Active noise reduction system, method, apparatus, device, storage medium, and program product

Abstract

The application provides an active noise reduction system, method, device, computer readable storage medium and computer program product; the system comprises a reference sensor, a controller and a secondary loudspeaker, the controller comprises a plurality of sub-band filters; wherein the reference sensor is used for collecting a noise signal and sending the noise signal to the controller; the controller is used for extracting a space-time feature of the noise signal, determining a weight of each sub-band filter according to the space-time feature; generating a control filter according to a current filter parameter of the sub-band filter whose weight is a target value, generating an inverted sound wave signal used for canceling the noise signal according to a filtering result of the noise signal by the control filter, and sending the inverted sound wave signal to the secondary loudspeaker; wherein a residual signal after the inverted sound wave signal and the noise signal cancel each other is less than or equal to a preset signal threshold; and the secondary loudspeaker is used for playing the inverted sound wave signal. In this way, the noise reduction effect of the system can be improved.

Description

Technical Field

[0001] This application relates to the field of noise reduction technology, and in particular to an active noise reduction system, method, apparatus, computer-readable storage medium, and computer program product. Background Technology

[0002] Currently, intelligent vehicles can only achieve noise reduction by recognizing noise and automatically closing windows. However, this method already causes noise interference for users inside the vehicle, affecting their driving experience. Therefore, this application provides an active noise reduction system and method to solve the above problems. Summary of the Invention

[0003] This application provides an active noise cancellation system, method, apparatus, computer-readable storage medium, and computer program product.

[0004] The technical solution of this application embodiment is implemented as follows:

[0005] This application provides an active noise cancellation system, which includes a reference sensor, a controller, and a secondary speaker. The controller is connected to both the reference sensor and the secondary speaker, and includes multiple sub-band filters.

[0006] The reference sensor is used to collect noise signals and send the noise signals to the controller;

[0007] The controller is configured to extract the spatiotemporal features of the noise signal, determine the weights of each sub-band filter based on the spatiotemporal features, generate a control filter based on the current filter parameters of the sub-band filter with a target weight, generate an anti-phase acoustic signal to cancel the noise signal based on the filtering result of the control filter on the noise signal, and send the anti-phase acoustic signal to the secondary loudspeaker; wherein the residual signal after the anti-phase acoustic signal and the noise signal cancel each other out is less than or equal to a preset signal threshold.

[0008] The secondary speaker is used to play the anti-phase sound wave signal.

[0009] This application also provides an active noise cancellation method applied to a controller in an active noise cancellation system. The active noise cancellation system further includes a reference sensor and a secondary speaker. The controller is connected to both the reference sensor and the secondary speaker. The controller includes multiple sub-band filters. The method includes:

[0010] Acquire the noise signal collected by the reference sensor;

[0011] The spatiotemporal features of the noise signal are extracted, and the weights of each sub-band filter are determined based on the spatiotemporal features. A control filter is generated based on the current filter parameters of the sub-band filter with a weight of the target value. Based on the filtering result of the control filter on the noise signal, an anti-phase acoustic signal is generated to cancel the noise signal. The anti-phase acoustic signal is sent to the secondary speaker, so that the secondary speaker plays the anti-phase acoustic signal. The residual signal after the anti-phase acoustic signal and the noise signal cancel each other out is less than or equal to a preset signal threshold.

[0012] This application also provides an active noise cancellation device applied to an active noise cancellation system. The system includes a reference sensor, a controller, and a secondary speaker. The controller is connected to both the reference sensor and the secondary speaker, and includes multiple sub-band filters. The device includes:

[0013] The acquisition module is used to acquire the noise signal collected by the reference sensor;

[0014] The processing module is used to extract the spatiotemporal features of the noise signal, determine the weights of each sub-band filter based on the spatiotemporal features, generate a control filter based on the current filter parameters of the sub-band filter with a target weight, generate an anti-phase acoustic signal to cancel the noise signal based on the filtering result of the control filter on the noise signal, and send the anti-phase acoustic signal to the secondary speaker so that the secondary speaker plays the anti-phase acoustic signal; wherein, the residual signal after the anti-phase acoustic signal and the noise signal cancel each other out is less than or equal to a preset signal threshold.

[0015] This application also provides an electronic device, the device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the active noise reduction method as described in any of the preceding claims.

[0016] This application also provides a computer-readable storage medium storing computer-executable instructions or a computer program, which, when executed by a processor, implement the active noise reduction method as described in any of the preceding claims.

[0017] This application also provides a computer program product, including computer-executable instructions or a computer program, which, when executed by a processor, can implement the active noise reduction method as described in any of the preceding claims.

[0018] The embodiments of this application have the following beneficial effects:

[0019] The active noise reduction system provided in this application, after the controller receives the noise signal collected by the reference sensor, determines the weights of each sub-band filter by extracting the spatiotemporal features from the noise signal, and dynamically generates a control filter based on the current filter parameters of the sub-band filter with the target weight. The control filter is then used to generate an inverse acoustic wave signal to cancel the noise signal, thus achieving active noise reduction. Since the system can dynamically determine the weights of each sub-band filter and dynamically generate the control filter based on the noise signal, it can ensure that the control filter can accurately respond to changes in the noise environment, improving the noise reduction effect. Furthermore, since each sub-band filter can decompose the noise signal during the filtering process, the computational complexity can be reduced, making it better adaptable to various scenarios requiring noise control. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of a first structure of an active noise reduction system provided in an embodiment of this application;

[0021] Figure 2 This application provides a schematic diagram of a process for processing noise signals using a controller, as illustrated in an embodiment of the present application.

[0022] Figure 3 This is a schematic diagram of a second structure of the active noise reduction system provided in the embodiments of this application;

[0023] Figure 4 This is a schematic diagram of a third structure of the active noise reduction system provided in the embodiments of this application;

[0024] Figure 5 A flowchart illustrating an active noise reduction method provided in an embodiment of this application;

[0025] Figure 6 This is a schematic diagram of the active noise cancellation device provided in the embodiments of this application;

[0026] Figure 7 This is a schematic diagram of the active noise cancellation device provided in an embodiment of this application. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0029] In this application embodiment, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0030] Unless otherwise defined, all technical and scientific terms used in the embodiments of this application have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in the embodiments of this application is for descriptive purposes only and is not intended to limit the scope of this application.

[0031] This application provides an active noise cancellation system, method, apparatus, computer-readable storage medium, and computer program product.

[0032] Figure 1 This is a schematic diagram of a first structure of the active noise reduction system provided in the embodiments of this application, as shown below. Figure 1 As shown, the active noise cancellation system 1 includes a reference sensor 10, a controller 11, and a secondary speaker 12. The controller 11 is connected to both the reference sensor 10 and the secondary speaker 12. The connection method can be wired or wireless, and no specific limitation is made here.

[0033] Reference sensor 10 is used to collect noise signals and send the noise signals to controller 11;

[0034] The controller 11 is used to extract the spatiotemporal features of the noise signal, determine the weights of each sub-band filter based on the spatiotemporal features, generate a control filter based on the current filter parameters of the sub-band filter with the target weight, generate an anti-phase acoustic signal to cancel the noise signal based on the filtering result of the control filter on the noise signal, and send the anti-phase acoustic signal to the secondary loudspeaker; wherein, the residual signal after the anti-phase acoustic signal and the noise signal cancel each other out is less than or equal to a preset signal threshold.

[0035] Secondary speaker 12 is used to play anti-phase sound wave signals.

[0036] In this embodiment, the active noise cancellation system can be applied to vehicles or wearable devices that require low-frequency noise reduction. The following description will use a vehicle as an example.

[0037] For example, if an active noise cancellation system is applied to a vehicle, it can be used to reduce in-vehicle noise and improve the user's driving experience. The noise signal can be an external noise signal that requires noise reduction; that is, the primary noise source of the noise signal is located outside the vehicle. For example, the noise signal can be road noise, engine noise, etc., generated during vehicle operation, without specific limitations here.

[0038] For example, the reference sensor 10 may be an acceleration sensor used to collect noise signals and send the collected noise signals to the controller 11; here, the deployment location of the reference sensor 10 is not specifically limited. For example, the reference sensor 10 may be deployed on the chassis or at the attachment point between the vehicle body and the chassis.

[0039] In this embodiment, after receiving the noise signal sent by the reference sensor 10, the controller 11 extracts the spatiotemporal features of the noise signal and determines the weights of each sub-band filter based on the spatiotemporal features. Based on the current filter parameters of the sub-band filter with the target weight, a control filter is generated. Based on the filtering result of the control filter on the noise signal, an anti-phase acoustic wave signal for canceling the noise signal is generated and sent to the secondary speaker. The secondary speaker will play the signal after receiving the anti-phase acoustic wave signal.

[0040] For example, the controller 11 may include an active noise cancellation (ANC) unit, a deep neural network, a target filter, and multiple subband filters; for ease of understanding, the process of generating the antiphase acoustic signal described above will be illustrated below with reference to the various parts included in the controller 11.

[0041] The reference sensor is connected to the input of the deep neural network, the output of the deep neural network is connected to the input of the target filter, the output of the target filter is connected to the input of multiple sub-band filters, the output of the multiple sub-band filters is connected to the input of the ANC unit, and the output of the ANC unit is connected to the secondary speaker.

[0042] For example, a deep neural network can be a two-dimensional convolutional neural network (2D CNN) or a recurrent neural network (RNN), etc., without being specifically limited here; considering the computational complexity of different networks, the following explanation will use a 2D CNN as an example of a deep neural network.

[0043] In some embodiments, the controller is further configured to: preprocess the noise signal using a deep neural network, extract the spatiotemporal features of the preprocessed noise signal, classify the spatiotemporal features, and obtain a classification result; the classification result can characterize the weights of each subband filter.

[0044] Here, the deep neural network refers to a deep neural network that has already been trained. Understandably, during the training process of a deep neural network, preprocessing the sample noise signals in the training dataset can effectively improve the stability of the training process and the feature learning effect.

[0045] In this embodiment of the application, after receiving the noise signal sent by the reference sensor, the deep neural network can first preprocess the noise signal to obtain the preprocessed noise signal; here, the preprocessing may include at least one of normalization processing and frame segmentation processing.

[0046] Here, normalization can eliminate differences in the order of magnitude of noise signals. The specific method of normalization is not limited; for example, it could be extreme value normalization or other types of normalization. Framing can decompose the chaotic noise signal into multiple individual frames for parallel processing, which not only reduces computational complexity but also improves processing efficiency.

[0047] For example, if the preprocessing includes normalization and framing, the preprocessing process can be as follows: normalize the noise signal so that the amplitude of the normalized noise signal is in the range of [-1, 1], and framing the normalized noise signal to divide the continuous noise signal into multiple noise frames of a set length, each noise frame being frame data including the set frame length.

[0048] In this embodiment of the application, the network structure of the deep neural network is not specifically limited; for example... Figure 2 As shown, in the case of a 2D CNN (Convolutional Neural Network), the network structure of a 2D CNN can include several Conv2D convolutional layers, several batch normalization (BN) layers, and several activation layers (corresponding to...). Figure 2ReLU), average pooling layer, fully connected layer (corresponding to) Figure 2 (linear mapping in the middle) and output layer (corresponding to) Figure 2 The Sigmoid function in [the context of the text].

[0049] The convolutional layer extracts spatiotemporal features from the preprocessed noisy signal. The batch normalization (BN) layer accelerates the training process, improves training stability, and enhances the model's generalization ability. The activation layer introduces nonlinear transformations to enhance the model's expressive power. The average pooling layer reduces computational cost, lowers the dimensionality of spatiotemporal features, and prevents overfitting. The fully connected layer acts as a classifier in the entire convolutional neural network. After passing through deep networks such as convolutional layers, activation layers, and average pooling layers, the fully connected layer classifies the processed spatiotemporal features to obtain an initial classification result. The sigmoid function in the output layer then processes the initial classification result according to the task type (e.g., multi-label classification) to obtain the final classification result.

[0050] Here, the classification results can characterize the weights of each subband filter in the controller (corresponding to...). Figure 2 The weights are binary weights, with values ​​of 0 or 1. It should be noted that the weight values ​​indicate whether the corresponding subband filter is used to generate the subsequent control filter. For example, a weight of 1 means that the subband filter is used to generate the subsequent control filter, and a weight of 0 means that the subband filter is not used to generate the subsequent control filter. The meaning of the weight values ​​is related to the training process of the deep neural network.

[0051] In this embodiment, the subband filter with a weight of the target value refers to the subband filter used to generate the control filter; that is, if the weight is 1, it means that the subband filter is used to generate the subsequent control filter, then the target value is 1.

[0052] In this embodiment, the type of target filter is not specifically limited. For example, the target filter can be a Kalman filter (KF), an extended Kalman filter (EKF), etc. Considering the noise reduction efficiency, the following description will take KF as the target filter.

[0053] For example, see Figure 2The target filter KF is connected after the deep neural network 2D CNN. The classification result output by the deep neural network 2D CNN is input into the target filter KF. In this way, after obtaining the classification result, the target filter KF can select the sub-band filter used to generate the control filter, that is, the sub-band filter with the weight of the target value, according to the classification result. Then, the control filter is generated according to the current filter parameters of the sub-band filter with the weight of the target value.

[0054] Understandably, in order to achieve efficient dynamic filtering, embodiments of this application may employ pre-trained binary weighted subband filters; the process of determining the subband filters is as follows: a broadband filter is pre-trained, the broadband filter is decomposed, and an orthogonal basis is established using the time-domain subband filters to obtain multiple subband filters.

[0055] In one embodiment, a pre-trained broadband filter C can be generated using the Filtered-x Least Mean Square (FxLMS) algorithm in the frequency domain. The pre-trained control filter C is then decomposed into multiple sub-band filters {C1, C2, C3, ..., CM} using a Discrete Fourier Transform. Each sub-band filter covers a specific frequency range, enabling precise management of different frequency components of the noise signal and maintaining signal integrity. An orthogonal basis is then established using the time-domain sub-band control filters {S1, S2, S3, ..., SM}, resulting in multiple sub-band filters. These multiple sub-band filters provide the foundation for generating the control filter in the system.

[0056] For example, the process of generating a control filter based on the current filter parameters of the subband filter with the target weight can be as follows: combine the subband filters with the target weight based on the current filter parameters of the subband filter with the target weight to generate a control filter.

[0057] Here, the current filter parameters include at least the gain; the subband filters are combined by a convolution operation; for example, the subband filters with weights equal to the target value can be updated according to the gain of the subband filters with weights equal to the target value, and convolution operations can be performed on these updated subband filters to generate a control filter.

[0058] Furthermore, after obtaining the control filter, the noise signal can be filtered using the control filter to obtain the filtering result; here, the filtering result is the filtered noise signal; then, the filtered noise signal can be inverted using the ANC unit to obtain the inverted acoustic signal.

[0059] In this embodiment, when the active noise cancellation system is first started, the current filter parameters of each sub-band filter can be a fixed value preset based on experience. In subsequent steps, the current filter parameters of each sub-band filter can be adaptively adjusted using the feedback residual signal and the inverted acoustic wave signal to dynamically generate control filters and achieve more accurate noise reduction; the parameter adjustment process will be described later and will not be repeated here.

[0060] Here, the value of the preset signal threshold is not specifically limited. To achieve a good noise reduction effect, the preset signal threshold can be set according to the user's actual perception. In this embodiment, the residual signal after the anti-phase sound wave signal and the noise signal cancel each other out is less than or equal to the preset signal threshold. Theoretically, the active noise cancellation system aims to completely cancel out the anti-phase sound wave signal and the noise signal, that is, the residual signal value is 0. However, in actual operation, the probability of complete cancellation is low, and when the residual signal value is less than or equal to the preset signal threshold, the user in the vehicle can hardly perceive the residual signal. Therefore, when the residual signal is less than or equal to the preset signal threshold, the active noise cancellation system can achieve a good sound cancellation effect.

[0061] For example, see Figure 3 The noise reduction system may also include an error microphone 13 connected to the controller; the error microphone 13 is used to collect residual signals and feed the residual signals back to the controller 11; the controller is also used to adjust the current filter parameters of the subband filter with the target value as the weight if it is determined that the residual signal is greater than a preset signal threshold, so as to obtain the adjusted filter parameters; and update the current filter parameters to the adjusted filter parameters.

[0062] In this embodiment, the error microphone 13 is used to collect the residual signal after the anti-phase sound wave signal and the noise signal cancel each other out; the residual signal can characterize the noise situation after noise reduction in the vehicle; here, the deployment position of the error microphone 13 is not specifically limited. In order to ensure that the noise reduction requirements are met, the error microphone 13 can be placed in a position close to the human ear. For example, the error microphone 13 can be deployed around the seats in the vehicle or on the roof of the vehicle.

[0063] For example, the residual signal is also called the error signal. After the error microphone 13 collects the residual signal, it feeds the residual signal back to the controller 11. When the controller 11 receives the residual signal, it compares the residual signal with a preset signal threshold. If the residual signal is less than or equal to the preset signal threshold, it means that the current denoising effect of the system is good. At this time, active denoising can continue to be performed by the control filter generated according to the current filter parameters. Conversely, if the residual signal is greater than the preset signal threshold, it means that the current denoising effect of the system is poor. At this time, it is necessary to adjust the current filter parameters of the subband filter with the target value as the weight to obtain the adjusted filter parameters, and update the current filter parameters to the adjusted filter parameters. The process of adjusting the filter parameters is described below as an example.

[0064] In some embodiments, the controller is further configured to: if it is determined that the residual signal is greater than a preset signal threshold, obtain the residual signal and the antiphase acoustic signal through the target filter, and adjust the current filter parameters of the subband filter with a weight of the target value according to the residual signal and the antiphase acoustic signal to obtain the adjusted filter parameters.

[0065] Understandably, the residual signal can characterize the difference between the anti-phase acoustic signal and the noise signal. If the residual signal is determined to be greater than the preset signal threshold, it indicates that the difference between the anti-phase acoustic signal and the noise signal is large, and the current denoising effect of the system is not good. At this time, the controller can adjust the current filter parameters of the subband filter with the target value through the target filter.

[0066] For example, the process of adjusting the current filter parameters can be as follows: Obtain the residual signal and the anti-phase acoustic signal through the target filter; determine the desired output anti-phase acoustic signal based on the residual signal and the anti-phase acoustic signal; and adjust the current filter parameters of the sub-band filter with the target weight based on the desired output anti-phase acoustic signal to obtain the adjusted filter parameters. Subsequently, the control filter can be regenerated based on the adjusted filter parameters of the sub-band filter with the target weight; and the anti-phase acoustic signal used to cancel the noise signal can be generated again based on the filtering result of the regenerated control filter on the noise signal. It should be noted that the difference between the generated anti-phase acoustic signal and the noise signal should be smaller than the difference between the previously generated anti-phase acoustic signal and the noise signal.

[0067] For illustrative purposes, the adjustment process of filter parameters is illustrated using KF as an example, with the target filter as the example. KF can dynamically adjust the control filter to adapt to changes in the noise environment, and connecting it after a 2D CNN can further improve the system's generalization performance. Its working principle is as follows: Define the system's state transition equation and observation equation. The state transition equation describes the evolution of the system state, and the observation equation describes the relationship between the observed values ​​and the system state. Based on the residual signal and the anti-phase acoustic wave signal, predict and update the system state using the state transition equation and observation equation. Then, based on the predicted system state and covariance estimation, adjust the current filter parameters of the sub-band filter with the target weight to obtain the adjusted filter parameters, and update the current filter parameters to the adjusted filter parameters. Through the recursive characteristics of KF, the control filter is continuously optimized, enabling it to quickly respond to changes in noise characteristics and improve the accuracy and efficiency of noise reduction.

[0068] As can be seen from the embodiments of this application, the target filter can dynamically adjust the filter parameters of the subband filter according to the actual noise reduction situation, realize the dynamic generation of the control filter, and enable the generated control filter to quickly respond to changes in the noise environment and improve the noise reduction accuracy.

[0069] Furthermore, after obtaining the adjusted filter parameters, the controller is also used to: combine the sub-band filters with target weights according to the adjusted filter parameters to generate a control filter.

[0070] For example, after updating the current filter parameters to the adjusted filter parameters, the control filter can be regenerated based on the adjusted filter parameters of the subband filter with the target weight, and active denoising can be performed using the regenerated control filter. The process of regenerating the control filter is similar to the process of generating the control filter based on the current filter parameters last time, and will not be described again here to avoid repetition.

[0071] Figure 4 This is a schematic diagram of a third structure of the active noise reduction system provided in the embodiments of this application, as shown below. Figure 4As shown, the noise signal generated by the primary noise source outside the vehicle propagates to the reference sensor 10, which then sends the acquired noise signal to the controller 11. The controller 11 generates an anti-phase sound wave signal and sends it to the secondary speaker 12. The anti-phase sound wave signal played by the secondary speaker 12 propagates to the error microphone 13 via the secondary path. During propagation, the anti-phase sound wave signal cancels out the noise signal that propagates to the error microphone 13 via the primary path. At this time, the error microphone 13 acquires the residual signal after the anti-phase sound wave signal propagating via the secondary path cancels out the noise signal propagating via the primary path, and feeds back the residual signal to the controller 11 to optimize the control filter and achieve efficient and real-time active noise control.

[0072] Based on the same technical concept, this application also provides an active noise reduction method, which can be applied to the controller in the active noise reduction system provided in the foregoing embodiments.

[0073] Figure 5 This is a flowchart illustrating an active noise reduction method provided in an embodiment of this application, as shown below. Figure 5 As shown, the method includes the following steps:

[0074] Step 201: Obtain the noise signal collected by the reference sensor.

[0075] Step 202: Extract the spatiotemporal features of the noise signal, and determine the weights of each sub-band filter based on the spatiotemporal features; generate a control filter based on the current filter parameters of the sub-band filter with the target weight; generate an anti-phase acoustic signal to cancel the noise signal based on the filtering result of the control filter on the noise signal; send the anti-phase acoustic signal to the secondary speaker so that the secondary speaker plays the anti-phase acoustic signal; wherein, the residual signal after the anti-phase acoustic signal and the noise signal cancel each other out is less than or equal to a preset signal threshold.

[0076] In some embodiments, the controller includes a deep neural network; extracting the spatiotemporal features of the noise signal and determining the weights of each sub-band filter based on the spatiotemporal features may include: preprocessing the noise signal using a deep neural network, extracting the spatiotemporal features of the preprocessed noise signal, classifying the spatiotemporal features, and obtaining a classification result; the classification result can characterize the weights of each sub-band filter.

[0077] In some embodiments, the active noise cancellation system further includes an error microphone connected to the controller; the method may further include: acquiring the residual signal collected by the error microphone; if it is determined that the residual signal is greater than a preset signal threshold, adjusting the current filter parameters of the subband filter with a weight of the target value to obtain the adjusted filter parameters; and updating the current filter parameters to the adjusted filter parameters.

[0078] In some embodiments, the controller further includes a target filter. If it is determined that the residual signal is greater than a preset signal threshold, the current filter parameters of the sub-band filter with the target weight are adjusted. This adjustment may include: if it is determined that the residual signal is greater than the preset signal threshold, obtaining the residual signal and the anti-phase acoustic wave signal through the target filter, and adjusting the current filter parameters of the sub-band filter with the target weight based on the residual signal and the anti-phase acoustic wave signal to obtain the adjusted filter parameters.

[0079] In some embodiments, after obtaining the adjusted filter parameters, the above method may further include: combining subband filters with target weights according to the adjusted filter parameters to generate a control filter.

[0080] For the above method embodiments, since they are basically similar to the system embodiments, the relevant parts can be referred to the corresponding descriptions of the system embodiments. To avoid repeated descriptions, they will not be repeated here.

[0081] Based on the foregoing embodiments, this application also provides an active noise cancellation device, applied to a controller in an active noise cancellation system. Figure 6 This is a schematic diagram of the active noise cancellation device provided in the embodiments of this application, as shown below. Figure 6 As shown, the active noise cancellation device 6 may include:

[0082] The acquisition module 301 is used to acquire the noise signal collected by the reference sensor;

[0083] The processing module 302 is used to extract the spatiotemporal features of the noise signal, determine the weights of each sub-band filter based on the spatiotemporal features, generate a control filter based on the current filter parameters of the sub-band filter with the target weight, generate an anti-phase acoustic wave signal to cancel the noise signal based on the filtering result of the control filter on the noise signal, and send the anti-phase acoustic wave signal to the secondary speaker so that the secondary speaker plays the anti-phase acoustic wave signal; wherein, the residual signal after the anti-phase acoustic wave signal and the noise signal cancel each other out is less than or equal to a preset signal threshold.

[0084] In some embodiments, the controller includes a deep neural network; the processing module 302 is further configured to: preprocess the noise signal through the deep neural network, extract the spatiotemporal features of the preprocessed noise signal, classify the spatiotemporal features, and obtain a classification result; the classification result can characterize the weight of each subband filter.

[0085] In some embodiments, the active noise cancellation system further includes an error microphone connected to the controller; the processing module 302 is further configured to: acquire the residual signal collected by the error microphone; if it is determined that the residual signal is greater than a preset signal threshold, adjust the current filter parameters of the subband filter with a weight of the target value to obtain the adjusted filter parameters; and update the current filter parameters to the adjusted filter parameters.

[0086] In some embodiments, the controller further includes a target filter, and the processing module 302 is further configured to: if it is determined that the residual signal is greater than a preset signal threshold, obtain the residual signal and the anti-phase acoustic wave signal through the target filter, and adjust the current filter parameters of the sub-band filter with a weight of the target value according to the residual signal and the anti-phase acoustic wave signal to obtain the adjusted filter parameters.

[0087] In some embodiments, after obtaining the adjusted filter parameters, the processing module 302 is further configured to: combine the sub-band filters with target weights according to the adjusted filter parameters to generate a control filter.

[0088] In practical applications, both the acquisition module 301 and the processing module 302 can be implemented by a processor located in the active noise reduction device. The processor can be at least one of the following: Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), Central Processing Unit (CPU), controller, microcontroller, and microprocessor.

[0089] Based on the foregoing embodiments, this application also provides an active noise cancellation device. Figure 7 This is a schematic diagram of the active noise cancellation device provided in the embodiments of this application, as shown below. Figure 7 As shown, the active noise cancellation device 7 may include a processor 401 and a memory 402; wherein,

[0090] Memory 402 is used to store computer-executable instructions;

[0091] The processor 401, when executing computer-executable instructions or computer programs stored in the memory 402, implements an active noise reduction method as described above.

[0092] Based on the foregoing embodiments, this application also provides a computer-readable storage medium storing computer-executable instructions or computer programs, which, when executed by a processor, implement the active noise reduction method as described above.

[0093] Based on the foregoing embodiments, this application also provides a computer program product, including computer-executable instructions or a computer program, which, when executed by a processor, can implement the active noise reduction method as described above.

[0094] In some embodiments, the computer-readable storage medium may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM), etc.; or it may be a device that includes one or any combination of the above-mentioned memories.

[0095] In some embodiments, computer-executable instructions may take the form of programs, software, software modules, scripts, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as stand-alone programs or as modules, components, subroutines, or other units suitable for use in a computing environment.

[0096] As an example, computer-executable instructions may, but do not necessarily, correspond to files in a file system, and may be stored in files of other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple co-located files (e.g., files storing one or more modules, subroutines, or code sections).

Claims

1. An active noise reduction system, characterized in that, The system includes a reference sensor, a controller, and a secondary loudspeaker. The controller is connected to both the reference sensor and the secondary loudspeaker, and includes multiple sub-band filters. The reference sensor is used to collect noise signals and send the noise signals to the controller; The controller is configured to extract the spatiotemporal features of the noise signal, determine the weights of each sub-band filter based on the spatiotemporal features, generate a control filter based on the current filter parameters of the sub-band filter with a target weight, generate an anti-phase acoustic wave signal to cancel the noise signal based on the filtering result of the control filter on the noise signal, and send the anti-phase acoustic wave signal to the secondary loudspeaker; wherein the residual signal after the anti-phase acoustic wave signal and the noise signal cancel each other out is less than or equal to a preset signal threshold. The secondary speaker is used to play the anti-phase sound wave signal.

2. The system according to claim 1, characterized in that, The controller includes a deep neural network; the controller is further configured to: The noise signal is preprocessed using the deep neural network to extract the spatiotemporal features of the preprocessed noise signal, and the spatiotemporal features are classified to obtain the classification result; the classification result can characterize the weight of each subband filter.

3. The system according to claim 1 or 2, characterized in that, The system also includes an error microphone connected to the controller; The error microphone is used to collect the residual signal and feed the residual signal back to the controller; The controller is further configured to, if it is determined that the residual signal is greater than the preset signal threshold, adjust the current filter parameters of the subband filter with the target weight to obtain the adjusted filter parameters; and update the current filter parameters to the adjusted filter parameters.

4. The system according to claim 3, characterized in that, The controller further includes a target filter, and the controller is further configured to: If it is determined that the residual signal is greater than the preset signal threshold, the residual signal and the antiphase acoustic signal are obtained through the target filter, and the current filter parameters of the subband filter with the target weight are adjusted according to the residual signal and the antiphase acoustic signal to obtain the adjusted filter parameters.

5. The system according to claim 3 or 4, characterized in that, After obtaining the adjusted filter parameters, the controller is further configured to: Based on the adjusted filter parameters, the sub-band filters with weights equal to the target values ​​are combined to generate the control filter.

6. An active noise reduction method, characterized in that, A controller is used in an active noise cancellation system, the active noise cancellation system further including a reference sensor and a secondary speaker, the controller being connected to the reference sensor and the secondary speaker respectively, the controller including a plurality of sub-band filters, the method comprising: Acquire the noise signal collected by the reference sensor; The spatiotemporal features of the noise signal are extracted, and the weights of each sub-band filter are determined based on the spatiotemporal features. A control filter is generated based on the current filter parameters of the sub-band filter with a weight of the target value. Based on the filtering result of the control filter on the noise signal, an anti-phase acoustic signal is generated to cancel the noise signal. The anti-phase acoustic signal is sent to the secondary speaker, so that the secondary speaker plays the anti-phase acoustic signal. The residual signal after the anti-phase acoustic signal and the noise signal cancel each other out is less than or equal to a preset signal threshold.

7. The method according to claim 6, characterized in that, The controller includes a deep neural network; the step of extracting the spatiotemporal features of the noise signal and determining the weights of each sub-band filter based on the spatiotemporal features includes: The noise signal is preprocessed using the deep neural network to extract the spatiotemporal features of the preprocessed noise signal, and the spatiotemporal features are classified to obtain the classification result; the classification result represents the weight of each sub-band filter.

8. The method according to claim 6 or 7, characterized in that, The active noise cancellation system further includes an error microphone connected to the controller; the method further includes: Acquire the residual signal collected by the error microphone; If it is determined that the residual signal is greater than the preset signal threshold, the current filter parameters of the subband filter with the target weight are adjusted to obtain the adjusted filter parameters; the current filter parameters are then updated to the adjusted filter parameters.

9. The method according to claim 8, characterized in that, The controller further includes a target filter, wherein if it is determined that the residual signal is greater than the preset signal threshold, the current filter parameters of the sub-band filter with the target weight are adjusted, including: If it is determined that the residual signal is greater than the preset signal threshold, the residual signal and the antiphase acoustic signal are obtained through the target filter, and the current filter parameters of the subband filter with the target weight are adjusted according to the residual signal and the antiphase acoustic signal to obtain the adjusted filter parameters.

10. The method according to claim 8 or 9, characterized in that, After obtaining the adjusted filter parameters, the method further includes: Based on the adjusted filter parameters, the sub-band filters with weights equal to the target values ​​are combined to generate the control filter.

11. An active noise reduction device, characterized in that, A controller for use in an active noise cancellation system, the active noise cancellation system further including a reference sensor and a secondary speaker, the controller being connected to the reference sensor and the secondary speaker respectively, the controller including a plurality of sub-band filters, the device comprising: The acquisition module is used to acquire the noise signal collected by the reference sensor; The processing module is used to extract the spatiotemporal features of the noise signal, determine the weights of each sub-band filter based on the spatiotemporal features, generate a control filter based on the current filter parameters of the sub-band filter with a target weight, generate an anti-phase acoustic signal to cancel the noise signal based on the filtering result of the control filter on the noise signal, and send the anti-phase acoustic signal to the secondary speaker so that the secondary speaker plays the anti-phase acoustic signal; wherein, the residual signal after the anti-phase acoustic signal and the noise signal cancel each other out is less than or equal to a preset signal threshold.

12. An electronic device, characterized in that, The device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the active noise reduction method according to any one of claims 6 to 10.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the active noise reduction method as described in any one of claims 6 to 10.

14. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the active noise reduction method according to any one of claims 6 to 10.

Images