Method and apparatus for noise control

Abstract

The embodiment of the present application provides a noise control method and device in the field of signal processing, the method comprises the following steps: filtering a first reference signal by a control filter to obtain a control signal, the control signal is used for indicating a secondary noise signal; acquiring an error signal collected by an error sensor, the error signal is obtained by superimposing an original noise signal and the secondary noise signal; determining a second reference signal according to the first reference signal; updating P segment coefficients of the control filter based on at least one of P segment sub-signals of the error signal and the second reference signal. The scheme of the embodiment of the present application can reduce the calculation amount in the noise reduction process, reduce the demand for computing resources, and improve the efficiency of signal processing.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the field of signal processing, and more particularly, to a noise control method and device. BACKGROUND

[0002] With the continuous development of technology, users have higher and higher requirements for sound experience. Echo cancellation, noise reduction and other signal processing methods are important means to improve user experience. For example, road noise is generated during vehicle driving due to interaction with the ground, and the road noise enters the vehicle cabin through the suspension, vehicle body and other architectures. Through noise reduction processing in the car, the influence of noise can be reduced, and the driving experience of users can be improved. The noise reduction method based on Newton algorithm has obvious effect and fast tracking speed. However, the calculation amount of this method is large, especially in the case of multiple inputs and multiple outputs, the calculation amount of this method exceeds the calculation capacity of most real-time processors, and it is difficult to apply to scenarios with high real-time requirements. SUMMARY

[0003] Embodiments of the present application provide a noise control method and device, which improves the efficiency of signal processing and is beneficial to meet higher real-time requirements.

[0004] In a first aspect, a noise control method is provided, comprising: obtaining a first reference signal collected by a reference sensor; filtering the first reference signal through a control filter to obtain a control signal, the control signal being used to indicate a secondary noise signal; obtaining an error signal collected by an error sensor, the error signal being obtained by superimposing an original noise signal and the secondary noise signal; determining a second reference signal according to the first reference signal; updating P segment coefficients of the control filter based on at least one of P segment sub-signals of the second reference signal and the error signal, the P segment sub-signals being obtained based on sampling time of the second reference signal, the P segment coefficients being obtained based on length of the control filter, and P being a positive integer greater than 1.

[0005] According to the scheme of embodiments of the present application, the P segment coefficients of the control filter are updated respectively, that is, the P segment coefficients are updated independently, which is beneficial to reduce the calculation amount in the noise reduction process, reduce the demand for computing resources, and improve the efficiency of signal processing.

[0006] The P segment sub-signals of the second reference signal and the P segment coefficients of the controller can be one-to-one corresponding. The length of the P segment sub-signals of the second reference signal is the same as the length of the corresponding P segment coefficients.

[0007] In some implementations of the first aspect, the updating the P segments of coefficients of the control filter based on the error signal and at least one of the P segments of the second reference signal comprises: updating the P segments of coefficients of the control filter based on an inverse of an autocorrelation matrix of the at least one of the P segments of the second reference signal and P segments of a gradient of a mean square error of the error signal, or updating the P segments of coefficients of the control filter based on an inverse of an autocorrelation matrix of the at least one of the P segments of the first reference signal and P segments of a gradient of a mean square error of the error signal, the P segments of the first reference signal being segmented based on sampling instants of the first reference signal, and the P segments of the gradient of the mean square error of the error signal being determined according to the P segments of the second reference signal.

[0008] When the coefficients of the control filter are updated by using the LMS Newton method, an inverse operation of a matrix needs to be performed, and when the coefficients of the control filter are updated as a whole, an inverse of an autocorrelation matrix of the second reference signal needs to be calculated. According to the scheme of the embodiments of the present application, only the inverse of the autocorrelation matrix of at least one segment of the signal needs to be calculated, which reduces the dimension of the autocorrelation matrix involved in the calculation, thereby reducing the calculation complexity of the inverse operation of the matrix, reducing the calculation amount, reducing the demand for computing resources, improving the signal processing efficiency, and facilitating the real-time requirement in the noise reduction process.

[0009] In some implementations of the first aspect, the length of the P segments of coefficients is determined according to the correlation of the third reference signal, the third reference signal being determined according to a fourth reference signal collected by a reference sensor, and the sampling time period of the fourth reference signal is not later than the sampling time period of the first reference signal.

[0010] For example, the length of each segment of the P segments of coefficients is determined according to the correlation of the third reference signal.

[0011] Alternatively, the length of at least one segment of the P segments of coefficients is determined according to the correlation of the third reference signal.

[0012] The embodiments of the present application determine the length of the segmented coefficients of the control filter based on the correlation between signals, for example, determine the length of the segmented coefficients of the control filter according to the correlation of the filtered reference signal, and the correlation between different segments of the signal is weak, which can reduce the calculation amount while ensuring that the convergence speed and the noise reduction effect will not be affected. The scheme of the embodiments of the present application has similar convergence speed and noise reduction effect as the LMS Newton algorithm, and the calculation amount is much smaller than the LMS Newton algorithm, that is, the scheme of the embodiments of the present application can reduce the calculation amount while ensuring the noise reduction effect, reduce the demand for computing resources, and improve the signal processing efficiency.

[0013] With reference to the first aspect, in some implementations of the first aspect, the length of the P segments of coefficients is determined according to a correlation of the third reference signal, including: the length of the P segments of coefficients is determined according to a maximum value of a number of sampling points required for the autocorrelation coefficients and the cross-correlation coefficients of the multiple channels of the third reference signal to decay to a first threshold value.

[0014] For example, the length of the P segments of coefficients is greater than or equal to the maximum value of the number of sampling points required for the autocorrelation coefficients and the cross-correlation coefficients of the multiple channels of the third reference signal to decay to the first threshold value.

[0015] With reference to the first aspect, in some implementations of the first aspect, a time interval between a sampling period of the fourth reference signal and a sampling period of the first reference signal is less than or equal to a second threshold value.

[0016] According to the scheme of the embodiments of the present application, the length of each segment of coefficients of the control filter is dynamically adjusted according to the signal characteristics, so as to improve the adaptability of the method and facilitate to ensure the convergence speed of the update of the control filter, that is, to ensure a better noise reduction effect.

[0017] With reference to the first aspect, in some implementations of the first aspect, the P segments of coefficients of the control filter are updated based on at least one of the error signal and the P segments of sub-signals of the second reference signal, including: the P segments of coefficients are updated based on the error signal and the p-th segment of sub-signals of the second reference signal, p = 0, 1, 2, …, P-1.

[0018] In this way, the P segments of coefficients of the control filter can be updated based on the error signal and the P segments of sub-signals of the second reference signal respectively.

[0019] With reference to the first aspect, in some implementations of the first aspect, the P segments of coefficients of the control filter are updated based on at least one of the error signal and the P segments of sub-signals of the second reference signal, including: the P segments of coefficients are updated based on any segment of sub-signals of the error signal and the second reference signal respectively.

[0020] According to the scheme of the embodiments of the present application, the P segments of coefficients of the control filter are updated using the autocorrelation matrix of the same segment of sub-signals, which can further reduce the amount of calculation, reduce the demand for computing capacity, and improve the signal processing efficiency.

[0021] With reference to the first aspect, in some implementations of the first aspect, the second reference signal is obtained by filtering the first reference signal through a secondary path transfer function, and the third reference signal is obtained by filtering the fourth reference signal through the secondary path transfer function.

[0022] With reference to the first aspect, in some implementations of the first aspect, the second reference signal is obtained by delaying the first reference signal by J sampling instants, J being a positive integer, J being determined according to a length of the secondary path transfer function, and updating the P taps coefficients of the control filter based on at least one of the P taps sub-signals of the error signal and the second reference signal comprises: filtering the error signal by an inverse secondary path transfer function to obtain a filtered error signal; and updating the P taps coefficients of the control filter based on at least one of the P taps sub-signals of the filtered error signal and the second reference signal.

[0023] According to the scheme of the embodiments of the present application, the P taps coefficients of the control filter are updated respectively, i.e., the P taps coefficients are updated independently, which is beneficial to reduce the calculation amount in the signal processing process, reduce the demand for computing resources, and improve the efficiency of the signal processing.

[0024] According to the scheme of the embodiments of the present application, the P taps coefficients of the control filter are updated respectively, i.e., the P taps coefficients are updated independently, which is beneficial to reduce the calculation amount in the signal processing process, reduce the demand for computing resources, and improve the efficiency of the signal processing.

[0025] With reference to the second aspect, in some implementations of the second aspect, updating the P taps coefficients of the control filter based on at least one of the P taps sub-signals of the error signal and the second reference signal comprises: updating the P taps coefficients of the control filter based on at least one of a P taps of a gradient of a mean square error of the error signal and a P taps of an inverse matrix of an autocorrelation matrix of the P taps sub-signals of the second reference signal, or updating the P taps coefficients of the control filter based on at least one of a P taps of a gradient of a mean square error of the error signal and a P taps of an inverse matrix of an autocorrelation matrix of the P taps sub-signals of the first reference signal, the P taps sub-signals of the first reference signal being obtained by segmenting the first reference signal based on sampling instants of the first reference signal, and the P taps of the gradient of the mean square error of the error signal being determined according to the P taps sub-signals of the second reference signal.

[0026] When the LMS Newton method is used to update the coefficients of the control filter, the inverse operation of a matrix needs to be performed, and the inverse matrix of the autocorrelation matrix of the second reference signal needs to be calculated for updating the coefficients of the control filter as a whole. When the scheme of the embodiments of the present application is used, only the inverse matrix of the autocorrelation matrix of at least one sub-signal needs to be calculated, the dimension of the autocorrelation matrix participating in the calculation is reduced, the calculation complexity of the inverse operation of the matrix is reduced, the calculation amount is reduced, the demand for computing resources is reduced, the signal processing efficiency is improved, and the real-time requirement in the signal processing process is met.

[0027] With reference to the second aspect, in some implementations of the second aspect, a length of the P coefficients is determined according to a correlation of a third reference signal, the third reference signal is determined according to a fourth reference signal, the fourth reference signal and the first reference signal are from a same signal source, and a sampling time period of the fourth reference signal is not later than a sampling time period of the first reference signal.

[0028] With reference to the second aspect, in some implementations of the second aspect, a length of the P coefficients is determined according to a correlation of a third reference signal, including that the length of the P coefficients is determined according to a maximum value in a number of sampling points required for autocorrelation coefficients and cross-correlation coefficients of a plurality of channels of the third reference signal to decay to a first threshold value.

[0029] With reference to the second aspect, in some implementations of the second aspect, a time interval between the sampling time period of the fourth reference signal and the sampling time period of the first reference signal is less than or equal to a second threshold value.

[0030] With reference to the second aspect, in some implementations of the second aspect, updating the P coefficients of the control filter based on at least one of the P sub-signals of the second reference signal and the error signal includes updating a p-th coefficient based on a p-th sub-signal of the second reference signal and the error signal, p = 0, 1, 2, …, P-1.

[0031] With reference to the second aspect, in some implementations of the second aspect, updating the P coefficients of the control filter based on at least one of the P sub-signals of the second reference signal and the error signal includes updating the P coefficients based on any sub-signal of the second reference signal and the error signal, respectively.

[0032] In a third aspect, a device for noise control is provided, which includes units for performing the method of the first aspect and any implementation of the first aspect.

[0033] In a fourth aspect, a device for signal processing is provided, which comprises units for performing the method in the second aspect and any of the implementation manners of the second aspect.

[0034] It should be understood that the expansions, limitations, explanations and descriptions of the related content in the first aspect above also apply to the same content in the second aspect, the third aspect and the fourth aspect.

[0035] In a fifth aspect, a device for noise control is provided, which comprises a memory for storing a program, and a processor for executing the program stored in the memory, wherein when the program stored in the memory is executed, the processor is configured to perform the method in the first aspect and any of the implementation manners of the first aspect.

[0036] Exemplarily, the device can be a vehicle-mounted device, a vehicle-mounted chip or the like.

[0037] In a sixth aspect, a device for signal processing is provided, which comprises a memory for storing a program, and a processor for executing the program stored in the memory, wherein when the program stored in the memory is executed, the processor is configured to perform the method in the second aspect and any of the implementation manners of the second aspect.

[0038] In a seventh aspect, a computer readable medium is provided, which stores program codes for execution by a device, and the program codes comprise codes for performing the method in the first aspect or any of the implementation manners of the second aspect.

[0039] In an eighth aspect, a computer program product containing instructions is provided, which, when executed on a computer, causes the computer to perform the method in the first aspect or any of the implementation manners of the second aspect.

[0040] In a ninth aspect, a vehicle is provided, which comprises the device in any of the third aspect or the fifth aspect.

[0041] In a tenth aspect, a chip system is provided, which comprises a processor for invoking a computer program or computer instructions stored in a memory, so as to cause the processor to perform the method in any of the aspects above.

[0042] In combination with the tenth aspect, in a possible implementation manner, the processor is coupled with the memory through an interface.

[0043] In combination with the tenth aspect, in a possible implementation manner, the chip system further comprises the memory, and the memory stores the computer program or the computer instructions. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1A structural schematic diagram of a system architecture of the present application.

[0045] Figure 2 A schematic diagram of a noise control device of an embodiment of the present application.

[0046] Figure 3 A schematic diagram of a noise control method of an embodiment of the present application.

[0047] Figure 4 A schematic diagram of another noise control method of an embodiment of the present application.

[0048] Figure 5 A schematic flow chart of a method for controlling coefficient update of a filter of an embodiment of the present application.

[0049] Figure 6 A schematic diagram of a change trend of a correlation coefficient of an embodiment of the present application.

[0050] Figure 7 A schematic diagram of yet another noise control method of an embodiment of the present application.

[0051] Figure 8 A schematic diagram of simulation comparison of two schemes of an embodiment of the present application.

[0052] Figure 9 A schematic flow chart of another method for controlling coefficient update of a filter of an embodiment of the present application.

[0053] Figure 10 A schematic diagram of another noise control method of an embodiment of the present application.

[0054] Figure 11 A schematic flow chart of yet another method for controlling coefficient update of a filter of an embodiment of the present application.

[0055] Figure 12 A schematic diagram of a signal processing method of an embodiment of the present application.

[0056] Figure 13 A schematic flow chart of a noise control method of an embodiment of the present application.

[0057] Figure 14 A schematic flow chart of a signal processing method of an embodiment of the present application.

[0058] Figure 15 A schematic block diagram of a noise control device of an embodiment of the present application.

[0059] Figure 16 A schematic block diagram of a signal processing device of an embodiment of the present application.

[0060] Figure 17 An exemplary block diagram of another signal processing device of embodiments of the present application. DETAILED DESCRIPTION

[0061] The technical solutions in the embodiments of the present application will be described below with reference to the drawings.

[0062] The signal processing method of the embodiments of the present application can be applied to signal processing scenarios such as active noise control, echo cancellation, audio noise reduction, etc., to reduce the amount of calculation and improve the efficiency of signal processing.

[0063] The above scenarios can all use an adaptive control filter for signal processing. Generally speaking, the more the number of reference signals of the adaptive control filter, the better the effect of signal processing. However, with the increase in the number of signals, the convergence speed of the adaptive control filter will be affected, and the amount of calculation will also increase.

[0064] For the convenience of understanding and clear description, active noise control is taken as an example for description in the embodiments of the present application, which does not constitute a limitation on the solutions of the embodiments of the present application.

[0065] Active noise control can be applied to scenarios such as earphone noise reduction, vehicle interior noise reduction, and indoor noise reduction. Active noise control can also be referred to as active noise control, which uses the principle of wave interference to cancel noise signals, that is, by generating a signal with a polarity opposite to that of the external noise to achieve noise cancellation.

[0066] Figure 1 An exemplary block diagram of a noise control method provided by an embodiment of the present application is shown.

[0067] As shown in Figure 1 , N r channel reference signals x1, x2, …, x Nr are collected by a reference sensor. The reference signals are filtered and processed by a control filter in a control device to obtain N s channel control signals. N s channel control signals y1-y Ns After passing through a secondary sound source (for example, a secondary loudspeaker), an inverted noise, that is, a secondary noise signal, is emitted. The secondary noise signal and the original N e channel noise signal p1-p Ne in the target area are superimposed and canceled to obtain N e channel error signals e1-e Ne , that is, residual signals. Wherein, N r is a positive integer, N s is a positive integer, and N eis a positive integer. The error signal is fed back to the control device after being collected by the error sensor in the target area. The control device continuously adjusts the coefficients of the control filter, thereby changing the output of the control filter, so that the generated secondary noise cancels out the primary noise. For example, the control device adjusts the output of the control filter with the goal of reducing the mean square error between the secondary noise signal and the original noise signal. The original noise signal can also be referred to as the primary noise signal.

[0068] Currently, active noise control is mostly applied in scenarios with a small number of signal channels. For example, in the earphone noise reduction scenario, single-channel or double-channel reference signals and error signals can be used to implement active noise control. With an increase in the number of channels of the reference signals and the error signals, the amount of calculation in the signal processing process increases linearly. In actual engineering applications, in order to expand the noise reduction area and improve the control effect of the wideband noise, a large number of reference signals, error signals, and secondary sound sources need to be used. This will cause the amount of calculation of the active noise control algorithm to exceed the calculation capability of most real-time processors, making it difficult to be applied in scenarios with high real-time performance.

[0069] Taking automobile road noise as an example, in the driving process of the automobile, the automobile road noise is one of the main noises in the vehicle. The automobile road noise is generated by the interaction between the road surface and the tire. Since the automobile road noise is excited by four wheels, the transmission path is complex, and multiple channels of reference signals need to be used as input. Moreover, due to the influence of the automobile body structure and the like, the sound field in the vehicle is relatively complex, and multiple channels of secondary sound sources are used as output, which can further improve the noise reduction effect, thereby greatly increasing the amount of calculation.

[0070] The scheme of the embodiments of the present application provides a noise control method, which reduces the amount of calculation of signal processing and improves the efficiency of signal processing to meet the requirement of high real-time performance by segmenting the update of the coefficients of the control filter.

[0071] Figure 2 is a schematic structural diagram of a noise control device 100 provided by the embodiments of the present application. As shown in Figure 2 the device 100 can include a sensor 110 and a processor 120. The sensor 110 can include a reference sensor 111 and an error sensor 112. The reference sensor 111 is configured to acquire a reference signal and send the reference signal to the processor. The error sensor 112 is configured to acquire an error signal and send the error signal to the processor. The processor 120 is configured to process the reference signal and the error signal, and send the processed signal to a secondary sound source (for example, a loudspeaker).

[0072] Exemplarily, the reference sensor 111 can be a reference acceleration sensor or a reference microphone.

[0073] Exemplarily, the error sensor 112 can be an error microphone.

[0074] For example, the reference sensor can be used to collect acceleration signals. The error microphone can be arranged near the human ear, for example, the error microphone can be arranged at the headrest or the roof above the user's head. The secondary sound source can include a headrest loudspeaker, so that the transmission path from the headrest loudspeaker to the human ear can be regarded as a secondary path.

[0075] It should be understood that the above device 100 can be applied to an active noise control processing scene, in which case the device can also be referred to as a noise control device. Exemplarily, the above device 100 can be a device in an active noise reduction earphone, a vehicle noise active control system, a machine noise active control system, an active sound barrier, or a building opening active control system. For example, the device 100 is a device in a vehicle noise active control system, which can be arranged in an active noise reduction headrest.

[0076] Optionally, the above vehicle can include one or more different types of vehicles, and can also include one or more different types of transportation tools or movable objects operating or moving on land (for example, highways, roads, railways, etc.), water surface (for example: waterways, rivers, oceans, etc.) or space. For example, the vehicle can include a car, a bicycle, a motorcycle, a train, a subway, an airplane, a ship, a spacecraft, a robot or other types of transportation tools or movable objects, etc., and the embodiments of the present application are not limited thereto.

[0077] It should also be understood that Figure 2 The structure of the device 100 should not be understood as a limitation of the embodiments of the present application.

[0078] The process of the above processor for signal processing will be described in detail below with reference to the accompanying drawings.

[0079] Figure 3 A schematic structural diagram of a noise control method provided by an embodiment of the present application is shown. As Figure 3 shown, the reference signal x(k) collected by the reference sensor is processed by the primary path transfer function P(z) to obtain the original noise signal p(k) in the target area. The reference signal x(k) is filtered by the control filter W(z) to obtain the control signal y(k). The control signal y(k) is filtered by the secondary path transfer function S(z) to obtain the secondary noise signal s(k). The error signal e(k) can be obtained by superimposing the secondary noise signal and the original noise signal. The error signal e(k) is collected by the error sensor in the target area. The reference signal x(k) is filtered by the secondary path transfer function model the filtered reference signal secondary path transfer function model may be understood as an estimate of the secondary path transfer function. The secondary path transfer function model may be obtained by modeling. The processor may update the coefficients of the control filter based on the error signal and the filtered reference signal using a least mean square (LMS) algorithm. Specifically, the processor may update the coefficients of the control filter using an LMS-Newton method. Illustratively, the coefficients of the control filter W(z) are updated based on the error signal e(k), the filtered reference signal and the inverse of the autocorrelation matrix of the filtered reference signal . For example, a product operation is performed on the inverse of the autocorrelation matrix of the filtered reference signal and the filtered reference signal, and the coefficients of the control filter W(z) are updated based on the result of the product operation and the error signal.

[0080] The primary path refers to the transfer path between the reference signal and the original noise signal. The secondary path refers to the transfer path between the control signal output by the control filter and the error signal. The control signal y(k) is filtered by the secondary path transfer function to obtain the secondary noise signal, which can also be understood as that the control signal y(k) can control the secondary sound source to generate the secondary noise signal. k represents the kth moment. The kth moment can also be understood as the kth sampling point, or the sampling moment of the kth sampling point.

[0081] As shown in FIG. 1, Figure 3 the reference signal x(k) collected by the reference sensor can include N r reference signals x1(k), x2(k)...x Nr (k), where x1(k) represents the reference signal of the first channel at the kth moment, x2(k) represents the reference signal of the second channel at the kth moment, and so on, and x Nr (k) represents the reference signal of the N r channel at the kth moment. Wherein N r is a positive integer.

[0082] It should be noted that Figure 3 the method shown in FIG. 1 is only an example and does not limit the scheme of the embodiments of the present application. For example, Figure 3 the original noise signal y(k), the control signal y(k) or the error signal e(k) and the like in FIG. 1 can be one channel or multiple channels, Figure 3 the secondary sound source in the method shown in FIG. 1 can be one or more, and the error sensor can be one or more, which are not all shown in the figure.

[0083] The coefficients W of the control filter W(z) satisfy the following formula:

[0084]

[0085] wherein L w denotes the length of the control filter W(z). w 1,1 (0), w 1,1 (1),..., w 1,1 (L w -1) respectively denote the L w coefficients of the control filter corresponding to the first secondary sound source for the reference signal of the first channel. respectively denote the L s coefficients of the control filter corresponding to the N w th secondary sound source for the reference signal of the first channel. w 2,1 (0),..., w 2,1 (L w -1) respectively denote the L w coefficients of the control filter corresponding to the first secondary sound source for the reference signal of the second channel, and so on. For the sake of understanding and clear description, the control filter W(z) can be regarded as N r *N s control filters. The length of each control filter is L w , i.e. the coefficients of each control filter are L w . L w is a positive integer greater than 1, and N s is a positive integer.

[0086] The filtered reference signal satisfies the following formula:

[0087]

[0088] wherein, denotes the reference signal of the first channel at time k, which is filtered by the secondary path transfer function between the first secondary sound source and the first error microphone. denotes the reference signal of the first channel at time k-1, which is filtered by the secondary path transfer function between the first secondary sound source and the first error microphone. Similarly, denotes the reference signal of the N w th channel at time k-L r +1, which is filtered by the secondary path transfer function between the N s th secondary sound source and the N e th error sensor.

[0089] Figure 4 A schematic structural diagram of another noise control method provided in an embodiment of this application is shown. Figure 4 and Figure 3 The difference between the noise control methods shown lies in the way the coefficients of the control filter W(z) are updated. To avoid repetition, in the description... Figure 4 Sometimes, certain parts of the description may be omitted. For example... Figure 4 As shown, the reference signal x(k) acquired by the reference sensor is processed by the primary path transfer function P(z) to obtain the original noise signal p(k) of the target area. The control filter W(z) filters the reference signal x(k) to obtain the control signal y(k). The control signal y(k) is filtered by the secondary path transfer function S(z) to obtain the secondary noise signal s(k). The error signal e(k) can be obtained by superimposing the secondary noise signal and the original noise signal. The error signal e(k) is acquired by the error sensor in the target area. The reference signal x(k) is delayed by J sampling times to obtain the delayed reference signal x(kJ). The error signal e(k) is filtered by the time-reversed secondary path transfer function to obtain the filtered error signal. The processor can use the adaptive LMS algorithm to update the coefficients of the control filter based on the filtered error signal and the delayed reference signal x(kJ). Specifically, the coefficients of the control filter can be updated using Newton's method. For example, as shown... Figure 4 As shown, the coefficients of the control filter W(z) are updated based on the inverse matrix of the autocorrelation matrix of the filtered error signal and the delayed reference signal, and the inverse matrix R of the autocorrelation matrix of the delayed reference signal x(kJ). -1 The coefficients of the control filter W(z) are updated based on the product operation and the result of the product operation and the filtered error signal.

[0090] exist Figure 3 and Figure 4 In the noise control method shown, the update of the coefficients of the control filter W(z) can satisfy the following formula:

[0091]

[0092] Among them, W new The coefficients of the updated control filter W(z) are given by W. old The coefficients of the control filter W(z) before the update are given, and μ is the iteration step size.

[0093] exist Figure 3 In the noise control method shown, R -1 (k) can be the filtered reference signal. The inverse matrix of the autocorrelation matrix, This is the gradient of the mean square error of the error signal, or in other words, This represents the gradient of the mean square error performance surface of the error signal. In the embodiments of this application, the mean square error performance surface can also be simply referred to as the performance surface. Figure 3 In the noise control method shown, when the coefficients of the control filter are updated in real time, The following formula can be approximated:

[0094]

[0095] When the coefficients of the control filter are updated in blocks of data, It can be the average gradient corresponding to the sampling time of the block data. It approximately satisfies the following formula:

[0096]

[0097] Where, N B This indicates the number of sampling moments in the block of data, or in other words, the number of sampling points.

[0098] In other words, in Figure 3 In the method shown, the gradient of the performance surface It can be based on the error signal e(k) and the filtered reference signal. It is certain. That is, in Figure 3 In the method shown, the error signal e(k) and the filtered reference signal can be used as a basis. The coefficients of the control filter W(z) are updated using the inverse matrix of the autocorrelation matrix of the filtered reference signal.

[0099] exist Figure 4 In the noise control method shown, the inverse matrix of the autocorrelation matrix of the reference signal x(k) and the inverse matrix of the autocorrelation matrix of the delayed reference signal x(k) are approximately equal, R -1 (k) can be the inverse of the autocorrelation matrix of the reference signal x(k), or, R -1 (k) can be the inverse of the autocorrelation matrix of the delayed reference signal x(kJ). It can be determined based on the filtered error signal and the delayed reference signal x (kJ). That is, in Figure 4 In the method shown, the coefficients of the control filter W(z) can be updated based on the inverse matrix of the autocorrelation matrix of the filtered error signal, the delayed reference signal x(kJ), and the autocorrelation matrix of the delayed reference signal x(kJ). Alternatively, in... Figure 4In the method shown, the coefficients of the control filter W(z) can be updated based on the filtered error signal, the delayed reference signal x(kJ), and the inverse matrix of the autocorrelation matrix of the reference signal x(k).

[0100] As can be seen from the above formula, updating the coefficients of the control filter W(z) requires matrix inversion. Since the computational complexity of matrix inversion is the cube of the matrix dimension, the computational complexity increases with the number of signal channels and the length of the control filter. For example, the filtered reference signal... The dimension of the autocorrelation matrix is ​​N. r *N s *L w The computational complexity of the inverse operation is O((N) r *N s *L w ) 3 ).

[0101] This application provides a noise control method that reduces computation and improves signal processing efficiency by updating the coefficients of the control filter in segments.

[0102] Figure 5 This illustration shows a schematic flowchart of a method for updating the coefficients of a control filter according to an embodiment of this application. Figure 5 The method shown can be divided into three stages: the first stage, the second stage, and the third stage.

[0103] For example, Figure 5 The method shown can be applied to vehicle noise control. In this case, Figure 5 The method shown can be performed by electronic devices on the vehicle side, which may specifically include one or more devices such as the vehicle, on-board chips, or on-board units (e.g., vehicle infotainment system, on-board computer). Alternatively, Figure 5 The method shown can also be executed by a cloud service device. In this embodiment, the vehicle-side electronic device can also be referred to as a vehicle-side device, or vehicle-side; the cloud service device can also be referred to as a cloud server, cloud device, or cloud. Alternatively, Figure 5 The method shown can also be executed by a system consisting of cloud service devices and vehicle-mounted electronic devices. For example, the first and second stages are executed by cloud service devices in the cloud, and the third stage is executed by vehicle-mounted electronic devices.

[0104] For example, Figure 5 The method shown can be derived from Figure 2 The processor executes the commands.

[0105] The following is about Figure 5 The method shown will be explained. It should be noted that... Figure 5The step numbers in the method shown are for descriptive convenience only and do not constitute a limitation on the execution order of the steps.

[0106] The first stage involves determining the control filter W. s (z) Length.

[0107] For example, the length L of the control filter is calculated based on the reference signal, the error signal, and the secondary path transfer function. w .

[0108] For example, such as Figure 5 As shown, the first stage may include the following steps:

[0109] 501, Obtain the secondary path transfer function through modeling.

[0110] 502, Obtain the reference signal 1# picked up by the reference sensor.

[0111] 503, acquire the error signal 1# picked up by the error sensor.

[0112] 504, Calculate the control filter W s The length L of (z) w That is, the length L of the control filter is calculated based on the reference signal 1#, the error signal 1#, and the secondary path transfer function. w .

[0113] The number of reference sensors in step 502 can be one or more. The number of channels for reference signal 1# can be one or more. For example, step 502 may include acquiring reference signals from multiple channels picked up by the array of reference sensors.

[0114] The number of error sensors in step 503 can be one or more. The number of channels for error signal 1# can also be one or more. For example, step 503 may include acquiring error signals from multiple channels picked up by the array of error sensors.

[0115] In step 504, the control filter W can be determined through offline simulation based on the reference signal 1#, the error signal 1#, and the secondary path transfer function. s The length L of (z) w Alternatively, in step 504, the length L of the control filter can be determined online in real time through statistical analysis based on the reference signal 1#, the error signal 1#, and the secondary path transfer function. w This application does not limit the scope of the embodiments.

[0116] The number of channels for the reference signal is N r There are N secondary sound sources with N channels. sFor the convenience of understanding and clear description, in this case, the number of control filters can be regarded as N r *N s In the first stage, the length L w of each control filter, i.e. the number of coefficients L w of each control filter, can be determined.

[0117] The first stage can be performed offline, or the first stage can also be performed online.

[0118] It should be understood that the above is only an example, and the length L s of the control filter W w (z) can also be determined in other ways in the first stage, which is not limited by the embodiments of the application.

[0119] In the second stage, the control filter W s (z) is segmented.

[0120] The control filter W s (z) is segmented based on the length of the control filter W s (z).

[0121] For example, as shown in FIG. 5, the second stage can include the following steps: Figure 5

[0122] 505, obtaining a reference signal 2# picked up by a reference sensor.

[0123] 506, filtering the reference signal 2# to obtain a filtered reference signal 2#.

[0124] Specifically, the reference signal 2# is filtered by a secondary path transfer function to obtain the filtered reference signal 2#.

[0125] 507, calculating the correlation coefficient of the filtered reference signal 2#.

[0126] 508, determining the length L ws of each segment coefficient of the control filter W s (z).

[0127] Specifically, the length L ws of each segment coefficient of the control filter W s (z) is determined according to the correlation coefficient of the filtered reference signal 2#. L ws is a positive integer.

[0128] 509, segmenting the coefficients of the control filter W s (z).

[0129] Specifically, the coefficients of the control filter W s (z) are segmented based on the length of the control filter W s (z).s the length L of each segment of the coefficients of W ws The coefficients of the control filter are segmented.

[0130] It should be noted that the reference signal 2# in the second stage and the reference signal 1# in the first stage can be collected in the same period of time, or can be collected in different periods of time. In other words, the reference signal 2# used for determining the length of the control filter W s (z) and the reference signal used for determining the length of the control filter W s (z) can be the same or different. In the case that the reference signal 2# in the second stage and the reference signal 1# in the first stage are the same, the step 505 and the step 502 can also be regarded as the same step.

[0131] The second stage can be performed offline, or the second stage can also be performed online.

[0132] The secondary path transfer function in the step 506 can also be understood as a secondary path transfer function model. Exemplarily, the secondary path transfer function in the step 506 can be obtained by modeling in the step 501.

[0133] In the step 507, autocorrelation calculation and cross-correlation calculation are performed on the filtered reference signal 2#, or in other words, autocorrelation responses and cross-correlation responses between the filtered reference signal 2# are calculated to obtain correlation coefficients of the filtered reference signal 2#. The correlation coefficients of the filtered reference signal 2# can be used to indicate the correlation of the filtered reference signal 2#.

[0134] In the step 508, a set of the number of sampling points required for the correlation coefficients of the filtered reference signal 2# to decay to a first threshold value is counted, and the length L of each segment of the coefficients of the control filter W s (z) is determined according to the set. ws .

[0135] The correlation coefficients decaying to the first threshold value can be understood as the correlation coefficients being less than or equal to the first threshold value.

[0136] Exemplarily, the first threshold value can be between 30%-5% of the peak value of the correlation coefficients. The first threshold values between different channels can be the same or different, and the embodiments of the present application do not make any limitation. For example, the first threshold value 1# between the channel 1# and the channel 2# can be determined according to 30%-5% of the peak value of the correlation coefficients between the channel 1# and the channel 2#; the first threshold value 2# between the channel 3# and the channel 4# can be determined according to 30%-5% of the peak value of the correlation coefficients between the channel 3# and the channel 4#. The first threshold value 1# and the first threshold value 2# can be the same or different.

[0137] For example, the length L of each coefficient segment of the control filter ws It can be determined based on the maximum value in that set. For example, the length L of each coefficient segment of the control filter. ws It is greater than or equal to the maximum value in the set.

[0138] Alternatively, the length L of each coefficient segment of the control filter can be adjusted. ws Greater than or equal to most values ​​in the set. For example, the length L of each segment of the control filter coefficients. ws Values ​​that are greater than or equal to 50% or more of the values ​​in the set.

[0139] In some cases, the correlation coefficient between some channels decays very slowly, while the correlation coefficient between other channels decays very quickly. The length L of each coefficient segment of the control filter... ws It can be set to a value greater than or equal to most values ​​in the set. This can prevent channels with slow correlation coefficient decay from affecting the length of each coefficient segment of the control filter, thus affecting the computational load.

[0140] Figure 6 This example illustrates the variation of the correlation coefficients between the two channels of the filtered reference signal #2 with the delay in sampling time. From... Figure 6 It can be seen that the correlation coefficient between the two channels decreases with increasing time delay. If the first threshold is δ, it is 4.5 * 10^- ... -6 After a delay of approximately 500 sampling times, or in other words, after approximately 500 sampling points, the correlation coefficient decays to the first threshold. If the number of sampling points required for the correlation coefficients of other channels in the filtered reference signal 2# to decay to the first threshold is less than or equal to 500, then the length L of each coefficient segment of the control filter can be adjusted. ws Set it to 500.

[0141] For example, in step 509, the length L of each segment coefficient of the control filter is... ws Divide the coefficients of the control filter into P segments, that is, satisfy L w =P*L ws In this case, the length L of the P-segment coefficients of the control filter... ws They are the same.

[0142] It should be noted that the length L of the P-segment coefficients of the control filter... ws The values ​​can also be different, and this application embodiment does not limit this. For ease of understanding and description, this application embodiment uses the length L of each coefficient segment. ws The examples are the same and are used for illustration only, and do not constitute a limitation on the solutions of the embodiments in this application.

[0143] Dividing the coefficients of the control filter into P segments can also be called dividing the control filter into P segments. P is a positive integer greater than 1. For example, P can be 2, 4, 8, 16, etc. For example, L... ws It can be 512, 256, 128, 64, etc.

[0144] Alternatively, step 509 can be understood as, putting N r *N s The coefficients of each control filter in the control filter are divided into P segments. In other words, each segment of coefficients includes N segments. r *N s The coefficients of the corresponding segment in each control filter.

[0145] For example, the p-th coefficient W in the P-th segment coefficients s,p It can be represented as:

[0146]

[0147] Where p = 0, 1, 2, ..., P-1. 1,1 (pL ws ),…,w 1,1 ((p+1)L ws -1) represents the L of the p-th segment of the control filter corresponding to the first secondary sound source from the reference signal of the first channel. ws A coefficient. Specifically, w 1,1 (pL ws ) represents the pL-th control filter corresponding to the first secondary sound source from the reference signal of the first channel. ws Each coefficient, w 1,1 ((p+1)L ws -1) represents the (p+1)Lth level of the control filter corresponding to the first secondary sound source from the reference signal of the first channel. ws -1 coefficient. w 2,1 (pL ws ),…,w 2,1 ((p+1)L ws -1) represents the L of the p-th segment of the control filter corresponding to the first secondary sound source from the reference signal of the second channel. ws Each coefficient, w 2,1 (pL ws ) represents the pL-th control filter of the control filter corresponding to the first secondary sound source from the reference signal of the second channel. ws Each coefficient, w 2,1 ((p+1)L ws -1) represents the (p+1)Lth level of the control filter corresponding to the first secondary sound source from the reference signal of the second channel. ws -1 coefficient, and so on. Control filter Ws The coefficient W of (z) s Satisfy W s =[W s,0 W s,1 ,…,W s,P-1 ] T .

[0148] The reference signal channels exhibit strong correlation. Separating strongly correlated signals for processing may affect the algorithm's convergence speed and noise reduction performance. In this embodiment, the segment length is determined based on the correlation of the filtered reference signal. The correlation between sub-signals in different segments is weak, thus reducing computational load while ensuring that convergence speed and noise reduction performance are not affected.

[0149] Furthermore, the second phase can be repeated.

[0150] For example, the second-stage process is executed once every time interval T to update the length of each segment coefficient of the control filter and to re-segment the control filter.

[0151] For example, the reference signal 2# used when performing the second phase includes part or all of the reference signal collected within the time interval T before the current time.

[0152] This allows for dynamic adjustment of the length of each coefficient segment of the control filter based on signal characteristics, thereby improving the adaptability of the method and ensuring the convergence speed of the control filter update, thus guaranteeing a better noise reduction effect.

[0153] In the third stage, the control filter W s The coefficients of (z) are updated in segments.

[0154] Control filter W s The update process of the coefficients of each segment of (z) is independent.

[0155] For example, such as Figure 5 As shown, the third stage may include the following steps:

[0156] 510, acquire the reference signal 3# picked up by the reference sensor.

[0157] 511. Filter the reference signal 3# to obtain the filtered reference signal 3#.

[0158] Specifically, the reference signal 3# is filtered through the secondary path transfer function to obtain the filtered reference signal 3#.

[0159] 512, the filtered reference signal 3# is segmented to obtain the P segment sub-signal of the filtered reference signal 3#.

[0160] Specifically, the filtered reference signal 3# is segmented according to the length L ws to obtain P sub-signals.

[0161] 513, the error signal 2# picked up by the error sensor is obtained.

[0162] 514, P coefficients of the control filter are respectively updated based on at least one of the sub-signals in the error signal 2# and the filtered reference signal 3#.

[0163] It should be noted that the reference signal 3# in the third stage and the reference signal 1# in the first stage can be collected in the same time period or in different time periods. If the reference signal 3# in the third stage and the reference signal 1# in the first stage are collected in the same time period, the step 510 and the step 502 can be the same step. The reference signal 3# in the third stage and the reference signal 2# in the second stage can be collected in the same time period or in different time periods. If the reference signal 3# in the third stage and the reference signal 1# in the first stage are collected in the same time period, the step 510 and the step 505 can be the same step. Exemplarily, the sampling time period of the reference signal 2# is not later than the sampling time period of the reference signal 3#. Exemplarily, the sampling time period of the reference signal 1# is not later than the sampling time period of the reference signal 3#. The error signal 2# in the third stage and the error signal 1# in the first stage can be collected in the same time period or in different time periods. If the error signal 2# in the third stage and the error signal 1# in the first stage are collected in the same time period, the step 503 and the step 513 can be the same step. Exemplarily, the sampling time period of the error signal 1# is not later than the sampling time period of the error signal 2#.

[0164] For ease of description, in the embodiments of the present application, the reference signal 3# can also be referred to as a reference signal x(k), and the filtered reference signal 3# can also be referred to as a filtered reference signal x(k). The error signal 2# can also be referred to as an error signal e(k).

[0165] In the step 512, the filtered reference signal 3# is segmented according to the length L ws based on the sampling time. Alternatively, it can be understood that the signal of each channel in the filtered reference signal 3# is segmented according to the length L ws based on the sampling time.

[0166] The P sub-signals and the P coefficients of the control filter are one-to-one corresponding.

[0167] Exemplarily, the p-th sub-signal in the P sub-signals can be represented as:

[0168]

[0169] in, k-pL ws The reference signal of the first channel at any given time is the reference signal after being filtered by the secondary path transfer function between the first secondary sound source and the first error microphone. Represents k-(p+1)L ws The reference signal of the first channel at time +1 is the reference signal after being filtered by the secondary path transfer function between the first secondary sound source and the first error microphone. And so on. Represents k-(p+1)L ws +1 time N r The reference signal of the Nth channel passes through the Nth channel. s From the Nth secondary sound source e The reference signal after filtering by the secondary path transfer function between the error sensors.

[0170] In one embodiment, step 514 may include updating the p-th segment coefficients of the control filter based on the p-th segment sub-signal in the error signal 2# and the filtered reference signal 3#.

[0171] In this way, the control filter W can be updated based on the P segment sub-signals in the error signal 2# and the filtered reference signal 3# respectively. s The P-segment coefficients of (z).

[0172] Control filter W s The update of the coefficients of (z) can satisfy the following formula:

[0173]

[0174] in, For the p-th segment coefficients of the updated control filter, Let μ be the coefficient of the p-th segment of the control filter before the update, and μ be the iteration step size. Figure 5 In the method shown, It can be the filtered reference signal The p-th segment signal The inverse of the autocorrelation matrix. The gradient of the performance surface In segment p, the gradient of the performance surface can satisfy... Specifically, It can be composed of the error signal e(k) and the p-th segment of the filtered reference signal. Calculated. For example, It can be the p-th segment of the error signal e(k) and the filtered reference signal. the product of the error signal and the filtered reference signal.

[0175] In an embodiment, step 510 can comprise updating the P coefficients of the controller filter W s (z) based on any one of the P sub-signals in the error signal and the filtered reference signal.

[0176] For example, in the case that the reference signal is a slow time-varying signal, the autocorrelation matrices of the P sub-signals in the filtered reference signal are approximately equal, i.e. satisfy the following equation:

[0177] R s,0 (k)≈R s,1 (k)≈...≈R s,p-1 (k);

[0178] In the case that the reference signal is a slow time-varying signal, the P coefficients of the controller filter can be updated using the autocorrelation matrix of the same sub-signal.

[0179] Alternatively, in the case that the length of the coefficients of each segment of the segmented controller filter is short, the autocorrelation matrices of the P sub-signals in the filtered reference signal are approximately equal, and the P coefficients of the controller filter W s (z) can also be updated using the autocorrelation matrix of the same sub-signal.

[0180] For example, in the case that the length of the coefficients of each segment of the controller filter divided by the length of the data of the reference signal used to calculate the autocorrelation matrix is less than 20%, the autocorrelation matrices of the P sub-signals can be considered approximately equal.

[0181] Alternatively, in the case that the autocorrelation matrices of other P sub-signals are approximately equal, the P coefficients of the controller filter W s (z) can also be updated using the autocorrelation matrix of one sub-signal, which is not limited in the embodiments of the present application.

[0182] Thus, updating the P coefficients of the controller filter using the autocorrelation matrix of the same sub-signal can further reduce the calculation amount, reduce the demand for calculation capacity, and improve the signal processing efficiency.

[0183] It should be understood that the above is only an example, and does not limit the scheme of the embodiments of the present application. For example, step 514 can also comprise updating the Q adjacent coefficients in the controller filter based on one sub-signal in the error signal 2# and the filtered reference signal 3#. Q is a positive integer less than P and greater than 1. For example, P is 9 and Q is 3, and the 9 coefficients are divided into 3 groups, each group including 3 adjacent coefficients, and the 3 coefficients in the same group are updated based on the autocorrelation matrix of the same sub-signal, and the coefficients in different groups are updated based on the autocorrelation matrices of different sub-signals.

[0184] The following will be described in combination withFigure 7 An exemplary description of a noise control method provided in an embodiment of this application will be given.

[0185] Reference signal x(k) acquired by reference sensor ( Figure 7 In After passing through the primary path transfer function P(z), the original noise signal p(k) at the error sensor is obtained. Control filter W s (z) The reference signal x(k) is filtered to obtain the control signal y(k). The control signal y(k) is then filtered by the secondary path transfer function S(z) to obtain the secondary noise signal s(k). The error signal e(k) can be obtained by superimposing the secondary noise signal s(k) and the original noise signal p(k). The error signal e(k) is acquired by an error sensor. The reference signal x(k) is processed by the secondary path transfer function model. The filtered reference signal is obtained after filtering. Secondary path transfer function model This can also be understood as an estimation of the secondary path transfer function. For the filtered reference signal... According to the length L of each coefficient segment in the control filter ws Segmentation, i.e. Figure 7 In To obtain the P-segment signal, i.e. Figure 7 In The P segment signal respectively The inverse matrix of the autocorrelation matrix, i.e. Figure 7 In and the P segment signal Perform the product operation. Based on the product of the error signal e(k) and the P-segment sub-signal, update the coefficients of the corresponding segments in the control filter, i.e., update W. s,0 W s,1 ,…,W s,P-1 To facilitate understanding and clear description, Figure 7 The number of secondary sound sources is set to 1, which does not limit the scheme of the embodiments of this application. Figure 7 w1(0),...,w1(L) ws -1) represents the coefficient of the p-th segment of the control filter corresponding to the reference signal of the first channel, where p is 0, and so on.

[0186] The length of each coefficient segment in the control filter can be determined according to... Figure 5 The method shown is used to determine this.

[0187] It should be understood that Figure 7 The noise control method shown here, which updates the coefficients of the control filter, is merely an example. Other methods can also be used to update the coefficients of the control filter; for details, please refer to [link to relevant documentation].Figure 5 The third stage of the method shown will not be described here to avoid repetition. Figure 3 The method of noise control shown and Figure 7 The difference between the method of noise control shown lies in the way of updating the coefficients of the control filter, and other descriptions can be referred to the method 300, which will not be described here to avoid repetition.

[0188] Figure 8 The simulation comparison of the scheme of the embodiment of the application and the LMS Newton algorithm is shown. Figure 8 (a) of the simulation comparison of the scheme of the embodiment of the application and the LMS Newton algorithm is shown, Figure 8 (b) of the simulation comparison of the scheme of the embodiment of the application and the LMS Newton algorithm is shown. As Figure 8 As shown, in the case of using the same amount of data, the scheme of the embodiment of the application and the LMS Newton algorithm can achieve the same noise reduction effect, that is, the noise reduction amount is 6.2 dB, and the consumption time of the simulation of the scheme of the embodiment of the application is 257 seconds, and the consumption time of the simulation of the LMS Newton algorithm is 2306 seconds. In other words, the scheme of the embodiment of the application can achieve the same noise reduction effect as the LMS Newton algorithm, and the calculation amount is only 11.14% of the LMS Newton algorithm.

[0189] Figure 9 The schematic flowchart of another method of updating the coefficients of the control filter provided by the embodiment of the application is shown. Figure 9 The method shown can be divided into three stages: the first stage, the second stage and the third stage. Among them, the first stage and the second stage are similar to Figure 5 the method shown, and in order to avoid repetition, part of the description will be omitted when describing Figure 9 the method shown.

[0190] Exemplarily, Figure 9 The method shown can be applied to the field of vehicle noise control. In this case, Figure 9 The method shown can be executed by an electronic device at the vehicle end, which can specifically include one or more of a vehicle, a vehicle-mounted chip or a vehicle-mounted device (such as a vehicle machine, a vehicle-mounted computer) and the like. Or, Figure 9 The method shown can also be executed by a cloud service device. In the embodiment of the application, the electronic device at the vehicle end can also be referred to as a vehicle end device, or a vehicle end; the cloud service device can also be referred to as a cloud end server, a cloud end device or a cloud end. Or, Figure 9 The method shown can also be executed by a system composed of a cloud service device and an electronic device at the vehicle end. For example, the first stage and the second stage are executed by the cloud service device at the cloud end, and the third stage is executed by the electronic device at the vehicle end.

[0191] Exemplarily, Figure 9 The method shown can be executed by Figure 2The processor shown is executed. It should be noted that Figure 9 The step number in the method shown is only for convenience of description, and does not limit the execution order of the steps.

[0192] In the first stage, the length of the control filter is determined.

[0193] Exemplarily, the length L of the control filter is calculated based on the reference signal, the error signal and the secondary path transfer function w .

[0194] For example, as Figure 9 shown, the first stage can include the following steps:

[0195] 901, obtaining the secondary path transfer function by modeling.

[0196] 902, obtaining the reference signal 1# picked up by the reference sensor.

[0197] 903, obtaining the error signal 1# picked up by the error sensor.

[0198] 904, calculating the length L of the control filter w , that is, calculating the length L of the control filter based on the reference signal 1#, the error signal 1# and the secondary path transfer function w .

[0199] Figure 9 The first stage shown is the same as the first stage in the method shown Figure 5 , steps 901 to 904 correspond to steps 501 to 504 respectively, and the specific description can be referred to Figure 5 , which will not be repeated here.

[0200] The second stage is to segment the control filter.

[0201] Segmenting the control filter based on the length of the control filter.

[0202] For example, as Figure 9 shown, the second stage can include the following steps:

[0203] 905, obtaining the reference signal 2# picked up by the reference sensor.

[0204] 906, calculating the correlation coefficient of the reference signal 2#.

[0205] 907, determining the length L ws of each segment coefficient of the control filter.

[0206] Specifically, the length L ws of each segment coefficient of the control filter is determined according to the correlation coefficient of the reference signal 2#. L wsIt is a positive integer.

[0207] 908, segment the coefficients of the control filter.

[0208] Specifically, based on the length L of each segment of the control filter coefficients ws The coefficients of the control filter are segmented.

[0209] Figure 9 The second stage shown and Figure 5 The difference in the second stage shown is that, Figure 9 The length L of each coefficient segment of the control filter in the process ws It is determined based on the correlation coefficient of reference signal 2#. Figure 5 The length L of each coefficient segment of the control filter in the process ws The correlation coefficient is determined based on the filtered reference signal 2#. Steps 906 to 908 are obtained by replacing the filtered reference signal 2# in steps 507, 508, and 509 with the actual reference signal 2#. Other descriptions can be found in [reference needed]. Figure 5 This will not be elaborated upon here.

[0210] Alternatively, the second stage may include delaying the reference signal 2# to obtain a delayed reference signal 2#. In this case, step 907 may determine the length L of each coefficient of the control filter based on the delayed reference signal 2#. ws In other words, the filtered reference signal 2# in steps 507, 508, and 509 can be replaced with the delayed reference signal 2#.

[0211] In the third stage, the coefficients of the control filter are updated in segments.

[0212] In other words, the coefficients of each segment of the control filter are updated separately.

[0213] For example, such as Figure 9 As shown, the third stage may include the following steps:

[0214] 909, acquire the reference signal 3# picked up by the reference sensor.

[0215] 910. The reference signal 3# is delayed to obtain the delayed reference signal 3#.

[0216] 911, the delayed reference signal 3# is segmented to obtain the P segment sub-signal of the delayed reference signal 3#.

[0217] Specifically, the delayed reference signal 3# is processed according to length L. ws The signal is segmented to obtain the P segment sub-signal of the delayed reference signal 3#.

[0218] 912, acquire the error signal 2# picked up by the error sensor.

[0219] 913, filter the error signal 2# by a time-reversed secondary path transfer function to obtain a filtered error signal 2#.

[0220] 914, update P coefficients of the control filter based on at least one sub-signal in the filtered error signal 2# and the delayed reference signal 3# respectively.

[0221] It should be noted that the reference signal 3# in the third stage and the reference signal 1# in the first stage can be collected in the same time period, or can be collected in different time periods. If the reference signal 3# in the third stage and the reference signal 1# in the first stage are collected in the same time period, step 909 and step 902 can be the same step. The reference signal 3# in the third stage and the reference signal 2# in the second stage can be collected in the same time period, or can be collected in different time periods. If the reference signal 3# in the third stage and the reference signal 1# in the first stage are collected in the same time period, step 909 and step 905 can be the same step. Exemplarily, the sampling time period of the reference signal 2# is not later than the sampling time period of the reference signal 3#. Exemplarily, the sampling time period of the reference signal 1# is not later than the sampling time period of the reference signal 3#. The error signal 2# in the third stage and the error signal 1# in the first stage can be collected in different time periods, or can be collected in the same time period. If the error signal 2# in the third stage and the error signal 1# in the first stage are collected in the same time period, step 903 and step 912 can be the same step. Exemplarily, the sampling time period of the error signal 1# is not later than the sampling time period of the error signal 2#.

[0222] In step 910, the reference signal 3# is delayed for J sampling time periods to obtain a delayed reference signal. Wherein, J is determined according to the length of the secondary path transfer function model. Or, J is determined according to the length of the secondary path modeling filter.

[0223] For ease of description, in the embodiments of the present application, the reference signal 3# can also be referred to as a reference signal x(k), the delayed reference signal 3# can also be referred to as a delayed reference signal x(k-J), and the error signal 2# can also be referred to as an error signal e(k).

[0224] In step 911, the delayed reference signal 3# is segmented according to the sampling time period based on the length L ws of each channel of the filtered delayed reference signal 3#. Or, it can be understood that the signal of each channel of the filtered delayed reference signal 3# is segmented according to the sampling time period based on the length L ws .

[0225] The P-segment signal and the P-segment coefficient of the controller are in one-to-one correspondence.

[0226] In one embodiment, step 914 may include updating the p-th segment coefficients of the control filter based on the p-th segment sub-signal in the filtered error signal 2# and the delayed reference signal 3#.

[0227] In this way, the P-segment coefficients of the control filter can be updated according to the P-segment sub-signals in the filtered error signal 23 and the delayed reference signal 3#, respectively.

[0228] Control filter W s The update of the coefficients of (z) can satisfy the following formula:

[0229]

[0230] in, For the p-th segment coefficients of the updated control filter, Let μ be the coefficient of the p-th segment of the control filter before the update, and μ be the iteration step size. Figure 9 In the method shown, It can be the inverse matrix of the autocorrelation matrix of the p-th segment of the delayed reference signal x(kJ). The gradient of the performance surface In segment p, the gradient of the performance surface satisfies Specifically, It can be calculated from the p-th segment of the filtered error signal 2# and the delayed reference signal 3#. For example, It can be the product of the filtered error signal and the p-th segment of the delayed reference signal 3#.

[0231] Furthermore, since the inverse matrix of the autocorrelation matrix of the p-th segment of the delayed reference signal x(k) is approximately equal to the inverse matrix of the autocorrelation matrix of the p-th segment of the reference signal x(k), in Figure 9 In the method shown, It can also be the inverse matrix of the autocorrelation matrix of the p-th sub-signal of the reference signal x(k). In this case, step 911 further includes: adjusting the reference signal 3# according to length L ws The signal is segmented to obtain the P-segment sub-signals of the reference signal 3#. Step 914 may include updating the P-segment coefficients of the control filter based on at least one segment of the filtered error signal 2#, the reference signal 3#, and at least one segment of the delayed reference signal 3#. At least one segment of the reference signal 3# and at least one segment of the delayed reference signal 3# correspond to each other.

[0232] In one embodiment, step 914 may include updating the P-segment coefficients of the controller filter based on any one of the sub-signals in the filtered error signal 2# and the delayed reference signal 3#.

[0233] For example, when the reference signal is a slow time-varying signal, the autocorrelation matrices of the P-segment sub-signals in the reference signal are approximately equal.

[0234] When the reference signal is a slow time-varying signal, the P-segment coefficients of the control filter can be updated using the autocorrelation matrix of the same sub-segment signal.

[0235] Alternatively, when the coefficients of the segmented control filter are short, the autocorrelation matrices of the P-segment sub-signals in the reference signal are approximately equal, and the autocorrelation matrices of the same sub-signal can be used to update the P-segment coefficients of the control filter.

[0236] Alternatively, if the autocorrelation matrices of other P-segment sub-signals are approximately equal, the autocorrelation matrix of one sub-signal can be used to update the P-segment coefficients of the filter. This application does not limit this approach.

[0237] By using the autocorrelation matrix of the same sub-signal to update the P-segment coefficients of the control filter, the computational load can be further reduced, the computational power requirement can be lowered, and the signal processing efficiency can be improved.

[0238] It should be understood that the above are merely examples and do not constitute a limitation on the solutions of the embodiments of this application.

[0239] The following is combined Figure 10 An exemplary description of a noise control method provided in an embodiment of this application will be given.

[0240] Reference signal x(k) acquired by reference sensor ( Figure 10 In After passing through the primary path transfer function P(z), the original noise signal p(k) at the error sensor is obtained. Control filter W s (z) The reference signal x(k) is filtered to obtain the control signal y(k). The control signal y(k) is then filtered by the secondary path transfer function S(z) to obtain the secondary noise signal s(k). The error signal e(k) can be obtained by superimposing the secondary noise signal s(k) and the original noise signal p(k). The error signal e(k) is acquired by an error sensor. The reference signal x(k) is delayed by J sampling times to obtain the delayed reference signal x(kJ). The delayed reference signal x(kJ) is then processed according to the length L of each coefficient in the control filter. ws Segmentation, i.e. Figure 10 In P segment sub-signals of the delayed reference signal. The inverse matrix of the autocorrelation matrix of the P segment sub-signals is multiplied by the P segment sub-signals respectively, i.e. Figure 10 The error signal e(k) is filtered by the inverse secondary path transfer function, and a filtered error signal is obtained. The product of the filtered error signal and the P segment sub-signals is used to update the coefficients of the corresponding segment of the control filter, i.e. W s,0 ,W s,1 ,…,W s,P-1 For the sake of understanding and clear description, Figure 10 The number of secondary sound sources in

[0241] The length of each segment coefficient of the control filter can be determined according to the method shown in Figure 9

[0242] It should be understood that Figure 10 The updating method of the coefficients of the control filter in the noise control method shown in Figure 9 The third stage of the method shown in Figure 4 The difference between the noise control method shown in Figure 10 and the noise control method shown in the method 400 is the updating method of the coefficients of the control filter. For the sake of brevity, the other descriptions can be referred to the method 400.

[0243] According to the scheme of the embodiments of the present application, the P segment coefficients of the control filter are updated respectively, which reduces the dimension of the autocorrelation matrix, thereby reducing the computational complexity of the inverse operation of the autocorrelation matrix, reducing the computational load, reducing the demand for computing resources, improving the signal processing efficiency, and being conducive to meeting the real-time requirements. For example, taking the method shown in Figure 5 or Figure 6 as an example, the p-th segment coefficient of the control filter is updated based on the error signal and the p-th segment sub-signal of the filtered reference signal. In this way, in the updating process of the control filter, the dimension of the autocorrelation matrix is reduced from N r *N s *L w to N r *N s *L w / P, and the computational complexity of the inverse operation is reduced to 1 / (P 3 ) of the overall updating scheme of the control filter.

[0244] ​​Furthermore, the embodiments of this application determine the length of the segmented coefficients of the control filter based on the correlation between signals. For example, the length of the segmented coefficients of the control filter is determined according to the correlation of the filtered reference signal. The correlation between sub-signals in different segments is weak, which can reduce the amount of computation while ensuring that the convergence speed and noise reduction effect are not affected. The scheme of the embodiments of this application has a similar convergence speed and noise reduction effect to the LMS Newton algorithm, and the amount of computation is much less than that of the LWS Newton algorithm. That is, the scheme of the embodiments of this application can reduce the amount of computation, reduce the demand for computing resources, and improve signal processing efficiency while ensuring the noise reduction effect.

[0245] The signal processing method provided in this application embodiment can also be applied to other signal processing fields besides active noise control. For signal processing fields that do not include secondary path steps, such as echo cancellation or audio noise reduction, the steps involving secondary paths and signal delay can be omitted when performing the above method.

[0246] Figure 11 A schematic diagram of another method for updating the coefficients of a control filter provided in an embodiment of this application is shown. Figure 11 The method shown can be divided into three stages: the first stage, the second stage, and the third stage. Figure 11 The method shown is the same as Figure 5 The method shown and Figure 9 The methods shown are similar. To avoid repetition, in the description... Figure 11 When illustrating the method, some descriptions may be omitted as appropriate.

[0247] It should be noted that, Figure 11 The step numbers in the method shown are for descriptive convenience only and do not constitute a limitation on the execution order of the steps.

[0248] by Figure 11 Taking the method illustrated in the field of echo cancellation as an example, the reference signal can be a far-end signal, and the error signal can be obtained by superimposing the desired signal and the output signal of the control filter. The desired signal can include the echo signal. The control filter processes the far-end speech signal to obtain the output signal of the control filter, i.e., the control signal, or in other words, the estimated value of the echo signal. The output signal and the echo signal are superimposed to obtain the error signal. For example, the echo signal can be a near-end signal containing the echo signal, acquired by a near-end microphone. The control filter updates its coefficients to reduce the error signal, so that the output signal of the control filter can better cancel the echo signal.

[0249] The first stage involves determining the length of the control filter.

[0250] For example, the length L of the control filter is calculated based on the reference signal and the error signal. w.

[0251] For example, such as Figure 11 As shown, the first stage may include the following steps:

[0252] 1101, Obtain reference signal 1#.

[0253] 1102, Obtain error signal 1#.

[0254] 1103, Calculate the length L of the control filter. w That is, the length L of the control filter is calculated based on the reference signal 1# and the error signal 1#. w .

[0255] The second stage involves segmenting the control filter.

[0256] The control filter is segmented based on its length.

[0257] For example, such as Figure 11 As shown, the second stage may include the following steps:

[0258] 1104, obtain reference signal 2#.

[0259] Reference signal 2# and reference signal 1# can come from the same one or more signal sources.

[0260] 1105, Calculate the correlation coefficient of reference signal 2#.

[0261] 1106, Determine the length L of each coefficient segment of the control filter. ws .

[0262] Specifically, the length L of each segment of the control filter coefficients is determined based on the correlation coefficient of reference signal 2#. ws L ws It is a positive integer.

[0263] 1107, segment the coefficients of the control filter.

[0264] Specifically, based on the length L of each segment of the control filter coefficients ws The coefficients of the control filter are segmented.

[0265] In the third stage, the coefficients of the control filter are updated in segments.

[0266] In other words, the coefficients of each segment of the control filter are updated separately.

[0267] For example, such as Figure 11 As shown, the third stage may include the following steps:

[0268] 1108, Obtain reference signal 3#. Reference signal 3#, reference signal 2#, and reference signal 1# can come from the same one or more signal sources.

[0269] 1109, the reference signal 3# is segmented to obtain the P segment sub-signal of the reference signal 3#.

[0270] Specifically, for reference signal 3# according to length L ws The signal is segmented to obtain the P segment sub-signal of reference signal 3#.

[0271] 1110, acquire error signal 2#. Error signal 2# and error signal 1# can come from the same one or more signal sources.

[0272] 1111, update the P-segment coefficients of the control filter based on at least one sub-signal from error signal 2# and reference signal 3# respectively.

[0273] The P-segment signal and the P-segment coefficient of the controller are in one-to-one correspondence.

[0274] In one embodiment, step 1111 may include updating the p-th segment coefficients of the control filter based on the p-th segment sub-signal in the error signal 2# and the reference signal 3#.

[0275] In this way, the P-segment coefficients of the control filter can be updated according to the P-segment sub-signals in error signal 2# and reference signal 3# respectively.

[0276] Control filter W s The update of the coefficients of (z) can satisfy the following formula:

[0277]

[0278] in, For the p-th segment coefficients of the updated control filter, Let μ be the coefficient of the p-th segment of the control filter before the update, and μ be the iteration step size. Figure 11 In the method shown, It can be the inverse matrix of the autocorrelation matrix of the p-th sub-signal of the reference signal x(k). The gradient of the performance surface In segment p, the gradient of the performance surface satisfies Specifically, It can be calculated from the p-th segment of the error signal 2# and the reference signal 3#. For example, It can be the product of the error signal e(k) and the p-th segment of the reference signal.

[0279] from Figure 11 and Figure 5 The comparison shows thatFigure 11 The method shown is to... Figure 5 The method shown is obtained by skipping steps involving secondary paths. From Figure 11 and Figure 9 The comparison shows that Figure 11 The method shown is to... Figure 5 The method shown is obtained by skipping steps involving secondary paths and delays. For related descriptions, please refer to [reference needed]. Figure 5 or Figure 9 The description will not be repeated here.

[0280] Figure 12 A schematic flowchart of a signal processing method provided in an embodiment of this application is shown.

[0281] The reference signal x(k) is processed by filter H(z) to obtain the desired signal d(k). Control filter W... s (z) The reference signal x(k) is filtered to obtain the control signal y(k), i.e., the output signal. The error signal e(k) can be obtained by superimposing the output signal and the desired signal. The reference signal x(k) is filtered according to the length L of each segment of the coefficients in the control filter. ws Segmentation, i.e. Figure 12 In To obtain the P-segment sub-signal of the reference signal, the inverse matrix of the autocorrelation matrix of this P-segment sub-signal is obtained, i.e. Figure 10 In Perform a product operation with the P-segment sub-signal. Based on the product result of the error signal and the P-segment sub-signal, update the coefficients of the corresponding segments in the control filter, i.e., update W. s,0 W s,1 ,…,W s,P-1 .

[0282] The length of each coefficient segment in the control filter can be determined according to... Figure 11 The method shown is used to determine this.

[0283] It should be understood that Figure 12 The signal processing method shown here, which updates the coefficients of the control filter, is merely an example. Other methods can also be used to update the coefficients of the control filter; for details, please refer to [link to relevant documentation]. Figure 11 The third stage of the method shown will not be described again here to avoid repetition.

[0284] Figure 12 The method shown is to... Figure 7 The method shown or Figure 10 The method shown involves removing secondary path steps and delayed steps; for a detailed description, please refer to [reference needed]. Figure 7 or Figure 10 Related descriptions.

[0285] The method shown in Figure 12 is taken as an example applied to the field of echo cancellation. After the far-end signal x(k) passes through the echo path H(z), the echo signal d(k) is generated. The near-end signal containing the echo signal d(k) is obtained by the near-end microphone, which is equivalent to obtaining the echo signal d(k). The control filter W s (z) is used to filter the far-end signal x(k) to obtain the output signal y(k). The error signal e(k) is determined according to the echo signal d(k) and the output signal y(k). The far-end signal x(k) is segmented according to the length L ws of each segment of the coefficient of the control filter, that is, Figure 12 to obtain P sub-signals of the reference signal. The product operation is respectively performed on the inverse matrix of the autocorrelation matrix of the P sub-signals, that is, Figure 10 and the P sub-signals. The coefficients of the corresponding segments in the control filter are updated based on the product results of the error signal and the P sub-signals, that is, W s,0 ,W s,1 ,…,W s,P-1 .

[0286] It should be understood that the method shown in Figure 11 and Figure 12 is taken as an example applied to the echo cancellation scene for description in the embodiments of the present application, and the scheme of the embodiments of the present application is not limited thereto.

[0287] According to the scheme of the embodiments of the present application, the P segment coefficients of the control filter are updated respectively, the dimension of the autocorrelation matrix is reduced, the calculation complexity of the inverse operation of the autocorrelation matrix is reduced, the calculation amount is reduced, the demand for computing resources is reduced, the signal processing efficiency is improved, and the real-time requirement is met.

[0288] In addition, the length of the segmented coefficient of the control filter is determined based on the correlation between signals in the embodiments of the present application, for example, the length of the segmented coefficient of the control filter is determined according to the correlation of the reference signal. The correlation between the sub-signals of different segments is weak, so that the calculation amount is reduced while the convergence speed and the signal processing effect are guaranteed not to be affected. The scheme of the embodiments of the present application has similar convergence speed and signal processing effect as the LMS Newton algorithm, and the calculation amount is much smaller than the LMS Newton algorithm, that is, the scheme of the embodiments of the present application can reduce the calculation amount while guaranteeing the signal processing effect, reduce the demand for computing resources, and improve the signal processing efficiency.

[0289] Figure 13 A schematic flowchart of a noise control method provided by the embodiments of the present application is shown. Exemplarily, Figure 13 ​​The method 1300 shown can be applied to the field of vehicle noise control. In this case, the method 1300 can be performed by an electronic device at the vehicle end, which can specifically include one or more of devices such as a vehicle, a vehicle-mounted chip, or a vehicle-mounted device (for example, a vehicle machine, a vehicle-mounted computer), and the like. Alternatively, the method 1300 can also be performed by a cloud service device. In the embodiments of the present application, the electronic device at the vehicle end can also be referred to as a vehicle-end device, or a vehicle end; and the cloud service device can also be referred to as a cloud-end server, a cloud-end device, or a cloud end. Alternatively, the method 1300 can also be performed by a system composed of a cloud service device and an electronic device at the vehicle end. For example, the cloud service device sends the coefficients of the control filter to the electronic device at the vehicle end, and the electronic device at the vehicle end performs steps 1301 to 1304, and the cloud service device performs step 1305.

[0290] Exemplarily, the method 1300 can be performed by a processor in the electronic device. Figure 2

[0291] In a possible implementation, the method of noise control in the embodiments of the present application can be understood as an improvement on the method of implementing noise control based on the LMS Newton algorithm.

[0292] As shown in Figure 13 , the method 1300 includes steps 1301 to 1305.

[0293] 1301, a first reference signal collected by a reference sensor is obtained.

[0294] 1302, the first reference signal is filtered by a control filter to obtain a control signal, the control signal being used to indicate a secondary noise signal.

[0295] 1303, an error signal collected by an error sensor is obtained, the error signal being obtained by superimposing an original noise signal and the secondary noise signal.

[0296] 1304, a second reference signal is determined according to the first reference signal.

[0297] 1305, P coefficients of the control filter are updated based on at least one of P sub-signals of the second reference signal and the error signal, the P sub-signals of the second reference signal being obtained based on sampling time points of the second reference signal, and the P coefficients of the control filter being obtained based on the length of the control filter, P being an integer greater than 1.

[0298] The P sub-signals of the second reference signal and the P coefficients of the controller can be one-to-one corresponding.

[0299] Exemplarily, the reference sensor can be a reference acceleration sensor or a reference microphone.

[0300] ​The number of reference sensors can be one or multiple. The number of channels of the first reference signal can be one or multiple.

[0301] Exemplarily, the first reference signal can be the reference signal x(k) in the method shown in Figure 5 、 Figure 7 、 Figure 9 or Figure 10 .

[0302] Exemplarily, the control filter in the method 1300 can be the control filter W Figure 5 、 Figure 7 、 Figure 9 or Figure 10 (z) in the method shown in s .

[0303] The number of control signals can be one or multiple. Exemplarily, the control signal in the method 1300 can be the control signal y(k) in the method shown in Figure 5 、 Figure 7 、 Figure 9 or Figure 10 .

[0304] In step 1302, the control filter can output the control signal to the secondary sound source, and the secondary noise signal is output by the secondary sound source, that is, the control signal can be used to control the secondary sound source to generate the secondary noise signal. Alternatively, the secondary noise signal is obtained by filtering the control signal through the secondary path transfer function. The secondary path refers to the transmission path between the control signal output by the control filter and the error signal. Exemplarily, the secondary path transfer function can be the secondary path transfer function S(z) in the foregoing.

[0305] The secondary noise signal can be output to the target area and superimposed with the original noise signal in the target area. In step 1303, the error signal is obtained by collecting the error microphone in the target area.

[0306] Exemplarily, the error sensor can be an error microphone. The number of error sensors can be one or multiple. The number of channels of the error signal can be one or multiple.

[0307] Exemplarily, the error signal in the method 1300 can be the error signal e(k) in the method shown in Figure 5 、 Figure 7 、 Figure 9 or Figure 10 .

[0308] The length of the P segment sub-signal of the second reference signal is the same as the length of the corresponding P segment coefficient.

[0309] For example, the step 1305 comprises: updating P segments of coefficients of the control filter respectively based on the inverse matrix of the autocorrelation matrix of at least one of the P segments of the second reference signal and P segments of the gradient of the mean square error of the error signal, or updating P segments of coefficients of the control filter respectively based on the inverse matrix of the autocorrelation matrix of at least one of the P segments of the first reference signal and P segments of the gradient of the mean square error of the error signal, the P segments of the first reference signal are segmented based on sampling instants of the first reference signal, and the P segments of the gradient of the mean square error of the error signal are determined according to the P segments of the second reference signal.

[0310] According to the scheme of the embodiments of the present application, the P segments of the coefficients of the control filter are updated respectively, i.e., the P segments of the coefficients are updated independently, which can reduce the number of multiplication operations, reduce the calculation amount in the signal processing process, reduce the demand for computing resources, and improve the efficiency of signal processing. For example, when the LMS Newton method is used to update the coefficients of the control filter, the inverse operation of a matrix needs to be performed, and the coefficients of the control filter are updated as a whole, which requires calculating the inverse matrix of the autocorrelation matrix of the second reference signal. When the scheme of the embodiments of the present application is used, only the inverse matrix of the autocorrelation matrix of at least one segment of signal needs to be calculated, which reduces the dimension of the autocorrelation matrix involved in the calculation, thereby reducing the calculation complexity of the inverse operation of the matrix, reducing the calculation amount, reducing the demand for computing resources, improving the efficiency of signal processing, and being conducive to meeting the real-time requirements in the noise reduction process.

[0311] In a possible implementation, the length of the P segments of the coefficients is determined according to the correlation of the third reference signal, the third reference signal is determined according to a fourth reference signal collected by a reference sensor, and the sampling time period of the fourth reference signal is not later than the sampling time period of the first reference signal.

[0312] For example, the length of each segment of the P segments of the coefficients is determined according to the correlation of the third reference signal.

[0313] Alternatively, the length of at least one segment of the P segments of the coefficients is determined according to the correlation of the third reference signal. For example, P-1 segments of the P segments of the coefficients can be determined according to the correlation of the third reference signal, and the remaining one segment of the coefficients can be determined according to the length of the control filter and the total length of the P-1 segments of the coefficients.

[0314] The lengths of the segments of the P segments of the coefficients can be the same or different.

[0315] The fourth reference signal and the first reference signal can be collected by the same reference sensor. Alternatively, the fourth reference signal and the first reference signal can come from the same signal source.

[0316] Exemplarily, the third reference signal can be filtered from the fourth reference signal by a secondary path transfer function. For example, the third reference signal can be the filtered reference signal 2# shown in the method of FIG. 6, the fourth reference signal can be the reference signal 2# shown in the method of FIG. 6. Figure 5 Figure 5

[0317] Exemplarily, the third reference signal can be delayed from the fourth reference signal. For example, the third reference signal can be the delayed reference signal 2# shown in the method of FIG. 7, the fourth reference signal can be the reference signal 2# shown in the method of FIG. 7. Figure 9 Figure 9

[0318] Exemplarily, the third reference signal can be the fourth reference signal. For example, the third reference signal can be the reference signal 2# shown in the method of FIG. 8, the fourth reference signal can be the reference signal 2# shown in the method of FIG. 8. Figure 9 Figure 9

[0319] In a possible implementation, the length of the P-segment coefficient is determined according to the correlation of the third reference signal, including: the length of the P-segment coefficient is determined according to the maximum value of the number of sampling points required for the autocorrelation coefficients and the cross-correlation coefficients of the multiple channels of the third reference signal to decay to a first threshold value.

[0320] Exemplarily, the length of the P-segment coefficient is greater than or equal to the maximum value of the number of sampling points required for the autocorrelation coefficients and the cross-correlation coefficients of the multiple channels of the third reference signal to decay to a first threshold value.

[0321] The embodiments of the present application determine the length of the segmented coefficient of the control filter based on the correlation between signals, for example, determine the length of the segmented coefficient of the control filter according to the correlation of the filtered reference signal, and the correlation between the sub-signals of different segments is weak, so that the convergence speed and the noise reduction effect can be ensured without being affected while reducing the amount of calculation. The scheme of the embodiments of the present application has similar convergence speed and noise reduction effect as the LMS Newton algorithm, and the amount of calculation is much smaller than the LMS Newton algorithm, that is, the scheme of the embodiments of the present application can reduce the amount of calculation while ensuring the noise reduction effect, reduce the demand for computing resources, and improve the signal processing efficiency.

[0322] In a possible implementation, the time interval between the sampling period of the fourth reference signal and the sampling period of the first reference signal is less than or equal to a second threshold value.

[0323] The reference signal used to determine the length of each segment coefficient of the control filter can be adjusted over time, that is, the length of each segment coefficient of the control filter can be updated over time. ​​​​​​

[0324] Exemplarily, the second threshold value can be the time interval T in the foregoing, that is, the length of each segment of coefficients of the control filter is re-determined every time interval T.

[0325] In this way, the length of each segment of coefficients of the control filter can be dynamically adjusted according to the signal characteristics, so as to improve the adaptability of the method and facilitate guaranteeing the convergence speed of the update of the control filter, that is, guaranteeing a better noise reduction effect.

[0326] In a possible implementation, the second reference signal is obtained by filtering the first reference signal through a secondary path transfer function, and the third reference signal is obtained by filtering the fourth reference signal through the secondary path transfer function.

[0327] Exemplarily, the second reference signal can be a filtered reference signal 3# in the method shown in Figure 5 or Figure 7 , or in other words, a filtered reference signal x(k-J) in the method shown in The first reference signal can be a reference signal 3# in the method shown in Figure 5 or Figure 7 , or in other words, a reference signal x(k) in the method shown in Figure 5 or Figure 7 , or in other words, a secondary path transfer function model in the method shown in For example, the third reference signal can be a filtered reference signal 2# in the method shown in Figure 5 , and the fourth reference signal can be a reference signal 2# in the method shown in Figure 5 .

[0328] Exemplarily, in this case, the P segments in the gradient of the mean square error of the error signal can be respectively calculated according to P segment sub-signals of the second reference signal and the error signal.

[0329] In a possible implementation, the second reference signal is obtained by delaying the first reference signal by J sampling instants, where J is a positive integer and is determined according to the length of the secondary path transfer function.

[0330] In this case, the third reference signal can be obtained by delaying the fourth reference signal by J sampling instants, or the third reference signal can be the fourth reference signal.

[0331] Exemplarily, the second reference signal can be a delayed reference signal 3# in the method shown in Figure 9 or Figure 10 , or in other words, a delayed reference signal x(k-J) in the method shown in Figure 9 or Figure 10The reference signal 3# in the method shown, or the reference signal x(k). The secondary path transfer function can be Figure 9 or Figure 10 The secondary path transfer function model in the method shown For example, the third reference signal can be Figure 9 The delayed reference signal 2# in the method shown, and the fourth reference signal can be Figure 9 The reference signal 2# in the method shown. For another example, the third reference signal can be Figure 9 The reference signal 2# in the method shown, and the fourth reference signal can be Figure 9 The reference signal 2# in the method shown.

[0332] In this case, the step 1305 can include filtering the error signal by the inverse secondary path transfer function to obtain a filtered error signal, and updating the P coefficients of the control filter based on at least one of the P sub-signals of the second reference signal and the filtered error signal.

[0333] Exemplarily, in this case, the P segments in the gradient of the mean square error of the error signal can be calculated respectively according to the P sub-signals of the second reference signal and the filtered error signal.

[0334] In a possible implementation, the step 1305 can include updating the p-th coefficient based on the p-th sub-signal of the second reference signal and the error signal, p = 0, 1, 2…P-1.

[0335] Optionally, updating the P coefficients of the control filter based on the P segments in the gradient of the mean square error of the error signal and the inverse matrix of the autocorrelation matrix of at least one of the P sub-signals of the second reference signal respectively includes: updating the p-th coefficient of the control filter based on the p-th segment in the gradient of the mean square error of the error signal and the inverse matrix of the autocorrelation matrix of the p-th sub-signal of the P sub-signals of the second reference signal; or updating the P coefficients of the control filter based on the P segments in the gradient of the mean square error of the error signal and the inverse matrix of the autocorrelation matrix of at least one of the P sub-signals of the first reference signal respectively includes: updating the p-th coefficient of the control filter based on the p-th segment in the gradient of the mean square error of the error signal and the inverse matrix of the autocorrelation matrix of the p-th sub-signal of the P sub-signals of the first reference signal.

[0336] In this way, the P coefficients of the control filter can be updated based on the error signal and the P sub-signals of the second reference signal respectively.

[0337] In a possible implementation, the step 1305 can include updating the P coefficients based on the error signal and any sub-signal of the second reference signal respectively.

[0338] Optionally, the P segments of the P segments of the gradient of the inverse matrix of the autocorrelation matrix of the P segments of the second reference signal and the mean square error of the error signal are respectively used to update the P segments of the coefficients of the control filter, including: the P segments of the gradient of the inverse matrix of the autocorrelation matrix of any of the P segments of the second reference signal and the mean square error of the error signal are respectively used to update the P segments of the coefficients of the control filter; or the P segments of the gradient of the inverse matrix of the autocorrelation matrix of at least one of the P segments of the first reference signal and the mean square error of the error signal are respectively used to update the P segments of the coefficients of the control filter, including: the P segments of the gradient of the inverse matrix of the autocorrelation matrix of any of the P segments of the first reference signal and the mean square error of the error signal are respectively used to update the P segments of the coefficients of the control filter.

[0339] In this way, the P segments of the coefficients of the control filter are updated by using the autocorrelation matrix of the same segment of the sub-signal, so that the calculation amount can be further reduced, the demand for calculation capability can be reduced, and the signal processing efficiency can be improved.

[0340] Figure 14 A schematic flowchart of a method for signal processing provided by an embodiment of the present application is shown.

[0341] In a possible implementation manner, the method for signal processing in the embodiment of the present application can be understood as an improvement on the method for implementing signal processing based on the LMS Newton algorithm.

[0342] As shown in FIG. 14, the method 1400 includes steps 1401 to 1404. Figure 14

[0343] 1401, a first reference signal is acquired.

[0344] 1402, the first reference signal is filtered by a control filter to obtain a control signal, the control signal being used to indicate an output signal.

[0345] 1403, an error signal is acquired, the error signal being obtained by superimposing a desired signal and the output signal.

[0346] 1404, P segments of coefficients of the control filter are updated based on the error signal and at least one of P segments of sub-signals of a second reference signal, the second reference signal being determined according to the first reference signal, the P segments of sub-signals being obtained by segmenting sampling time of the second reference signal, the P segments of coefficients being obtained by segmenting a length of the control filter, and P being a positive integer greater than 1.

[0347] The P segments of sub-signals and the P segments of coefficients can be one-to-one corresponding.

[0348] ​The method 1300 can be regarded as a specific implementation of the method 1400 applied to the noise control field. Specifically, the output signal in the method 1400 can be the secondary noise signal in the method 1300, and the desired signal in the method 1400 can be the original noise signal in the method 1300. The specific description can refer to the method 1300, which will not be repeated here.

[0349] The method 1400 can also be applied to other signal processing fields, such as echo cancellation or audio noise reduction, which do not involve secondary paths.

[0350] The following takes the method 1400 applied to the signal processing field not involving the secondary path as an example for illustration.

[0351] The number of channels of the first reference signal can be one or multiple.

[0352] Exemplarily, the first reference signal in the method 1400 can be the reference signal x(k) in Figure 11 or Figure 12 , that is, the reference signal 3#.The control filter in the method 1400 can be the control filter W Figure 11 or Figure 12 (z). s

[0353] The number of control signals can be one or multiple. The control signal is the output signal.

[0354] Exemplarily, the control signal in the method 1400 can be the control signal y(k) in Figure 11 or Figure 12 .

[0355] The number of channels of the error signal can be one or multiple. Exemplarily, the error signal in the method 1400 can be the error signal e(k) in Figure 11 or Figure 12 .

[0356] In the case where the method 1400 is applied to the signal processing field not involving the secondary path, the second reference signal can be the first reference signal.

[0357] The length of the P segment sub-signal of the second reference signal is the same as the length of the P segment coefficient corresponding thereto.

[0358] Taking the method 1400 applied to the echo cancellation field as an example, the first reference signal can be a far-end signal, and the desired signal can be an echo signal. The output signal can be an estimated value of the echo signal output by the control filter.

[0359] It should be understood that the above is only an example, and the method 1400 does not constitute a limitation on the scheme of the embodiments of the present application, and the method 1400 can also be applied to other fields.

[0360] For example, the step 1404 includes: updating P segments of coefficients of the control filter respectively based on the inverse matrix of the autocorrelation matrix of at least one of the P segments of the second reference signal and P segments of the gradient of the mean square error of the error signal, or updating P segments of coefficients of the control filter respectively based on the inverse matrix of the autocorrelation matrix of at least one of the P segments of the first reference signal and P segments of the gradient of the mean square error of the error signal, the P segments of the first reference signal are segmented based on sampling time points of the first reference signal, and the P segments of the gradient of the mean square error of the error signal are determined according to the P segments of the second reference signal.

[0361] According to the scheme of the embodiments of the present application, the P segments of the coefficients of the control filter are updated respectively, that is, the P segments of the coefficients are updated independently, which can reduce the number of multiplication operations, reduce the amount of calculation in the signal processing process, reduce the demand for computing resources, and improve the efficiency of signal processing. For example, when the LMS Newton method is used to update the coefficients of the control filter, the inverse operation of the matrix needs to be performed, and the coefficients of the control filter are updated as a whole, so that the inverse matrix of the autocorrelation matrix of the second reference signal needs to be calculated. When the scheme of the embodiments of the present application is used, only the inverse matrix of the autocorrelation matrix of at least one segment of the signal needs to be calculated, which reduces the dimension of the autocorrelation matrix involved in the calculation, thereby reducing the calculation complexity of the inverse operation of the matrix, reducing the amount of calculation, reducing the demand for computing resources, and improving the efficiency of signal processing. It is beneficial to meet the real-time requirements in the signal processing process. For example, when the method 1400 is applied to the field of echo cancellation, the efficiency of echo cancellation can be improved, which is beneficial to meet the real-time requirements in the echo cancellation process and improve the user experience.

[0362] In a possible implementation, the length of the P segments of the coefficients is determined according to the correlation of the third reference signal, the third reference signal is determined according to the fourth reference signal, and the sampling time period of the fourth reference signal is not later than the sampling time period of the first reference signal. The fourth reference signal and the first reference signal come from the same signal source.

[0363] In the case where the method 1400 is applied to the field of signal processing not involving a secondary path, the third reference signal can be the fourth reference signal. For example, the third reference signal and the fourth reference signal can be the reference signal 2# in the following formula. Figure 11

[0364] ​In a possible implementation, the length of the P-section coefficients is determined according to the correlation of the third reference signal, including: the length of the P-section coefficients is determined according to the maximum value in the number of sampling points required for the autocorrelation coefficients and the cross-correlation coefficients of the multiple channels of the third reference signal to decay to the first threshold.

[0365] For example, the length of the P-section coefficients is greater than or equal to the maximum value in the number of sampling points required for the autocorrelation coefficients and the cross-correlation coefficients of the multiple channels of the third reference signal to decay to the first threshold.

[0366] The embodiments of the present application determine the length of the section coefficients of the control filter based on the correlation between signals, for example, determine the length of the section coefficients of the control filter according to the correlation of the filtered reference signal, and the correlation between different section sub-signals is weak, so that the convergence speed and signal processing effect can be ensured without being affected while reducing the amount of calculation. The scheme of the embodiments of the present application has similar convergence speed and signal processing effect as the LMS Newton algorithm, and the amount of calculation is much smaller than the LMS Newton algorithm, that is, the scheme of the embodiments of the present application can reduce the amount of calculation while ensuring the signal processing effect, reduce the demand for computing resources, and improve the signal processing efficiency.

[0367] In a possible implementation, the time interval between the sampling period of the fourth reference signal and the sampling period of the first reference signal is less than or equal to the second threshold.

[0368] The reference signal used to determine the length of each section coefficient of the control filter can be adjusted over time, that is, the length of each section coefficient of the control filter can be updated over time.

[0369] For example, the second threshold can be the time interval T in the foregoing, that is, the length of each section coefficient of the control filter is re-determined every time interval T.

[0370] In this way, the length of each section coefficient of the control filter can be dynamically adjusted according to the signal characteristics to improve the adaptability of the method, and the convergence speed of the update of the control filter can be ensured, that is, the optimal signal processing effect can be ensured.

[0371] In a possible implementation, the step 1404 can include: updating the pth section coefficient based on the error signal and the pth section sub-signal in the second reference signal, p=0,1,2…P-1.

[0372] Optionally, the P segments of the P segments of the coefficients of the control filter are respectively updated based on the inverse matrix of the autocorrelation matrix of the P segments of the second reference signal and the P segments of the gradient of the mean square error of the error signal, including: the pth segment of the P segments of the coefficients of the control filter is updated based on the pth segment of the P segments of the second reference signal and the pth segment of the gradient of the mean square error of the error signal; or, the P segments of the P segments of the coefficients of the control filter are respectively updated based on the inverse matrix of the autocorrelation matrix of the P segments of the first reference signal and the P segments of the gradient of the mean square error of the error signal, including: the pth segment of the P segments of the coefficients of the control filter is updated based on the pth segment of the P segments of the first reference signal and the pth segment of the gradient of the mean square error of the error signal.

[0373] In this way, the P segments of the coefficients of the control filter can be respectively updated based on the error signal and the P segments of the second reference signal.

[0374] In a possible implementation, the step 1404 can include: the P segments of the coefficients are respectively updated based on the error signal and any segment of the second reference signal.

[0375] Optionally, the P segments of the P segments of the coefficients of the control filter are respectively updated based on the inverse matrix of the autocorrelation matrix of the P segments of the second reference signal and the P segments of the gradient of the mean square error of the error signal, including: the pth segment of the P segments of the coefficients of the control filter is updated based on the pth segment of the P segments of the second reference signal and the pth segment of the gradient of the mean square error of the error signal; or, the P segments of the P segments of the coefficients of the control filter are respectively updated based on the inverse matrix of the autocorrelation matrix of the P segments of the first reference signal and the P segments of the gradient of the mean square error of the error signal, including: the pth segment of the P segments of the coefficients of the control filter is updated based on the pth segment of the P segments of the first reference signal and the pth segment of the gradient of the mean square error of the error signal.

[0376] In this way, the P segments of the coefficients of the control filter are updated based on the autocorrelation matrix of the same segment, which can further reduce the amount of calculation, reduce the demand for computing power, and improve the efficiency of signal processing.

[0377] The following describes the apparatus of the embodiments of the present application. Figures 15 to 17 The apparatus of the embodiments of the present application is described. It should be understood that the apparatus described below can perform the method of the foregoing embodiments of the present application, and to avoid unnecessary repetition, the description of the apparatus of the embodiments of the present application is appropriately omitted below.

[0378] Figure 15 is a schematic block diagram of the noise control apparatus of the embodiments of the present application. Figure 15The apparatus 3000 for noise control shown can be used to perform the method shown in Figure 5 、 Figure 7 、 Figure 9 、 Figure 10 or Figure 13 The apparatus 3000 includes a first obtaining unit 3010, a first processing unit 3020, a second obtaining unit 3030, a second processing unit 3040, and a third processing unit 3050.

[0379] The first obtaining unit 3010 is configured to obtain a first reference signal collected by a reference sensor.

[0380] The first processing unit 3020 is configured to filter the first reference signal by a control filter to obtain a control signal, the control signal being used to indicate a secondary noise signal.

[0381] The second obtaining unit 3030 is configured to obtain an error signal collected by an error sensor, the error signal being obtained by superimposing an original noise signal and the secondary noise signal.

[0382] The second processing unit 3040 is configured to determine a second reference signal according to the first reference signal.

[0383] The third processing unit 3050 is configured to update P segments of coefficients of the control filter based on at least one of P segments of the error signal and P segments of the second reference signal, the P segments of the second reference signal being obtained based on sampling time points of the second reference signal, the P segments of the coefficients of the control filter being obtained based on a length of the control filter, and P being a positive integer greater than 1.

[0384] Optionally, the third processing unit 3050 is specifically configured to update the P segments of the coefficients of the control filter based on P segments of an inverse matrix of an autocorrelation matrix of at least one of the P segments of the second reference signal and a gradient of a mean square error of the error signal, or update the P segments of the coefficients of the control filter based on P segments of an inverse matrix of an autocorrelation matrix of at least one of P segments of the first reference signal and a gradient of a mean square error of the error signal, the P segments of the first reference signal being obtained based on sampling time points of the first reference signal, and the P segments of the gradient of the mean square error of the error signal being determined according to the P segments of the second reference signal.

[0385] Optionally, a length of the P segments of the coefficients is determined according to a correlation of a third reference signal, the third reference signal being determined according to a fourth reference signal collected by the reference sensor, and a sampling time period of the fourth reference signal being no later than a sampling time period of the first reference signal.

[0386] Optionally, the length of the P segment coefficients is determined according to the correlation of the third reference signal, including: the length of the P segment coefficients is determined according to the maximum value of the sampling point numbers required for the autocorrelation coefficients and the cross-correlation coefficients of the multiple channels of the third reference signal to decay to a first threshold value.

[0387] Optionally, a time interval between a sampling period of the fourth reference signal and a sampling period of the first reference signal is less than or equal to a second threshold value.

[0388] Optionally, the third processing unit 3050 is specifically configured to update the P segment coefficients based on the error signal and any one of the P segment sub-signals in the second reference signal.

[0389] Optionally, the third processing unit 3050 is specifically configured to update the P segment coefficients based on the error signal and any one of the P segment sub-signals in the second reference signal.

[0390] Optionally, the second reference signal is obtained by filtering the first reference signal through a secondary path transfer function, and the third reference signal is obtained by filtering the fourth reference signal through the secondary path transfer function.

[0391] Optionally, the second reference signal is obtained by delaying the first reference signal by J sampling instants, J being a positive integer, J being determined according to the length of the secondary path transfer function, and the third processing unit 3050 is specifically configured to filter the error signal through the inverse secondary path transfer function to obtain a filtered error signal; and update the P segment coefficients of the control filter based on the filtered error signal and at least one of the P segment sub-signals in the second reference signal.

[0392] Figure 16 is a schematic block diagram of a signal processing device according to an embodiment of the present application. Figure 16 The signal processing device 4000 shown can be used to perform the method shown in Figure 5 , Figures 9 to 14 , Figure 17 The device 4000 includes a first acquisition unit 4010, a first processing unit 4020, a second acquisition unit 4030, and a second processing unit 4040.

[0393] The first acquisition unit 4010 is configured to acquire a first reference signal.

[0394] The first processing unit 4020 is configured to filter the first reference signal through a control filter to obtain a control signal, the control signal being used to indicate an output signal.

[0395] The second acquisition unit 4030 is configured to acquire an error signal, the error signal being obtained by superimposing a desired signal and the output signal.

[0396] The second processing unit 4040 is configured to update P segments of coefficients of the control filter based on at least one of P segments of the error signal and P segments of the second reference signal, the second reference signal being determined according to the first reference signal, the P segments of the second reference signal being segmented based on sampling instants of the second reference signal, and the P segments of the coefficients being segmented based on a length of the control filter, P being a positive integer greater than 1.

[0397] Optionally, the second processing unit 4040 is specifically configured to update the P segments of the coefficients of the control filter based on P segments of a gradient of a mean square error of the error signal and P segments of an inverse matrix of an autocorrelation matrix of at least one of the P segments of the second reference signal, or update the P segments of the coefficients of the control filter based on P segments of a gradient of a mean square error of the error signal and P segments of an inverse matrix of an autocorrelation matrix of at least one of P segments of the first reference signal, the P segments of the first reference signal being segmented based on sampling instants of the first reference signal, and the P segments of the gradient of the mean square error of the error signal being determined according to the P segments of the second reference signal.

[0398] Optionally, the length of the P segments of the coefficients is determined according to a correlation of a third reference signal, the third reference signal being determined according to a fourth reference signal, the fourth reference signal and the first reference signal being from a same signal source, and a sampling time period of the fourth reference signal being not later than a sampling time period of the first reference signal.

[0399] Optionally, the length of the P segments of the coefficients is determined according to the correlation of the third reference signal, and includes that the length of the P segments of the coefficients is determined according to a maximum value of a number of sampling points required for autocorrelation coefficients and cross-correlation coefficients of a plurality of channels of the third reference signal to decay to a first threshold value.

[0400] Optionally, a time interval between the sampling time period of the fourth reference signal and the sampling time period of the first reference signal is less than or equal to a second threshold value.

[0401] Optionally, the second processing unit 4040 is specifically configured to update a p-th segment of coefficients based on the error signal and a p-th segment of the second reference signal, p=0, 1, 2, …, P-1.

[0402] Optionally, the second processing unit 4040 is specifically configured to update the P segments of the coefficients based on any segment of the second reference signal and the error signal respectively.

[0403] It should be noted that the apparatus 3000 and the apparatus 4000 are embodied in the form of functional units. The term “unit” herein can be implemented by software and / or hardware, and is not limited in specific form.

[0404] For example, the "unit" can be a software program, hardware circuit, or a combination of both, which implements the above functions. The hardware circuit can include an application specific integrated circuit (ASIC), an electronic circuit, a processor (for example, a shared processor, a dedicated processor, or a group processor, etc.) and a memory for executing one or more software or firmware programs, a combination logic circuit, and / or other suitable components that support the described functions.

[0405] Therefore, the units of each example described in the embodiments of the present application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether the functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0406] The embodiments of the present application also provide a device, which comprises a processing unit and a storage unit, wherein the storage unit is used to store instructions, and the processing unit executes the instructions stored in the storage unit, so that the device executes the noise control method or the signal processing method in the above embodiments.

[0407] ​ A schematic block diagram of the device 5000 provided by the embodiments of the present application is shown, which comprises a memory 5001 used to store a computer program, and a processor 5002 used to execute the computer program stored in the memory, so that the device 5000 executes the noise control method or the signal processing method described above.

[0408] The embodiments of the present application also provide a terminal device, which can comprise the device 3000, the device 4000, or the device 5000.

[0409] Optionally, the terminal device can be a vehicle.

[0410] The embodiments of the present application also provide a computer readable medium, which stores program codes for execution by a device, and the program codes comprise codes for executing the noise control method or the signal processing method in the embodiments of the present application.

[0411] The embodiments of the present application also provide a computer program product comprising instructions which, when the computer program product is executed on a computer, cause the computer to execute the signal processing method in the embodiments of the present application.

[0412] It should be appreciated that the processor in the embodiments of the present application can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0413] It should also be understood that the memory in the embodiments of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memory. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically EPROM (EEPROM) or a flash memory. The volatile memory can be a random access memory (RAM) used as an external cache. By way of example, but not by way of limitation, many forms of random access memory (RAM) are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link dynamic random access memory (SLDRAM) and direct memory bus random access memory (DRAM) (DR RAM).

[0414] The above-described embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, the above-described embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another, for example, the computer instructions can be transferred from one website, computer, server, or data center to another via wired (for example, infrared, wireless, microwave, etc.) or wireless means. The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server, data center, or the like, which includes one or a set of available media. The available media can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0415] It should be understood that the term "and / or" herein merely describes an association relationship of associated objects, which means that there can be three relationships, for example, A and / or B can represent the following three cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents an "or" relationship between the associated objects before and after it, but it can also represent an "and / or" relationship. The specific meaning can be understood according to the context before and after it.

[0416] In this application, "at least one" means one or more, and "multiple" means two or more. "At least one of the following" or the like means any combination of the items, including any combination of single or multiple items. For example, at least one of a, b, or c can represent a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0417] It should be understood that in various embodiments of the present application, the size of the sequence number of the above-described processes does not mean the order of execution, and the execution order of the processes should be determined according to their functions and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0418] Those skilled in the art can clearly understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0419] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0420] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the above-described device embodiments are merely schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0421] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. According to actual needs, part or all of the units can be selected to achieve the purpose of the embodiment.

[0422] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit.

[0423] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0424] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for noise control, characterized in that, include: Acquire the first reference signal collected by the reference sensor; The first reference signal is filtered by a control filter to obtain a control signal, which is used to indicate secondary noise signals. An error signal acquired by an error sensor is obtained, wherein the error signal is obtained by superimposing the original noise signal and the secondary noise signal; The second reference signal is determined based on the first reference signal; The second reference signal is a filtered version of the first reference signal, or the second reference signal is a delayed version of the first reference signal; The P-segment coefficients of the control filter are updated based on at least one of the P-segment sub-signals of the error signal and the second reference signal. The P-segment sub-signals are obtained by segmenting based on the sampling time of the second reference signal, and the P-segment coefficients are obtained by segmenting based on the length of the control filter. P is a positive integer greater than 1.

2. The method according to claim 1, characterized in that, The updating of the P-segment coefficients of the control filter based on at least one sub-signal from the P-segment sub-signals of the error signal and the second reference signal includes: The P-segment coefficients of the control filter are updated based on the inverse matrix of the autocorrelation matrix of at least one sub-signal in the P-segment sub-signals of the second reference signal and the P-segment in the gradient of the mean square error of the error signal, respectively. Alternatively, the P-segment coefficients of the control filter are updated based on the inverse matrix of the autocorrelation matrix of at least one sub-signal in the P-segment sub-signals of the first reference signal and the P-segment in the gradient of the mean square error of the error signal, respectively. The P-segment sub-signals of the first reference signal are obtained by segmenting based on the sampling time of the first reference signal, and the P-segment in the gradient of the mean square error of the error signal is determined based on the P-segment sub-signals of the second reference signal.

3. The method according to claim 1 or 2, characterized in that, The length of the P-segment coefficient is determined based on the correlation of the third reference signal, which is determined based on the fourth reference signal acquired by the reference sensor. The sampling period of the fourth reference signal is no later than the sampling period of the first reference signal.

4. The method according to claim 3, characterized in that, The length of the P-segment coefficients is determined based on the correlation of the third reference signal, including: The length of the P-segment coefficients is determined based on the maximum value among the number of sampling points required for the autocorrelation coefficients and cross-correlation coefficients of multiple channels of the third reference signal to decay to the first threshold.

5. The method according to claim 3, characterized in that, The time interval between the sampling period of the fourth reference signal and the sampling period of the first reference signal is less than or equal to the second threshold.

6. The method according to claim 1 or 2, characterized in that, The updating of the P-segment coefficients of the control filter based on at least one sub-signal from the P-segment sub-signals of the error signal and the second reference signal includes: The coefficients of the p-th segment are updated based on the error signal and the p-th segment sub-signal in the second reference signal, where p = 0, 1, 2…P-1.

7. The method according to claim 1 or 2, characterized in that, The updating of the P-segment coefficients of the control filter based on at least one sub-signal from the P-segment sub-signals of the error signal and the second reference signal includes: The coefficients of segment P are updated based on any one of the sub-signals in the error signal and the second reference signal.

8. The method according to claim 5, characterized in that, The second reference signal is obtained by filtering the first reference signal through a secondary path transfer function, and the third reference signal is obtained by filtering the fourth reference signal through the secondary path transfer function.

9. The method according to claim 1 or 2, characterized in that, The second reference signal is obtained by delaying the first reference signal by J sampling times, where J is a positive integer, and J is determined based on the length of the secondary path transfer function. The updating of the P-segment coefficients of the control filter based on at least one sub-signal from the P-segment sub-signals of the error signal and the second reference signal includes: The error signal is filtered by the reversed secondary path transfer function to obtain the filtered error signal; The P-segment coefficients of the control filter are updated based on at least one of the P-segment sub-signals of the filtered error signal and the second reference signal.

10. A signal processing apparatus, characterized in that, include: The first acquisition unit is used to acquire the first reference signal collected by the reference sensor; The first processing unit is configured to filter the first reference signal using a control filter to obtain a control signal, the control signal being used to indicate a secondary noise signal; The second acquisition unit is used to acquire the error signal collected by the error sensor, wherein the error signal is obtained by superimposing the original noise signal and the secondary noise signal; The second processing unit is configured to determine the second reference signal based on the first reference signal; The third processing unit is used to update the P-segment coefficients of the control filter based on at least one segment of the P-segment sub-signals of the error signal and the second reference signal. The P-segment sub-signals are obtained by segmenting based on the sampling time of the second reference signal, and the P-segment coefficients are obtained by segmenting based on the length of the control filter, where P is a positive integer greater than 1.

11. The apparatus according to claim 10, characterized in that, The third processing unit is specifically used for: The P-segment coefficients of the control filter are updated based on the inverse matrix of the autocorrelation matrix of at least one sub-signal in the P-segment sub-signals of the second reference signal and the P-segment in the gradient of the mean square error of the error signal, respectively. Alternatively, the P-segment coefficients of the control filter are updated based on the inverse matrix of the autocorrelation matrix of at least one sub-signal in the P-segment sub-signals of the first reference signal and the P-segment in the gradient of the mean square error of the error signal, respectively. The P-segment sub-signals of the first reference signal are obtained by segmenting based on the sampling time of the first reference signal, and the P-segments in the gradient of the mean square error of the error signal are determined based on the P-segment sub-signals of the second reference signal.

12. The apparatus according to claim 10 or 11, characterized in that, The length of the P-segment coefficient is determined based on the correlation of the third reference signal, which is determined based on the fourth reference signal acquired by the reference sensor. The sampling period of the fourth reference signal is no later than the sampling period of the first reference signal.

13. The apparatus according to claim 12, characterized in that, The length of the P-segment coefficients is determined based on the correlation of the third reference signal, including: The length of the P-segment coefficients is determined based on the maximum value among the number of sampling points required for the autocorrelation coefficients and cross-correlation coefficients of multiple channels of the third reference signal to decay to the first threshold.

14. The apparatus according to claim 12, characterized in that, The time interval between the sampling period of the fourth reference signal and the sampling period of the first reference signal is less than or equal to the second threshold.

15. The apparatus according to claim 10 or 11, characterized in that, The third processing unit is specifically used for: The coefficients of the p-th segment are updated based on the error signal and the p-th segment sub-signal in the second reference signal, where p = 0, 1, 2…P-1.

16. The apparatus according to claim 10 or 11, characterized in that, The third processing unit is specifically used for: The coefficients of segment P are updated based on any one of the sub-signals in the error signal and the second reference signal.

17. The apparatus according to claim 14, characterized in that, The second reference signal is obtained by filtering the first reference signal through a secondary path transfer function, and the third reference signal is obtained by filtering the fourth reference signal through the secondary path transfer function.

18. The apparatus according to claim 10 or 11, characterized in that, The second reference signal is obtained by delaying the first reference signal by J sampling times, where J is a positive integer, and J is determined based on the length of the secondary path transfer function. The third processing unit is specifically used for: The error signal is filtered by the reversed secondary path transfer function to obtain the filtered error signal; The P-segment coefficients of the control filter are updated based on at least one of the P-segment sub-signals of the filtered error signal and the second reference signal.

19. A noise control device, characterized in that, It includes a processor and a memory, the memory being used to store program instructions, and the processor being used to invoke the program instructions to perform the method as described in any one of claims 1 to 9.

20. A computer-readable storage medium, characterized in that, The computer-readable medium stores program code for execution by the device, the program code including methods for performing any one of claims 1 to 9.

21. A computer program product containing instructions, characterized in that, When the computer program product is run on a computer, it causes the computer to perform the method as described in any one of claims 1 to 9.

Images