Noise reduction method, device and equipment and mobile carrier

Abstract

The embodiment of the invention provides a noise reduction method applied to a mobile carrier, and the method comprises the steps: continuously collecting a first environment noise signal through a first audio collection device, and inputting the first environment noise signal into a first filter to generate a first noise reduction signal; obtaining a noise reduction signal at the current moment, and outputting the noise reduction signal through audio signal output equipment; in the process that the audio signal output equipment continuously outputs the noise reduction signal, a first filter coefficient is generated according to the positions of the ears of the audience in the mobile carrier and the audio signal output equipment, and the first filter is adjusted. According to the noise reduction method provided by the embodiment of the invention, compared with a method for carrying out noise reduction calculation by using an accelerometer to detect the road bumping degree in the related technology, the system structure is simplified, elements on an existing carrier are reused, and the overall cost is reduced; a method of combining feedforward and feedback is adopted, so that the noise reduction capability is improved; in addition, according to the position of the audience, the output noise reduction signal can be adjusted in real time, and the stability of the noise reduction effect is maintained.

Description

Technical Field

[0001] The present disclosure relates to the field of acoustic technology, and in particular to a noise reduction method, device, equipment, and a mobile vehicle. Background Art

[0002] In the modern automotive industry, driving comfort is a key concern, and in-cabin noise levels are a key factor. Road noise generated by road surface excitation accounts for a significant portion of this. Existing active road noise control systems often rely on accelerometers. These sensors are typically installed in key locations, such as near the vehicle's suspension and on the vehicle body, to detect vibration excitation signals in real time. Based on the coherence between the vibration and the target noise, a feedforward adaptive method is then used to generate an inverse signal to offset the noise in the target area, thereby reducing road noise to a certain extent. However, in practical applications, these systems have drawbacks. Firstly, their structure is complex. Signal accuracy is significantly affected by the position and angle of the accelerometer's installation, requiring high installation precision and making subsequent maintenance difficult. Secondly, they are expensive. Accelerometers and their associated sensors are expensive, and the accumulation of multiple accelerometers significantly increases the total system cost, placing a heavy economic burden on large-scale production applications. In summary, while existing accelerometer-based active road noise control systems have shown some effectiveness, they struggle to meet demand due to structural and cost constraints. Summary of the Invention

[0003] The embodiments of the present disclosure provide a noise reduction method, device, equipment and mobile vehicle to solve the problem that existing vehicle noise reduction methods rely on accelerometers, require complex equipment, are difficult to maintain, and are costly.

[0004] Based on the above problems, in a first aspect, an embodiment of the present disclosure provides a noise reduction method, which is applied to a mobile vehicle, wherein the mobile vehicle is equipped with a first audio acquisition device, a second audio acquisition device, and an audio signal output device. The method includes:

[0005] Continuously collecting a first ambient noise signal through the first audio collection device and inputting the first ambient noise signal into a first filter to generate a first noise reduction signal;

[0006] superimposing the currently generated first noise reduction signal and the second noise reduction signal to obtain the noise reduction signal at the current moment, and outputting the signal through the audio signal output device;

[0007] The second noise reduction signal is obtained by performing noise reduction processing on the error audio signal, and the error audio signal is generated by superimposing the noise reduction signal output by the audio signal output device at a previous moment and the second ambient noise signal collected by the second audio collection device at that time;

[0008] While the audio signal output device continuously outputs the noise reduction signal, first filter coefficients are generated according to the positions of the listener's ears and the audio signal output device in the mobile vehicle, and the first filter is adjusted.

[0009] With reference to the first aspect, in one possible implementation, the first noise reduction signal and the first environmental noise signal have opposite phases and the same amplitudes;

[0010] The second noise reduction signal and the error audio signal have opposite phases and the same amplitudes.

[0011] With reference to the first aspect, in a possible implementation manner, noise reduction processing is performed on the error audio signal in the following manner to obtain the second noise reduction signal:

[0012] inputting the error audio signal into a second filter to generate the second noise reduction signal;

[0013] The method further comprises:

[0014] While the audio signal output device continuously outputs the noise reduction signal, the second filter coefficient of the second filter is determined according to the real-time speed of the mobile vehicle and the sound characteristics of the first ambient noise signal collected in real time, and the second filter is adjusted.

[0015] In conjunction with the first aspect, in one possible implementation, determining the second filter coefficient of the second filter according to the real-time speed of the mobile vehicle and the sound characteristics of the first ambient noise signal collected in real time includes:

[0016] Obtaining the real-time speed of the mobile vehicle and the sound characteristics of the first environmental noise signal collected in real time, and inputting them into a preset scene recognition model;

[0017] Determining, according to the scene information output by the preset scene recognition model, corresponding filter coefficients as the second filter coefficients;

[0018] Among them, the preset scene recognition model is a pre-trained neural network model that uses the speed information of the mobile vehicle and the sound characteristics of the ambient noise signal as input to predict the scene in which it is located; the filter coefficient is determined based on the pre-established correspondence between different scenes and filter coefficients.

[0019] In conjunction with the first aspect, in one possible implementation, the preset scene recognition model is trained in the following manner:

[0020] Constructing the preset scene recognition model;

[0021] Multiple groups of sample data containing speed information of mobile vehicles and sound characteristics of environmental noise are used as input, and the scene information corresponding to each group of sample data is used as the true value to train the preset scene recognition model to obtain the trained preset scene recognition model.

[0022] In conjunction with the first aspect, in one possible implementation, while the audio signal output device continuously outputs the noise reduction signal, generating first filter coefficients based on the positions of the listener's ear and the audio signal output device in the mobile vehicle, and adjusting the first filter include:

[0023] determining a transfer function according to a sound transmission characteristic of a secondary path between the audio signal output device and the second audio acquisition device;

[0024] During the process of the audio signal output device continuously outputting the noise reduction signal, the real-time determined ear position of the listener and the first ambient noise signal are input into the transfer function to output a reference signal;

[0025] Inputting the reference signal and the error audio signal into an adaptive filter to obtain a first filter coefficient of a first filter;

[0026] The first filter is adjusted using the first filter coefficients.

[0027] In combination with the first aspect, in a possible implementation, the listener's ear position is determined by a preset computer vision algorithm.

[0028] According to a second aspect of the embodiments of the present disclosure, a noise reduction device is provided, comprising: a noise reduction module and a head tracking module;

[0029] The noise reduction module includes: a first audio acquisition device, a second audio acquisition device and an audio signal output device;

[0030] The noise reduction module is configured to continuously collect a first ambient noise signal through the first audio acquisition device and input the signal into a first filter to generate a first noise reduction signal; superimpose the currently generated first noise reduction signal and the second noise reduction signal to obtain a noise reduction signal at a current moment, and output the signal through the audio signal output device; wherein the second noise reduction signal is obtained by performing noise reduction processing on an error audio signal, and the error audio signal is generated by superimposing the noise reduction signal outputted by the audio signal output device at a previous moment and the second ambient noise signal collected by the second audio acquisition device at that time;

[0031] The head tracking module is used to generate a first filter coefficient and adjust the first filter according to the position of the listener's ear and the audio signal output device in the mobile vehicle while the audio signal output device continuously outputs the noise reduction signal.

[0032] According to a third aspect of the embodiments of the present disclosure, a noise reduction device is provided, comprising the noise reduction apparatus described in the second aspect.

[0033] According to a fourth aspect of the embodiments of the present disclosure, a mobile vehicle is provided, comprising the noise reduction device described in the third aspect.

[0034] The beneficial effects of the embodiments of the present disclosure include:

[0035] Embodiments of the present disclosure provide a noise reduction method, apparatus, and device, as well as a mobile vehicle, which are applied to a mobile vehicle, wherein the mobile vehicle is equipped with a first audio acquisition device, a second audio acquisition device, and an audio signal output device. The method includes: continuously acquiring a first ambient noise signal through the first audio acquisition device, and inputting the signal into a first filter to generate a first noise reduction signal; superimposing the currently generated first noise reduction signal and the second noise reduction signal to obtain a noise reduction signal at the current moment, and outputting the signal through the audio signal output device; wherein the second noise reduction signal is obtained by performing noise reduction processing on an error audio signal, and the error audio signal is generated by superimposing the noise reduction signal outputted by the audio signal output device at a previous moment and the second ambient noise signal acquired by the second audio acquisition device at that time; while the audio signal output device continuously outputs the noise reduction signal, a first filter coefficient is generated according to the position of the listener's ear and the audio signal output device in the mobile vehicle, and the first filter is adjusted.

[0036] A noise reduction method provided by an embodiment of the present disclosure simplifies the system structure, reuses components on existing vehicles, and reduces overall costs compared to the method of using an accelerometer to detect the degree of road bumps for noise reduction calculation in related technologies; adopts a method that combines feedforward and feedback to improve the noise reduction capability; the embodiment of the present disclosure can also adjust the output noise reduction signal in real time according to the position of the listener to maintain the stability of the noise reduction effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A flowchart of a noise reduction method provided in an embodiment of the present disclosure;

[0038] Figure 2 A schematic structural diagram of a noise reduction device provided in an embodiment of the present disclosure;

[0039] Figure 3 This is an example of an implementation scenario of the noise reduction method and noise reduction device provided in the embodiments of the present disclosure. DETAILED DESCRIPTION

[0040] The embodiments of the present disclosure provide a noise reduction method, apparatus, device, and mobile vehicle. Preferred embodiments of the present disclosure are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are intended only to illustrate and explain the present disclosure and are not intended to limit the present disclosure. Furthermore, the embodiments and features of the embodiments in this application may be combined with each other unless there is a conflict.

[0041] The embodiment of the present disclosure provides a noise reduction method, such as Figure 1 As shown, the method is applied to a mobile vehicle, which is equipped with a first audio acquisition device, a second audio acquisition device, and an audio signal output device. The method includes:

[0042] S101: continuously collect a first ambient noise signal through a first audio collection device, and input the signal into a first filter to generate a first noise reduction signal;

[0043] S102: Superimpose the currently generated first noise reduction signal and the second noise reduction signal to obtain the current noise reduction signal, and output the signal through an audio signal output device;

[0044] The second noise reduction signal is obtained by performing noise reduction processing on the error audio signal, and the error audio signal is generated by superimposing the noise reduction signal output by the audio signal output device at the previous moment and the second ambient noise signal collected by the second audio collection device at that time;

[0045] S103 . While the audio signal output device continuously outputs the noise reduction signal, generate first filter coefficients according to the positions of the listener's ears and the audio signal output device in the mobile vehicle, and adjust the first filter.

[0046] The noise reduction method provided in the embodiments of the present disclosure can be applied to mobile vehicles. It is based on the principle of destructive interference of sound waves. By continuously executing the above-mentioned noise reduction signal generation process, it continuously generates sound waves opposite to the noise to offset the noise in the vehicle, thereby reducing the noise heard by the passengers in the passenger compartment of the mobile vehicle.

[0047] The first audio acquisition device can be a microphone for collecting ambient noise, which can be used to collect the main noise signals during the movement of the mobile vehicle. The mobile vehicle here can be a manned vehicle. During the actual movement of the vehicle, various types of ambient noise are generated, including engine noise, road noise, exhaust system noise, and wind noise.

[0048] The first audio collection device can be specifically positioned near different noise sources for different noise types. For example, for collecting engine noise, the first audio collection device can be placed in the engine compartment; for collecting wind noise, the first audio collection device can be placed on the windward side of the vehicle; for collecting road noise, the first audio collection device can be placed on the vehicle chassis near the tires.

[0049] According to actual needs, the number of first audio collection devices set up in the mobile vehicle can be determined according to the type of noise that needs to be targeted and eliminated. If multiple types of noise need to be eliminated, multiple first audio collection devices can be set up accordingly to collect multiple environmental noises for targeted noise reduction.

[0050] The original ambient noise signal of the first audio acquisition device is processed and input into the noise reduction algorithm of the first filter to generate a first noise reduction signal. The processing of the original ambient noise signal here can be converted into an electrical signal by a microphone (i.e., the first audio acquisition device), and then converted into a digital signal by an analog-to-digital converter, so that the first filter can process the signal. Among them, the execution hardware body for generating the noise reduction signal in the embodiment of the present disclosure can be a digital signal processor (DSP, Digital Signal Processor) carried by a mobile vehicle. The first filter generates a corresponding first noise reduction signal based on the first ambient noise signal.

[0051] Furthermore, the currently generated first and second noise reduction signals are superimposed to generate the current noise reduction signal. After the digital signal is converted to an analog signal via a digital-to-analog converter, the corresponding audio signal is output via an audio signal output device to cancel the corresponding ambient noise signal. The audio signal output device may be a speaker installed in the passenger compartment of a mobile vehicle. For example, in a manned vehicle, the audio signal output device may be an audio system including headrest speakers.

[0052] The second noise reduction signal here is obtained by performing noise reduction algorithm processing on the error audio signal; the error audio signal can be the noise reduction signal output by the audio signal output device at the previous moment and the audio signal collected by the second audio collection device after the second ambient noise signal is superimposed.

[0053] The signal noise reduction processing in the embodiment of the present disclosure is a cyclic process. The first audio acquisition device continuously acquires the first environmental noise signal and inputs it into the first filter. For the first noise reduction signal obtained for the first time, since the second noise reduction signal has not yet been generated as feedback, in this case, the first noise reduction signal can be output through the audio signal output device. The output noise reduction signal is superimposed on the second environmental noise signal collected by the second audio device to obtain an error audio signal. The error audio signal is then subjected to noise reduction processing to obtain a second noise reduction signal. Since the first noise reduction signal is continuously generated by continuously noise reduction processing of the continuously acquired first environmental signal, a new first noise reduction signal is also obtained while obtaining the second noise reduction signal. The new first noise reduction signal and the currently obtained second noise reduction signal are superimposed to obtain the noise reduction signal at the current moment, which is then output through the audio signal output device to continue to generate a new error audio signal. The above process is repeated to achieve a combination of feedforward and feedback, thereby improving the noise reduction capability.

[0054] Among them, the second audio acquisition device can be a microphone, which is used to collect the residual noise signal (i.e., the error audio signal) after noise reduction in the passenger compartment of the mobile vehicle, and feed the residual noise signal back into the noise reduction system to improve the noise reduction effect.

[0055] The second audio collection device is typically placed near the passenger compartment to simulate the vehicle's passenger's audio signal reception. For example, in a manned vehicle, the second audio collection device can be placed near the passenger's headrest or on the roof of the vehicle.

[0056] Furthermore, in actual application, the noise heard by passengers in a mobile vehicle may be uneven due to changes in the passenger position, head direction, or vehicle motion state. In the disclosed embodiment, the ear position of the passenger (i.e., the listener) in the vehicle can be tracked in real time by a sensor, the noise propagation path can be calculated in real time, and the filter coefficients of the noise reduction filter (i.e., the first filter and the second filter) can be adjusted to adjust the generated noise reduction signal to ensure that the noise reduction effect at the passenger's ear does not change due to reasons such as the movement of the passenger's position. The sensor here can be an existing sensor in the passenger compartment of a mobile vehicle, such as a driver monitoring system (DMS).

[0057] This disclosed embodiment utilizes the mobile vehicle's existing microphones (i.e., the first and second audio acquisition devices) and speaker system (i.e., the audio signal output device), while also reusing the vehicle's existing power amplifier equipment. Active noise reduction can be achieved by simply adding certain computing devices (such as the aforementioned DSP), significantly reducing system costs. This disclosed embodiment also utilizes head tracking technology to ensure the stability of the noise reduction effect even when the vehicle's occupants move, thereby improving system reliability.

[0058] In another embodiment provided by the present disclosure, the first noise reduction signal and the first environmental noise signal have opposite phases and the same amplitude;

[0059] The second noise reduction signal and the error audio signal have opposite phases and the same amplitudes.

[0060] In the embodiment of the present disclosure, a noise reduction signal may be generated based on the following formula (1).

[0061]

[0062] Where L represents the sampling period range of the filter; y(n) represents the noise reduction signal generated at period n; w k (n) represents the filter coefficient of the k-th order filter at cycle n; x(nk) represents the noise signal before the k-th cycle of the input. The noise signal here can be the original ambient noise signal directly obtained from the noise source, or it can be a noise signal estimated based on the noise source and the noise propagation path. Here, the timing characteristics of the historical noise signal values ​​are captured to generate a precise cancellation signal.

[0063] In another embodiment provided by the present disclosure, the error audio signal is subjected to noise reduction processing in the following manner to obtain a second noise reduction signal:

[0064] inputting the error audio signal into a second filter to generate a second noise reduction signal;

[0065] The above method further includes:

[0066] Step 1: While the audio signal output device continuously outputs the noise reduction signal, determine the second filter coefficient of the second filter according to the real-time speed of the mobile vehicle and the sound characteristics of the first ambient noise signal collected in real time, and adjust the second filter.

[0067] In the embodiment of the present disclosure, the error audio signal is input into the second filter, and the second filter can generate a corresponding second noise reduction signal based on the error audio signal. The ideal error audio signal should be the environmental sound with all noise signals eliminated. However, in actual applications, the noise part in the second environmental noise signal cannot be completely eliminated by the noise reduction signal, and a part of the noise is still retained in the second environmental noise signal (i.e., the error audio signal) after superimposing the noise reduction signal. The second filter generates a corresponding second noise reduction signal for this part of the noise content, and after superimposing it with the currently generated first noise reduction signal, a new current noise reduction signal can be obtained by outputting it through the audio signal output device. Through this method of combining feedback and feedforward, the processing capability of the noise signal is stronger, and a better noise reduction effect can be provided. Compared with the method of relying solely on feedforward noise reduction in the related art, the noise reduction capability is significantly improved.

[0068] Furthermore, in actual applications, the error audio signal collected by the second audio acquisition device may include normal ambient sounds within the passenger compartment of a mobile vehicle (e.g., conversations between passengers and music). To preserve these sounds and prevent them from being offset as noise signals, the second filter can selectively process the corresponding sound signal frequency band, processing only the noise-dominant frequency band while preserving the voice band. Alternatively, a corresponding prefilter can be added before the second filter to pre-separate the voice content from the noise content.

[0069] Furthermore, most noise generated by mobile vehicles during motion follows a relatively linear pattern, with regular patterns to follow. However, some noise is significantly affected by environmental factors, and the waveforms of the noise generated under different circumstances will have different characteristics. Therefore, addressing this noise, which is affected by various environmental factors, is also a key aspect of mobile vehicle noise reduction systems. For example, in manned vehicles, engine noise, exhaust noise, and wind noise generally show a significant correlation with driving speed; however, road noise is affected by a variety of factors, including driving speed, road conditions, weather conditions, and tire conditions.

[0070] In the disclosed embodiment, a special preset scene recognition model is set up to target this part of the noise that is significantly affected by environmental factors. The corresponding scene is determined by the sound characteristics of the collected first environmental noise signal. Different second filter coefficients are pre-set for different scenes. When it is detected that the mobile vehicle is in the corresponding scene, the second filter coefficient of the second filter is changed to correspond to the current scene, so that the second filter can improve the frequency domain response characteristics of the corresponding noise signal and enhance the noise reduction capability.

[0071] The scene recognition method provided by the disclosed embodiments can identify the current scene and automatically adjust noise reduction parameters (i.e., filter coefficients) based on real-time speed and noise characteristics input, thereby improving the system's adaptability and noise reduction effectiveness. It can utilize the vehicle's existing speed sensor, eliminating the need for additional accelerometers, simplifying the system architecture and reducing maintenance.

[0072] In another embodiment provided by the present disclosure, the above step 1 of “determining the second filter coefficient of the second filter based on the real-time speed of the mobile vehicle and the sound characteristics of the first ambient noise signal collected in real time” can be performed as follows:

[0073] Step 1: Obtain the real-time speed of the mobile vehicle and the sound characteristics of the first environmental noise signal collected in real time, and input them into a preset scene recognition model;

[0074] Step 2: Determine corresponding filter coefficients as second filter coefficients based on scene information output by a preset scene recognition model;

[0075] Among them, the preset scene recognition model is a pre-trained neural network model that uses the speed information of the mobile vehicle and the sound characteristics of the environmental noise signal as input to predict the scene in which it is located; the filter coefficient is determined based on the pre-established correspondence between different scenes and filter coefficients.

[0076] In the disclosed embodiment, the scene recognition operation may be performed by an onboard computing device mounted on a mobile vehicle. The computing device obtains a first ambient noise signal collected by a first audio collection device at the current moment and the real-time speed of the mobile vehicle at the current moment, and inputs the above two items into a preset scene recognition model.

[0077] The preset scene recognition model here can be a neural network model, which can take the speed of the mobile vehicle and the sound characteristics of the noise signal as input, predict the scene in which the current mobile vehicle is located, and output it.

[0078] In another embodiment provided by the present disclosure, the preset scene recognition model is trained in the following manner:

[0079] Step 1: Build a preset scene recognition model;

[0080] Step 2: Take multiple groups of sample data containing the speed information of the mobile vehicle and the sound characteristics of the ambient noise as input, take the scene information corresponding to each group of sample data as the true value, train the preset scene recognition model, and obtain the trained preset scene recognition model.

[0081] In the disclosed embodiment, a convolutional neural network can be trained. During the training process, the input layer of the neural network receives vehicle speed data and sound feature vectors, and the output layer classifies predefined scenes by labels. The collected noise data can be subjected to short-time Fourier transform and Mel Frequency Cepstral Coefficient (MFCC) analysis to extract the sound feature vectors.

[0082] Different mobile vehicles have different predefined scenarios. For example, for a passenger vehicle, predefined scenarios can include road conditions and road surface materials. Predefined scenario labels might include dry asphalt road, wet cement road, or snowy dirt road. These are examples only and are not intended to be limiting.

[0083] In another embodiment provided by the present disclosure, the aforementioned step S103 of "generating first filter coefficients and adjusting the first filter based on the positions of the listener's ear and the audio signal output device within the mobile vehicle while the audio signal output device continuously outputs the noise reduction signal" may be performed as follows:

[0084] Step 1: determining a transfer function based on the sound transmission characteristics of the secondary path between the audio signal output device and the second audio acquisition device;

[0085] Step 2: while the audio signal output device continuously outputs the noise reduction signal, the real-time determined ear position of the listener and the first ambient noise signal are input into the transfer function to output a reference signal;

[0086] Step 3: inputting the reference signal and the error audio signal into an adaptive filter to obtain a first filter coefficient of the first filter;

[0087] Step 4: Use the first filter coefficient to adjust the first filter.

[0088] In the disclosed embodiments, the sound propagation path (i.e., the secondary path) between the audio signal output device and the second audio capture device is affected by the physical propagation medium (e.g., air and conductive structures), thereby influencing the amplitude, phase, and frequency of the noise reduction signal. Ignoring the influence of the secondary path during noise reduction signal generation may result in a phase deviation between the noise reduction signal and the noise signal when the noise reduction signal reaches the target point (i.e., the listener's ear or the second audio capture device), thereby reducing the noise reduction effect.

[0089] In the disclosed embodiments, a transfer function is used to simulate the sound propagation characteristics of a secondary path. A first ambient noise signal processed by the transfer function is input as a reference signal into an adaptive filter. Combined with the error audio signal, the first filter coefficient of the first filter is updated to eliminate the influence of the physical medium on sound propagation, thereby improving the noise reduction effect. The transfer function can be determined using methods such as a frequency sweep method or a white noise method.

[0090] The adaptive filter can use the following formula (2) to update the filter coefficients.

[0091] w(n+1)=w(n)+μ×e(n)×x f (n) (2)

[0092] Where w(n) represents the filter coefficient at period n; μ represents the step size factor, and by changing the value of the step size factor, the update amplitude of the filter coefficient can be changed; e(n) represents the error audio signal at period n; x f (n) represents the first environmental noise signal at period n after being processed by the transfer function; w(n+1) represents the filter coefficient at period n+1 (ie, the updated filter coefficient).

[0093] Similarly, the movement of passengers in a mobile vehicle can alter the sound propagation path between the audio signal output device and the ear. This prevents the noise reduction signal received by the ear from canceling out the noise signal, reducing the noise reduction effect. Based on the same principle, the transfer function can be adjusted based on passenger movement, thereby changing the filter coefficients to compensate for the noise reduction effect. An example illustrates this compensation process. First, a camera in the passenger compartment of the mobile vehicle maps the listener's ear position to a predetermined physical propagation model of the secondary path. This physical propagation model contains pre-stored transfer functions for different locations. The camera captures the listener's ear position in real time based on computer vision and other calculations, and adjusts the ear position in real time within the physical propagation model. Finally, the closest pre-stored transfer function can be selected based on the listener's real-time ear position, or a current transfer function can be generated using an interpolation algorithm. By changing the transfer function, the first filter coefficient of the first filter is modified, thereby compensating for the noise reduction effect.

[0094] In another embodiment provided by the present disclosure, the ear position of the listener is determined by a preset computer vision algorithm.

[0095] The present disclosure also provides a noise reduction device, such as Figure 2 As shown, it includes: a noise reduction module 210 and a head tracking module 220;

[0096] The noise reduction module 210 includes: a first audio collection device, a second audio collection device, and an audio signal output device; the noise reduction module 210 is configured to continuously collect a first ambient noise signal through the first audio collection device, and input the signal into a first filter to generate a first noise reduction signal; superimpose the currently generated first noise reduction signal and the second noise reduction signal to obtain a noise reduction signal at the current moment, and output the signal through the audio signal output device; wherein the second noise reduction signal is obtained by performing noise reduction processing on an error audio signal, and the error audio signal is generated by superimposing the noise reduction signal outputted by the audio signal output device at a previous moment and the second ambient noise signal collected by the second audio collection device at that time; wherein the first noise reduction signal and the first ambient noise signal have opposite phases and the same amplitudes; and the second noise reduction signal and the error audio signal have opposite phases and the same amplitudes.

[0097] The head tracking module 220 is used to generate first filter coefficients and adjust the first filter according to the positions of the listener's ears and the audio signal output device in the mobile vehicle while the audio signal output device continuously outputs the noise reduction signal.

[0098] In another embodiment provided by the present disclosure, the noise reduction module 210 is further configured to input the error audio signal into a second filter to generate a second noise reduction signal;

[0099] The noise reduction device provided by the embodiment of the present disclosure further includes a scene recognition module 230;

[0100] A scene recognition module 230 is configured to determine a second filter coefficient of a second filter based on the real-time speed of the mobile vehicle and the sound characteristics of the first ambient noise signal collected in real time, and adjust the second filter while the audio signal output device continuously outputs the noise reduction signal;

[0101] In another embodiment provided by the present disclosure, the scene recognition module 230 is further configured to obtain the real-time speed of the mobile vehicle and the sound characteristics of the first ambient noise signal collected in real time, and input them into a preset scene recognition model; and determine corresponding filter coefficients as second filter coefficients based on the scene information output by the preset scene recognition model; wherein the preset scene recognition model is a pre-trained neural network model that uses the speed information of the mobile vehicle and the sound characteristics of the ambient noise signal as input to predict the scene in which it is located; and the filter coefficients are determined based on a pre-established correspondence between different scenes and the filter coefficients;

[0102] Among them, the preset scene recognition model is trained using the following method: constructing a preset scene recognition model; taking multiple groups of sample data containing the speed information of the mobile vehicle and the sound characteristics of the environmental noise as input, and taking the scene information corresponding to each group of sample data as the true value, training the preset scene recognition model, and obtaining the trained preset scene recognition model.

[0103] In another embodiment provided by the present disclosure, the head tracking module 220 is further used to determine a transfer function based on the sound transmission characteristics of the secondary path between the audio signal output device and the second audio acquisition device; while the audio signal output device continuously outputs the noise reduction signal, the real-time determined ear position of the listener and the first ambient noise signal are input into the transfer function to output a reference signal; the reference signal and the error audio signal are input into an adaptive filter to obtain a first filter coefficient of the first filter; and the first filter coefficient is used to adjust the first filter; wherein the ear position of the listener is determined by a preset computer vision algorithm.

[0104] The noise reduction module 210 in the embodiment of the present disclosure may include: a first audio acquisition device, a second audio acquisition device, an audio signal output device, and a computing device carried by a mobile vehicle; wherein the computing device may be the executor of the first filter, the second filter, the adaptive filter, and the transfer function.

[0105] The head tracking module 220 may include a camera device for identifying the positions of the listener's ears and a computer device for calculating position data.

[0106] The scene recognition module 230 may include a computer device that stores a scene recognition model and performs scene recognition.

[0107] Here is an example of an implementation scenario of the noise reduction method and noise reduction device provided in the embodiments of the present disclosure, such as Figure 3 shown.

[0108] The first audio capture device 211 continuously collects a first ambient noise signal, converts it into analog-to-digital signals, and then inputs it into the first filter 214, the transfer function 217, and the scene recognition module 230. The first filter 214 applies a noise reduction algorithm to the input first ambient noise signal and outputs a first noise reduction signal. This signal is then superimposed with the second noise reduction signal output by the second filter 215 to obtain the current noise reduction signal. The audio signal output device 213 obtains the noise reduction signal, performs digital-to-analog conversion on it, and outputs the noise reduction signal for noise reduction. After being transmitted through a secondary path, the noise reduction signal is superimposed with the second ambient noise signal at the second audio capture device 212. The superimposed signal is collected by the second audio capture device 212 to obtain an error audio signal. The error audio signal is then input into the second filter 215 and the adaptive filter 216. The second filter 215 generates a second noise reduction signal based on the error audio signal and outputs the second noise reduction signal. The adaptive filter 216 obtains the error audio signal and the reference signal obtained by the transfer function 217 respectively. According to the two signals, the adaptive filter 216 updates the first filter coefficient of the first filter 214 through an updating algorithm to provide a better noise reduction effect.

[0109] The scene recognition module 230 determines the current scene of the mobile vehicle based on the real-time speed and the first environmental noise signal, and updates the second filter coefficient for the second filter 216 based on the predetermined correspondence between the scene and the second filter coefficient to adapt to the noise characteristics of different environments.

[0110] The head tracking module 220 obtains the listener's ear position in real time through the camera, generates compensation data, and adjusts the transfer function 217 to change the first filter coefficient, thereby compensating the noise reduction signal and ensuring that the noise reduction effect is not reduced due to changes in the listener's position.

[0111] An embodiment of the present disclosure further provides a noise reduction device, comprising the noise reduction device in the above embodiment.

[0112] An embodiment of the present disclosure also provides a mobile vehicle, comprising the noise reduction device in the above embodiment.

[0113] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments of the present disclosure can be implemented through hardware or by means of software plus a necessary general hardware platform. Based on this understanding, the technical solutions of the embodiments of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present disclosure.

[0114] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes in the accompanying drawings are not necessarily required for implementing the present disclosure.

[0115] Those skilled in the art will appreciate that the modules in the devices of the embodiments may be distributed in the devices of the embodiments as described in the embodiments, or may be located in one or more devices different from the embodiments with corresponding changes. The modules of the above embodiments may be combined into one module or further split into multiple submodules.

[0116] The serial numbers of the above-mentioned embodiments of the present disclosure are for description only and do not represent the advantages or disadvantages of the embodiments.

[0117] Obviously, those skilled in the art may make various changes and modifications to the present disclosure without departing from the spirit and scope of the present disclosure. Thus, if these modifications and variations of the present disclosure fall within the scope of the claims of the present disclosure and their equivalents, the present disclosure is intended to include these modifications and variations.

Claims

1. A noise reduction method, characterized in that: Applied to a mobile vehicle, the mobile vehicle is equipped with a first audio acquisition device, a second audio acquisition device, and an audio signal output device, and the method includes: Continuously collecting a first ambient noise signal through the first audio collection device and inputting the first ambient noise signal into a first filter to generate a first noise reduction signal; superimposing the currently generated first noise reduction signal and the second noise reduction signal to obtain the noise reduction signal at the current moment, and outputting the signal through the audio signal output device; The second noise reduction signal is obtained by performing noise reduction processing on the error audio signal, and the error audio signal is generated by superimposing the noise reduction signal output by the audio signal output device at a previous moment and the second ambient noise signal collected by the second audio collection device at that time; While the audio signal output device continuously outputs the noise reduction signal, first filter coefficients are generated according to the positions of the listener's ears and the audio signal output device in the mobile vehicle, and the first filter is adjusted.

2. The method according to claim 1, wherein The first noise reduction signal and the first environmental noise signal have opposite phases and the same amplitude; The second noise reduction signal and the error audio signal have opposite phases and the same amplitudes.

3. The method according to claim 1, wherein The error audio signal is subjected to noise reduction processing in the following manner to obtain the second noise reduction signal: inputting the error audio signal into a second filter to generate the second noise reduction signal; The method further comprises: While the audio signal output device continuously outputs the noise reduction signal, the second filter coefficient of the second filter is determined according to the real-time speed of the mobile vehicle and the sound characteristics of the first ambient noise signal collected in real time, and the second filter is adjusted.

4. The method according to claim 3, wherein Determining a second filter coefficient of the second filter according to the real-time speed of the mobile vehicle and the sound characteristics of the first ambient noise signal collected in real time includes: Obtaining the real-time speed of the mobile vehicle and the sound characteristics of the first environmental noise signal collected in real time, and inputting them into a preset scene recognition model; Determining, according to the scene information output by the preset scene recognition model, corresponding filter coefficients as the second filter coefficients; Among them, the preset scene recognition model is a pre-trained neural network model that uses the speed information of the mobile vehicle and the sound characteristics of the ambient noise signal as input to predict the scene in which it is located; the filter coefficient is determined based on the pre-established correspondence between different scenes and filter coefficients.

5. The method according to claim 4, wherein The preset scene recognition model is trained in the following manner: Constructing the preset scene recognition model; Multiple groups of sample data containing speed information of mobile vehicles and sound characteristics of environmental noise are used as input, and the scene information corresponding to each group of sample data is used as the true value to train the preset scene recognition model to obtain the trained preset scene recognition model.

6. The method according to claim 1, wherein During the process of the audio signal output device continuously outputting the noise reduction signal, generating a first filter coefficient according to the position of the listener's ear and the audio signal output device in the mobile vehicle, and adjusting the first filter, including: determining a transfer function according to a sound transmission characteristic of a secondary path between the audio signal output device and the second audio acquisition device; During the process of the audio signal output device continuously outputting the noise reduction signal, the real-time determined ear position of the listener and the first ambient noise signal are input into the transfer function to output a reference signal; Inputting the reference signal and the error audio signal into an adaptive filter to obtain a first filter coefficient of a first filter; The first filter is adjusted using the first filter coefficients.

7. The method according to claim 6, wherein The ear positions of the listeners are determined by a preset computer vision algorithm.

8. A noise reduction device, characterized in that: include: Noise reduction module and head tracking module; The noise reduction module includes: a first audio acquisition device, a second audio acquisition device and an audio signal output device; The noise reduction module is configured to continuously collect a first ambient noise signal through the first audio acquisition device and input the signal into a first filter to generate a first noise reduction signal; superimpose the currently generated first noise reduction signal and the second noise reduction signal to obtain a noise reduction signal at a current moment, and output the signal through the audio signal output device; wherein the second noise reduction signal is obtained by performing noise reduction processing on an error audio signal, and the error audio signal is generated by superimposing the noise reduction signal outputted by the audio signal output device at a previous moment and the second ambient noise signal collected by the second audio acquisition device at that time; The head tracking module is used to generate a first filter coefficient and adjust the first filter according to the position of the listener's ear and the audio signal output device in the mobile vehicle while the audio signal output device continuously outputs the noise reduction signal.

9. A noise reduction device, characterized in that: Comprising the noise reduction device as claimed in claim 8.

10. A mobile vehicle, characterized in that: Comprising the noise reduction device as claimed in claim 9.

Images