Active noise reduction system and active noise reduction method

Abstract

The present invention discloses an active noise reduction system and an active noise reduction method, wherein the active noise reduction system comprises a plurality of feedback noise reduction paths and a processor; the feedback noise reduction paths are used to collect audio signals in real time and perform signal processing to obtain noise signals; the processor is used to extract the characteristics of the noise signals in real time, adjust the gains of each feedback noise reduction path according to the characteristics, and perform weighted processing on the noise signals based on the adjusted gains to generate noise reduction signals. The present invention can collect and process audio signals in real time, thereby achieving more accurate noise reduction control.

Description

Technical Field

[0001] The present invention relates to the technical field of active noise reduction, and in particular to an active noise reduction system and an active noise reduction method. Background Art

[0002] Active noise reduction headphones usually adopt a hybrid structure of feedback active noise reduction path and feedforward active noise reduction path. The feedback active noise reduction path includes a feedback microphone and a feedback noise reduction filter, and the feedback noise reduction filter performs noise reduction processing on the ear noise collected by the feedback microphone; the feedforward active noise reduction path includes a feedforward microphone and a feedforward noise reduction filter, and the feedforward noise reduction filter performs noise reduction processing on the ambient noise collected by the feedforward microphone.

[0003] With the popularity of active noise-cancelling headphones and the gradual maturity of the industry chain, the number of people using active noise-cancelling headphones is increasing, which means that users will use the active noise-cancelling function of headphones in more scenarios and with more ambient noise intensities. At present, users' demands and goals for noise reduction effects are constantly increasing, while high signal-to-noise ratio microphones still have limitations in terms of cost and technical maturity. In development and design, due to the limitations of the microphone's signal-to-noise ratio, it is usually necessary to find a balance between the background noise and the noise reduction performance during debugging to ensure the user's noise reduction comfort in a noise environment with lower loudness, making it impossible to adjust the noise reduction performance to the optimal level.

[0004] At present, the existing technology usually sets up multiple feedforward active noise reduction channels in headphones to improve the noise reduction ability of active noise reduction headphones against environmental noise in different directions. Although this will have a more obvious benefit in reducing the background noise within a specific frequency band, in a wider octave range, the background noise of the feedback noise reduction channel is still the primary contributor to the user's subjective perception of the background noise. However, during the long-term use of headphones, the feedback noise reduction channel may experience abnormal conditions such as microphone blockage and failure. These problems will cause a significant decrease in the noise reduction effect and affect the user experience. The existing technology lacks an effective mechanism to monitor and handle these abnormal conditions, and cannot guarantee the stability of the noise reduction performance.

[0005] Therefore, it is urgent to propose an active noise reduction system and an active noise reduction method to solve the above problems. Summary of the invention

[0006] The purpose of the present invention is to provide an active noise reduction system and an active noise reduction method, which can collect and process audio signals in real time, thereby achieving more accurate noise reduction control.

[0007] In order to solve the above technical problems, the present invention provides an active noise reduction system, comprising a plurality of feedback noise reduction paths and a processor;

[0008] The feedback noise reduction path is used to collect audio signals in real time and perform signal processing to obtain noise signals;

[0009] The processor is used to extract the characteristics of the noise signal in real time, adjust the gain of each feedback noise reduction path according to the characteristics, and perform weighted processing on the noise signal based on the adjusted gain to generate a noise reduction signal.

[0010] Further, the feedback noise reduction path includes: a feedback microphone, a primary signal processing module and a secondary signal processing module;

[0011] The feedback microphone is used to collect the audio signal in real time;

[0012] The primary signal processing module is used to perform analog gain and analog-to-digital conversion processing on the audio signal, and transmit the processed audio signal to the secondary signal processing module;

[0013] The secondary signal processing module is used to filter and gain-adjust the processed audio signal to obtain a noise signal, and transmit the noise signal to the processor.

[0014] Further, the processor includes a feature extraction module, a gain processing module and a weighted processing module;

[0015] The feature extraction module is used to extract features from the noise signal;

[0016] The gain processing module is used to calculate the gain parameter of the feedback noise reduction path in real time according to the extracted features, and adjust the gain of the feedback noise reduction path in real time according to the gain parameter;

[0017] The weighted processing module is used to perform weighted processing on the noise signal according to the adjusted gain of the feedback noise reduction path, and to superimpose the processed signals to generate a noise reduction signal.

[0018] Furthermore, the processor further includes an anomaly detection module, which includes a multi-dimensional feature extraction module, an anomaly degree calculation module, and a weight adjustment module;

[0019] The multidimensional feature extraction module is used to extract multidimensional feature vectors from the signal collected by the feedback noise reduction path in real time;

[0020] The abnormality degree calculation module is used to compare the multi-dimensional feature vector extracted in real time with a reference threshold to determine the abnormality degree of the feedback noise reduction path;

[0021] The weight adjustment module is used to adjust the weight coefficient of the feedback noise reduction path according to the abnormality degree;

[0022] The weighted processing module is further used to perform weighted processing on the noise signal of the feedback noise reduction path according to the adjusted weight coefficient.

[0023] In addition, the present invention also proposes an active noise reduction method, using the active noise reduction system as described above, which specifically includes the following:

[0024] Obtain the noise signal processed by the feedback noise reduction path in real time;

[0025] Extracting the characteristics of the noise signal in real time;

[0026] adjusting the gain of each feedback noise reduction path according to the characteristics;

[0027] The noise signal is weighted based on the adjusted gain to generate a noise reduction signal.

[0028] Furthermore, the real-time acquisition of the noise signal processed by the feedback noise reduction path specifically includes: synchronously collecting audio signals in the environment through multiple feedback noise reduction paths; performing analog gain and analog-to-digital conversion processing on the audio signals; and filtering and gain adjusting the processed audio signals to obtain the noise signal.

[0029] Further, adjusting the gain of each feedback noise reduction path according to the feature specifically includes: calculating the gain parameter of the feedback noise reduction path in real time according to the extracted feature, and adjusting the gain of the feedback noise reduction path in real time according to the gain parameter;

[0030] The weighted processing of the noise signal based on the adjusted gain to generate the noise reduction signal specifically includes: weighted processing of the noise signal according to the adjusted gain of the feedback noise reduction path, and superimposing the processed signals to generate the noise reduction signal.

[0031] Furthermore, it also includes: performing abnormality detection on the features extracted from the feedback noise reduction path, and adjusting the weight coefficient in the gain of the feedback noise reduction path according to the detection result; then performing weighted processing on the noise signal of the feedback noise reduction path according to the adjusted weight coefficient, and superimposing the processed signal to generate a noise reduction signal.

[0032] Furthermore, multidimensional feature vectors are extracted in real time from the signal collected by the feedback noise reduction path; the multidimensional feature vectors extracted in real time are compared with a reference threshold to determine the degree of abnormality of the feedback noise reduction path; the weight coefficient of the feedback noise reduction path is adjusted according to the degree of abnormality; the noise signal of the feedback noise reduction path is weighted according to the adjusted weight coefficient, and the processed signal is superimposed to generate a noise reduction signal.

[0033] Furthermore, if the ratio of the multidimensional feature vector to the reference threshold is greater than a predetermined value, it indicates that the degree of abnormality is high, otherwise the degree of abnormality is low; wherein, a first weight coefficient is assigned to the feedback noise reduction path with a high degree of abnormality, and a second weight coefficient is assigned to the feedback noise reduction path with a low degree of abnormality; the first weight coefficient is smaller than the second weight coefficient; when the signal-to-noise ratio in the feedback noise reduction path is lower than a specified value, the corresponding weight coefficient is reduced.

[0034] Through the above technical solution, the present invention has the following beneficial effects:

[0035] By setting up multiple feedback noise reduction paths and processors, and the feedback noise reduction paths are used to collect audio signals in real time and perform signal processing to obtain noise signals; the processor is used to extract the characteristics of the noise signal in real time, adjust the gain of each feedback noise reduction path according to the characteristics, and perform weighted processing on the noise signal based on the adjusted gain to generate a noise reduction signal. This device improves the accuracy of noise collection by using multiple feedback noise reduction paths, and improves the noise reduction effect by real-time feature extraction and gain adjustment, thereby achieving more accurate noise reduction control.

[0036] In addition, the device also effectively improves the noise reduction effect and reliability of use through the setting of the abnormal detection module. And through the real-time calculation and adjustment of the gain processing module, the dynamic adaptability of the system is enhanced. The device improves the overall processing efficiency through the collaborative work between modules. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic diagram of the overall structure of an active noise reduction system in one embodiment of the present invention;

[0038] Figure 2 It is a structural schematic diagram of a feedback noise reduction path in an active noise reduction system in one embodiment of the present invention;

[0039] Figure 3 It is a structural schematic diagram of a secondary signal processing module in an active noise reduction system in one embodiment of the present invention;

[0040] Figure 4 is a flow chart of an active noise reduction method in one embodiment of the present invention;

[0041] Figure 5 The figure is a flow chart of an abnormality detection process of an active noise reduction method in one embodiment of the present invention. DETAILED DESCRIPTION

[0042] The following will be described in more detail with reference to the accompanying drawings, wherein preferred embodiments of the present invention are shown. It should be understood that those skilled in the art may modify the present invention described herein while still achieving the advantageous effects of the present invention. Therefore, the following description should be understood as being widely known to those skilled in the art and not as a limitation of the present invention.

[0043] The present invention is described in more detail in the following paragraphs by way of example with reference to the accompanying drawings. The advantages and features of the present invention will become more apparent from the following description. It should be noted that the accompanying drawings are in very simplified form and are not in exact proportions, and are only used to facilitate and clearly assist in illustrating the purpose of the embodiments of the present invention.

[0044] Since traditional active noise reduction headphones mainly rely on fixed feedback (FB) path configuration, they achieve noise reduction by collecting external environmental noise and generating a cancellation signal. However, in the face of complex and changeable noise environments, fixed feedback paths may not achieve the best noise reduction effect. This embodiment can use a deep learning model to dynamically analyze the audio signals collected by each path, so as to automatically adjust the gain parameters of the feedback path, achieve real-time adaptive noise reduction effects, and improve user experience.

[0045] Specifically, Figure 1 As shown, an embodiment of the present invention provides an active noise reduction system, including multiple feedback noise reduction paths and a processor. The multiple feedback noise reduction paths can be built into different positions of a cavity of an electronic device (such as headphones) to achieve all-round collection of ambient noise.

[0046] Specifically, the feedback noise reduction path is used to collect audio signals in real time and perform signal processing to obtain noise signals; the processor is used to extract characteristics of the noise signals in real time, adjust the gains of each feedback noise reduction path according to the characteristics, and perform weighted processing on the noise signals based on the adjusted gains to generate noise reduction signals.

[0047] This embodiment improves the spatial coverage of noise collection by setting up multiple feedback noise reduction channels and avoids the limitations of single-channel collection; this embodiment collects and processes audio signals in real time, enhancing the real-time response capability of the system; the feature-based gain adjustment mechanism in this embodiment improves the accuracy and adaptability of noise reduction; in addition, this embodiment adopts a weighted processing strategy to enhance the system's processing capabilities for different noise sources.

[0048] Among them, the processor of this embodiment integrates a deep learning model, which can extract features from noise signals collected by each feedback noise reduction path. It can also identify different noise features and the background noise prediction results of different feedback noise reduction paths and generate dynamic gain parameters to ensure that the gain allocation meets the current noise environment. In addition, the deep learning model can adopt a convolutional neural network (CNN) or a recurrent neural network (RNN) structure, and train the model in advance so that it can recognize various typical noise scenes (such as human voice, traffic noise, mechanical noise, etc.) and original noise sources in different directions.

[0049] In a specific example, the extracted features include spectrum energy distribution, frequency components, and spatiotemporal features of noise, etc. Those skilled in the art will appreciate that the features may be set according to actual needs.

[0050] In a specific example, the number of the feedback noise reduction paths may be 3-8, preferably 4, which are arranged at the front, back, top and bottom of the earphone cavity to provide 360-degree noise collection coverage. It is known to those skilled in the art that the specific number of the feedback noise reduction paths may be set according to actual needs.

[0051] In one embodiment, if Figure 2 and Figure 3 As shown, the feedback noise reduction path includes: a feedback microphone, a primary signal processing module and a secondary signal processing module.

[0052] Specifically, the feedback microphone is used to collect the audio signal in real time; the primary signal processing module is used to perform analog gain and analog-to-digital conversion on the audio signal, and transmit the processed audio signal to the secondary signal processing module; the secondary signal processing module is used to filter and gain adjust the processed audio signal to obtain a noise signal, and transmit it to the processor.

[0053] This embodiment improves the reliability of signal processing by adopting a hierarchical architecture of feedback microphones and primary and secondary signal processing; improves the digital quality of the signal through the analog gain and analog-to-digital conversion of the primary signal processing module; enhances the accuracy of signal processing through filtering and gain adjustment of the secondary signal processing module of this embodiment; and the modular setting of this embodiment can improve the maintainability and upgrade convenience of the system.

[0054] In a specific example, the feedback microphone can be a high-sensitivity MEMS microphone, or can be set according to actual needs. In this embodiment, the headset body has multiple feedback noise reduction paths built in, and each feedback noise reduction path includes an independent feedback microphone (and may also include a collection circuit) to collect noise signals from different directions or positions.

[0055] In a specific example, the secondary signal processing module includes a feedback noise reduction filter and a gain adjustment module.

[0056] In a specific embodiment, the sampling rate of the primary signal processing module can be set to 44.1kHz (can be set according to actual needs), and the analog-to-digital conversion accuracy is 24 bits (can be set according to actual needs) to provide high-quality digital signal output. The secondary signal processing module uses a digital filter group, including a high-pass filter, a low-pass filter and a band-pass filter, which can perform targeted processing according to the noise characteristics of different frequency bands.

[0057] In one embodiment, the processor includes a feature extraction module, a gain processing module and a weighting processing module.

[0058] Specifically, the feature extraction module is used to extract features from the noise signal; the gain processing module is used to calculate the gain parameters of the feedback noise reduction path in real time according to the extracted features, and adjust the gain of the feedback noise reduction path in real time according to the gain parameters; wherein each feedback noise reduction path is dynamically adjusted in real time according to the calculated gain parameters, so that noises in different directions and frequency bands can be optimally offset.

[0059] During the calculation of the gain parameters, a weighting mechanism based on spectral characteristics can be introduced to make the headphones more sensitive to noise in a specific frequency band (such as low-frequency noise), thereby improving the noise reduction effect.

[0060] Furthermore, the weighted processing module is used to perform weighted processing on the noise signal according to the adjusted gain of the feedback noise reduction path, and to superimpose the processed signals to generate a noise reduction signal. More specifically, the processed signals are superimposed to generate a noise reduction signal, which is output through the speaker of the headset to offset the ambient noise. The device can periodically re-collect and analyze the ambient noise to ensure that the gain parameters can be updated in time to cope with dynamically changing noise scenarios.

[0061] This embodiment realizes accurate identification of noise features through the feature extraction module, thereby improving the targetedness of noise reduction; the real-time calculation and adjustment of the gain processing module of this embodiment can enhance the dynamic adaptability of the system; this embodiment improves the smoothness of the noise reduction effect through the signal superposition processing of the weighted processing module; and this embodiment can improve the overall processing efficiency through the collaborative work between modules.

[0062] The processor of this embodiment can dynamically calculate the gain parameter of each feedback noise reduction path according to the characteristics output by the deep learning model. The gain parameter is used to adjust the signal strength of each feedback noise reduction path to generate an optimized noise reduction signal.

[0063] In a specific example, the feature extraction adopts the short-time Fourier transform method, the window length is 512 points (can be set according to actual needs), and the overlap rate is 50% (can be set according to actual needs), which can ensure both real-time performance and the accuracy of spectrum analysis.

[0064] In summary, this embodiment can realize dynamic gain control of active noise reduction headphones by integrating the above-mentioned multiple feedback noise reduction paths and processors. By identifying the noise characteristics of each feedback noise reduction path, the noise reduction parameters can be adaptively adjusted according to the real-time noise environment, providing a better noise reduction effect and user experience.

[0065] In multi-channel active noise reduction headphones, each feedback noise reduction channel needs to work normally to ensure the best noise reduction effect. However, in the case of long-term use or complex external environment, some channels may have abnormalities, such as microphone damage, sound receiving hole blockage or collected signal distortion, resulting in a decrease in the overall noise reduction performance. Therefore, the processor described in this embodiment also includes an abnormality detection module, which calculates the degree of abnormality using the characteristics of the noise signal collected by each feedback noise reduction channel, and dynamically adjusts the weight coefficient of each feedback noise reduction channel to ensure that the headphones can still provide a good noise reduction effect when an abnormality occurs.

[0066] Specifically, the anomaly detection module is used to perform anomaly detection on the features extracted by the feedback noise reduction path, and adjust the weight coefficient in the gain of the feedback noise reduction path according to the detection result.

[0067] In a specific embodiment, the anomaly detection module includes a multi-dimensional feature extraction module, an anomaly degree calculation module and a weight adjustment module.

[0068] Specifically, the multidimensional feature extraction module is used to extract in real time the multidimensional feature vectors in the signal collected by the feedback noise reduction path; the abnormality degree calculation module is used to compare the multidimensional feature vectors extracted in real time with a reference threshold to determine the abnormality degree of the feedback noise reduction path; the weight adjustment module is used to adjust the weight coefficient of the feedback noise reduction path according to the abnormality degree to reduce the negative impact of the abnormal path on the noise reduction effect; the weighted processing module is also used to perform weighted processing on the noise signal of the feedback noise reduction path according to the adjusted weight coefficient.

[0069] This embodiment can improve the accuracy of anomaly detection by extracting multi-dimensional features; this embodiment can enhance the system's fault identification capability by adopting a real-time comparison and judgment mechanism; this embodiment can provide an optimized processing solution under abnormal conditions through a weight adjustment mechanism; and this embodiment can also improve the reliability of the system in complex environments.

[0070] In this embodiment, the abnormality degree calculation module can use a reference threshold or a deep learning model to determine the abnormality degree and dynamically adjust the weight coefficient of the feedback noise reduction path.

[0071] In addition, if Figure 4 As shown, this embodiment also proposes an active noise reduction method, using the active noise reduction system as described above, specifically including the following steps:

[0072] S1, obtaining the noise signal processed by the feedback noise reduction path in real time;

[0073] S2, extracting the characteristics of the noise signal in real time;

[0074] S3, adjusting the gain of each feedback noise reduction path according to the characteristics; and

[0075] S4. Perform weighted processing on the noise signal based on the adjusted gain to generate a noise reduction signal.

[0076] The real-time processing mechanism in this embodiment can ensure the timeliness of noise reduction; the combination of feature extraction and gain adjustment improves the noise reduction effect; and the weighted processing strategy enhances the adaptability of this method.

[0077] In step S1, the real-time acquisition of the noise signal processed by the feedback noise reduction path specifically includes: synchronously collecting audio signals in the environment through multiple feedback noise reduction paths; performing analog gain and analog-to-digital conversion processing on the audio signals; filtering and gain adjustment on the processed audio signals to obtain noise signals. The synchronous acquisition mechanism in this embodiment can improve the timing consistency of multi-path signals; the processing of analog gain and analog-to-digital conversion can improve signal quality; filtering and gain adjustment can optimize signal processing effects; and this embodiment can also improve the reliability of signal processing.

[0078] In step S3, adjusting the gain of each feedback noise reduction path according to the feature specifically includes: calculating the gain parameter of the feedback noise reduction path in real time according to the extracted feature, and adjusting the gain of the feedback noise reduction path in real time according to the gain parameter.

[0079] In step S4, the noise signal is weighted based on the adjusted gain to generate a noise reduction signal, which specifically includes: weighting the noise signal according to the adjusted gain of the feedback noise reduction path, and superimposing the processed signals to generate a noise reduction signal.

[0080] Steps S3 and S4 of this embodiment can improve the timeliness of gain adjustment through real-time calculation mechanism; parameterized adjustment can enhance the precise control capability of the system; weighted processing method can enhance the smoothness of noise reduction effect; and signal superposition strategy can optimize the final noise reduction effect.

[0081] As a preferred embodiment, Figure 5 As shown, this embodiment also includes: performing abnormality detection on the features extracted from the feedback noise reduction path, and adjusting the weight coefficient in the gain of the feedback noise reduction path according to the detection result; then performing weighted processing on the noise signal of the feedback noise reduction path according to the adjusted weight coefficient, and superimposing the processed signal to generate a noise reduction signal.

[0082] Furthermore, multidimensional feature vectors are extracted in real time from the signal collected by the feedback noise reduction path; the multidimensional feature vectors extracted in real time are compared with a reference threshold to determine the degree of abnormality of the feedback noise reduction path; the weight coefficient of the feedback noise reduction path is adjusted according to the degree of abnormality; the noise signal of the feedback noise reduction path is weighted according to the adjusted weight coefficient, and the processed signal is superimposed to generate a noise reduction signal.

[0083] The multidimensional feature vector may include, for example, parameters such as spectrum characteristics, signal amplitude, signal-to-noise ratio, etc. Those skilled in the art may appreciate that the multidimensional feature vector may be set according to actual needs.

[0084] In a specific example, spectrum characteristics: the spectrum information of each feedback noise reduction path signal can be extracted through fast Fourier transform (FFT) to capture the energy distribution of different frequency bands. Signal amplitude: the average amplitude value of the signal can be calculated to determine the strength level of the feedback noise reduction path signal. Signal-to-noise ratio (SNR): the signal-to-noise ratio of each feedback noise reduction path can be calculated to evaluate the signal quality. Time domain characteristics: the amplitude change rate and zero crossing rate of the signal can be calculated to detect possible abnormal fluctuations. At least one of the above characteristics is combined into a multidimensional feature vector to represent the noise signal characteristics of each feedback noise reduction path.

[0085] Furthermore, if the ratio of the multidimensional feature vector to the reference threshold (i.e., the abnormality degree value) is greater than a predetermined value, it indicates that the abnormality degree is high, otherwise the abnormality degree is low; wherein, a first weight coefficient is assigned to a feedback noise reduction path with a high abnormality degree, and a second weight coefficient is assigned to a feedback noise reduction path with a low abnormality degree; the first weight coefficient is less than the second weight coefficient; when the signal-to-noise ratio in the feedback noise reduction path is lower than a specified value, the corresponding weight coefficient is reduced.

[0086] In a specific example, the degree of abnormality can be calculated by distance calculation (such as Euclidean distance, cosine similarity) or machine learning model (such as anomaly detection network). The higher the value of the degree of abnormality, the greater the deviation of the signal of the feedback noise reduction path from the normal state, and the higher the possibility of abnormality.

[0087] In addition, this embodiment sets a dynamic weight adjustment mechanism according to the abnormality level of each feedback noise reduction path. For example, a feedback noise reduction path with a low abnormality level is assigned a higher weight (i.e., the second weight coefficient) to ensure that its signal has a greater impact on the noise reduction output. The feedback noise reduction path with a high abnormality level has its weight reduced (i.e., the first weight coefficient is reduced) to reduce the impact of the abnormal signal on the noise reduction effect. The weight adjustment process should be optimized according to different abnormality types. For example, if the signal-to-noise ratio of the feedback noise reduction path is low, its weight coefficient can be appropriately reduced to avoid excessive noise affecting the noise reduction effect.

[0088] Finally, the signals of each feedback noise reduction path are weighted according to the adjusted weight coefficients. The optimized noise reduction signal can be generated by superimposing the weighted signals and output to the headphone speaker to achieve real-time noise reduction. This embodiment can periodically recalculate the multidimensional feature vector and weight coefficient to ensure the best noise reduction effect in a dynamically changing noise environment.

[0089] Therefore, this embodiment detects abnormalities in the feedback noise reduction path, so that the device can identify abnormal paths in real time and dynamically adjust their weight coefficients, thereby reducing the impact of abnormal signals on the noise reduction effect. This method can be applied to complex and changeable noise environments, providing users with a more stable and reliable noise reduction experience.

[0090] In this embodiment, the device realizes all-round collection of environmental noise by arranging multiple feedback noise reduction paths at different positions of the earphone cavity. Each feedback noise reduction path uses a highly sensitive MEMS microphone to collect audio signals, and completes analog gain and analog-to-digital conversion through the first-level signal processing module to convert the analog signal into a high-quality digital signal. Subsequently, the second-level signal processing module uses a digital filter group to filter and gain-adjust the signal to obtain the noise signal to be processed.

[0091] Then, the processor first analyzes the noise signal through the feature extraction module to extract characteristic parameters such as spectral energy distribution, frequency components, and spatiotemporal characteristics of noise. Based on these characteristics, the gain processing module can use an adaptive algorithm to calculate and adjust the gain parameters of each feedback noise reduction path in real time. At the same time, the anomaly detection module continuously monitors the signal quality of each feedback noise reduction path, determines the degree of abnormality of each feedback noise reduction path by comparing the multidimensional feature vector with the reference threshold, and dynamically adjusts the weight coefficient accordingly. Finally, the weighted processing module performs weighted superposition on the noise signals of each feedback noise reduction path according to the adjusted gain and weight coefficient to generate the final noise reduction signal.

[0092] This multi-channel coordination and adaptive adjustment setting can not only improve the accuracy and spatial coverage of noise reduction, but also enhance the reliability of the system in complex environments through anomaly detection mechanisms. The processor's real-time feature extraction and gain adjustment can ensure that the system can quickly respond to changes in environmental noise and achieve precise noise reduction control.

[0093] In summary, the active noise reduction system and active noise reduction method proposed in the present invention have the following advantages:

[0094] By setting up multiple feedback noise reduction paths and processors, and the feedback noise reduction paths are used to collect audio signals in real time and perform signal processing to obtain noise signals; the processor is used to extract the characteristics of the noise signal in real time, adjust the gain of each feedback noise reduction path according to the characteristics, and perform weighted processing on the noise signal based on the adjusted gain to generate a noise reduction signal. This device improves the accuracy of noise collection by using multiple feedback noise reduction paths, and improves the noise reduction effect by real-time feature extraction and gain adjustment, thereby achieving more accurate noise reduction control.

[0095] In addition, the device also effectively improves the noise reduction effect and reliability of use through the setting of the abnormal detection module. And through the real-time calculation and adjustment of the gain processing module, the dynamic adaptability of the system is enhanced. The device improves the overall processing efficiency through the collaborative work between modules.

[0096] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An active noise reduction system, characterized in that: Includes multiple feedback noise reduction paths and processors; The feedback noise reduction path is used to collect audio signals in real time and perform signal processing to obtain noise signals; The processor is used to extract the characteristics of the noise signal in real time, adjust the gain of each feedback noise reduction path according to the characteristics, and perform weighted processing on the noise signal based on the adjusted gain to generate a noise reduction signal.

2. The active noise reduction system according to claim 1, characterized in that: The feedback noise reduction path includes: a feedback microphone, a primary signal processing module and a secondary signal processing module; The feedback microphone is used to collect the audio signal in real time; The primary signal processing module is used to perform analog gain and analog-to-digital conversion processing on the audio signal, and transmit the processed audio signal to the secondary signal processing module; The secondary signal processing module is used to filter and gain-adjust the processed audio signal to obtain a noise signal, and transmit the noise signal to the processor.

3. The active noise reduction system according to claim 1, characterized in that: The processor includes a feature extraction module, a gain processing module and a weighted processing module; The feature extraction module is used to extract features from the noise signal; The gain processing module is used to calculate the gain parameter of the feedback noise reduction path in real time according to the extracted features, and adjust the gain of the feedback noise reduction path in real time according to the gain parameter; The weighted processing module is used to perform weighted processing on the noise signal according to the adjusted gain of the feedback noise reduction path, and to superimpose the processed signals to generate a noise reduction signal.

4. The active noise reduction system according to claim 3, characterized in that: The processor further includes an anomaly detection module, which includes a multi-dimensional feature extraction module, an anomaly degree calculation module, and a weight adjustment module; The multidimensional feature extraction module is used to extract multidimensional feature vectors from the signal collected by the feedback noise reduction path in real time; The abnormality degree calculation module is used to compare the multi-dimensional feature vector extracted in real time with a reference threshold to determine the abnormality degree of the feedback noise reduction path; The weight adjustment module is used to adjust the weight coefficient of the feedback noise reduction path according to the abnormality degree; The weighted processing module is further used to perform weighted processing on the noise signal of the feedback noise reduction path according to the adjusted weight coefficient.

5. An active noise reduction method, using the active noise reduction system according to any one of claims 1 to 4, characterized in that: The details include: Obtain the noise signal processed by the feedback noise reduction path in real time; Extracting the characteristics of the noise signal in real time; adjusting the gain of each feedback noise reduction path according to the characteristics; The noise signal is weighted based on the adjusted gain to generate a noise reduction signal.

6. The active noise reduction method according to claim 5, characterized in that: The real-time acquisition of the noise signal processed by the feedback noise reduction path specifically includes: synchronously collecting audio signals in the environment through multiple feedback noise reduction paths; performing analog gain and analog-to-digital conversion processing on the audio signals; and filtering and gain adjusting the processed audio signals to obtain the noise signal.

7. The active noise reduction method according to claim 5, characterized in that: The step of adjusting the gain of each feedback noise reduction path according to the feature specifically includes: calculating the gain parameter of the feedback noise reduction path in real time according to the extracted feature, and adjusting the gain of the feedback noise reduction path in real time according to the gain parameter; The weighted processing of the noise signal based on the adjusted gain to generate the noise reduction signal specifically includes: weighted processing of the noise signal according to the adjusted gain of the feedback noise reduction path, and superimposing the processed signals to generate the noise reduction signal.

8. The active noise reduction method according to claim 5, characterized in that: Also includes: Performing anomaly detection on the features extracted by the feedback noise reduction path, and adjusting the weight coefficient in the gain of the feedback noise reduction path according to the detection result; Then, the noise signal of the feedback noise reduction path is weighted according to the adjusted weight coefficient, and the processed signals are superimposed to generate a noise reduction signal.

9. The active noise reduction method according to claim 8, characterized in that: Extracting multidimensional feature vectors from the signal collected by the feedback noise reduction path in real time; comparing the multidimensional feature vectors extracted in real time with a reference threshold to determine the abnormality of the feedback noise reduction path; adjusting the weight coefficient of the feedback noise reduction path according to the abnormality; performing weighted processing on the noise signal of the feedback noise reduction path according to the adjusted weight coefficient, and superimposing the processed signals to generate a noise reduction signal.

10. The active noise reduction method according to claim 9, characterized in that: If the ratio of the multidimensional feature vector to the reference threshold is greater than a predetermined value, it indicates that the abnormality is high, otherwise it is low; wherein, a first weight coefficient is assigned to a feedback noise reduction path with a high abnormality, and a second weight coefficient is assigned to a feedback noise reduction path with a low abnormality; the first weight coefficient is smaller than the second weight coefficient; when the signal-to-noise ratio in the feedback noise reduction path is lower than a specified value, the corresponding weight coefficient is reduced.

Images