-rank decomposition according to claim 1, characterized in that, In step S2), the measured sound pressure at the m-th microphone in the bright or dark area is modeled as follows: This represents the sound pressure level collected at the m-th microphone within the bright / dark area at the n-th time sampling point. For the input signal sampled at the nth time, T For transpose, K and J are the room impulse response length and the control filter length of a single channel, respectively. It is a matrix composed of the room impulse responses between the m-th microphone in the bright / dark zone and each loudspeaker. For length is The control filter vector, This represents the number of channels.

4. The adaptive acoustic zone control method based on low-rank decomposition according to claim 1, characterized in that: In step S3), the control filter of length JL is represented as a combination of short sub-filters using the principle of the nearest Kronecker integral solution, so as to reduce the computational complexity of adaptive update. Perform low-rank decomposition on the control filter: in, , , and They are respectively of length , and Sub-filters, For low-rank estimation parameters, For the Kronecker product; based on ,get: in, , and For dimension , and The identity matrix, , , , , , , , , , , , , , .

5. The adaptive acoustic zone control method based on low-rank decomposition according to claim 1, characterized in that: In step S4), the adaptive sound zone control mathematical model is based on the forgetting factor. The weighted recursive least squares form has an objective function that includes: the weighted cumulative sum of squares of the difference between the measured sound pressure and the desired sound pressure at each microphone in the bright area, the weighted cumulative sum of squares of the measured sound pressure energy at each microphone in the dark area, and a regularization penalty term for the control filter; wherein, the desired sound pressure in the bright area is obtained by convolving the desired room impulse response with the input signal.

6. The adaptive acoustic zone control method based on low-rank decomposition according to claim 5, characterized in that, In step S4), the mathematical model for adaptive sound zone control is as follows: in, and This refers to the number of microphones in the bright and dark areas. Forgetting factor, and For regularization parameters, Let m be the desired sound pressure level at the m-th microphone within the bright area. It is a matrix composed of the desired room impulse response; Substituting the microphone measurement signals from the bright / dark areas, we obtain a mathematical model with three sub-filters as variables: in, , , , , , , , , , , , , , , , yes The element rearrangement form.

7. The adaptive acoustic zone control method based on low-rank decomposition according to claim 1, characterized in that, In step S5), the recursive solution formulas for the three optimal sub-filters are as follows: in, , , , , , .

8. The adaptive acoustic zone control method based on low-rank decomposition according to claim 7, characterized in that, In step S6), the three types of sub-filters are integrated into the final control filter according to the following structure: in, yes The j2th block of length J2 in the middle, yes The Middle A length of J 12 The block, yes The Middle A length of J 11 L's block.

9. The adaptive acoustic zone control method based on low-rank decomposition according to claim 1, characterized in that: The room impulse response matrix is ​​obtained through actual measurement or simulation using the mirror source method; when the sound field environment undergoes abrupt changes, a forgetting factor is used. By controlling the weight decay of historical data, rapid tracking and adaptive adjustment of sound field changes can be achieved.

10. The adaptive acoustic zone control method based on low-rank decomposition according to any one of claims 1 to 9, characterized in that: The sound zone control system also includes a virtual sound source; the virtual sound source is used to generate a desired sound field signal; the desired sound pressure in the bright zone is obtained by convolving the output signal of the virtual sound source with the desired room impulse response, and is used as a reference signal for calculating the sound pressure error in the bright zone.

Images