Method and system for adaptively adjusting ambient noise of Bluetooth headset

Abstract

The invention relates to the technical field of audio noise reduction, and discloses a Bluetooth headset environment noise adaptive adjustment method and system, and the method comprises the steps: collecting multi-band sound data of headset environment noise through a preset acoustic sensor array, generating a noise spectrogram, extracting a noise masking threshold through time-frequency joint analysis, and collecting a headset audio signal; next, noise suppression is carried out on the audio signal by using a multi-order digital filter bank to obtain a primary noise reduction signal, the primary noise reduction signal is optimized into an optimized noise reduction signal based on a signal-to-noise ratio, and the frequency response matching degree of the optimized noise reduction signal and the original audio signal is calculated; then, performing phase correction and amplitude-frequency equalization processing according to the frequency response matching degree to obtain an audio equalization parameter; and finally, monitoring residual noise energy based on the parameter, generating a noise regulation and control instruction, driving adaptive sampling, and generating an adaptive noise reduction scheme. According to the invention, the audio noise reduction effect of the Bluetooth earphone can be improved.

Description

Technical Field

[0001] The present invention relates to a method and system for adaptively adjusting the environmental noise of a Bluetooth headset, belonging to the technical field of audio noise reduction. Background Art

[0002] As a portable audio device, Bluetooth headsets are widely used in communication, entertainment, and office scenarios. Under complex environmental noise conditions, traditional Bluetooth headsets usually adopt fixed noise reduction modes or manual adjustment methods, making it difficult to adapt to the dynamically changing noise environment in real time, resulting in a decline in call quality or a poor music playback experience.

[0003] Currently, the noise adjustment technology of Bluetooth headsets mainly relies on the combination of active noise cancellation (ANC) and passive noise isolation, and at the same time, it combines the environmental noise signals collected by microphones for feedback control. However, the existing technology has limitations in the recognition accuracy and response speed of environmental noise, and lacks an intelligent adaptive adjustment mechanism, resulting in unstable noise reduction effects. Therefore, a method for adaptively adjusting the environmental noise of Bluetooth headsets is needed to improve the audio noise reduction effect of Bluetooth headsets. Summary of the Invention

[0004] The present invention provides a method and system for adaptively adjusting the environmental noise of a Bluetooth headset, and its main purpose is to improve the audio noise reduction effect of the Bluetooth headset.

[0005] To achieve the above object, a method for adaptively adjusting the environmental noise of a Bluetooth headset provided by the present invention includes: Obtain the headset environmental noise corresponding to the target Bluetooth headset, collect multi-band sound data corresponding to the headset environmental noise by using a preset acoustic sensor array, and generate a noise spectrogram corresponding to the target environmental noise based on the multi-band sound data; Perform time-frequency joint analysis on the noise spectrogram to obtain spectral analysis parameters, extract the noise masking threshold in the spectral analysis parameters, and collect the headset audio signal in the target environmental noise based on the noise masking threshold; Based on a preset multi-order digital filter bank, perform noise suppression on the headset audio signal to obtain a primary noise reduction signal, calculate the signal-to-noise ratio corresponding to the primary noise reduction signal, optimize the primary noise reduction signal based on the signal-to-noise ratio to obtain an optimized noise reduction signal, and calculate the frequency response matching degree between the optimized noise reduction signal and the original audio signal; Based on the frequency response matching degree, perform phase correction on the optimized noise reduction signal to obtain a corrected audio signal, and perform amplitude-frequency equalization processing on the corrected audio signal to obtain audio equalization parameters; Based on the audio equalization parameters, the residual noise energy in the user's ear canal is monitored in real time, a noise regulation instruction corresponding to the residual noise energy is generated, based on the noise regulation instruction, the acoustic sensor array is driven to perform adaptive sampling of the noise frequency to obtain a noise sampling log, and based on the noise sampling log, an adaptive noise reduction scheme corresponding to the target Bluetooth headset is synchronously generated.

[0006] Optionally, the generating a noise spectrogram corresponding to the target ambient noise based on the multi-band sound data includes: Determine the noise analysis requirements corresponding to the multi-band sound data; Based on the noise analysis requirements, extract the sound pressure signal in the target ambient noise; Perform time-frequency transformation on the sound pressure signal to obtain an initial spectrum matrix; Remove the background interference components in the initial spectrum matrix to obtain optimized spectrum data; Based on the optimized spectrum data, generate a noise spectrogram corresponding to the target ambient noise.

[0007] Optionally, the collecting the headphone audio signal in the target ambient noise based on the noise masking threshold includes: According to the noise masking threshold, set the audio collection range corresponding to the target Bluetooth headset; Configure the filtering parameters corresponding to the audio collection range; Based on the filtering parameters, extract the effective noise frames in the target ambient noise; Perform superposition processing on the effective noise frames to obtain a superposed frequency band sample; Based on the superposed frequency band sample, collect the headphone audio signal in the target ambient noise.

[0008] Optionally, the performing noise suppression on the headphone audio signal based on a preset multi-order digital filter bank to obtain a primary noise reduction signal includes: Extract the time-frequency features in the headphone audio signal and construct a time-frequency distribution matrix corresponding to the time-frequency features; Based on the time-frequency distribution matrix, identify the noise-dominated frequency bands in the headphone audio signal; According to the noise-dominated frequency bands, dynamically adjust the cut-off frequency parameters of the preset multi-order digital filter bank; Use the adjusted multi-order digital filter bank to perform band-pass filtering on the headphone audio signal to obtain a frequency-domain filtered signal; Perform noise suppression on the frequency-domain filtered signal to obtain the primary noise reduction signal.

[0009] Optionally, calculating the signal-to-noise ratio corresponding to the primary noise reduction signal includes: Calculating the signal-to-noise ratio corresponding to the primary noise reduction signal by using the following formula: wherein represents the signal-to-noise ratio corresponding to the primary noise reduction signal, and respectively represent the start time and the end time of a defined time interval, represents a time variable, represents the useful audio signal component in the primary noise reduction signal at the time variable, represents the residual noise signal component in the primary noise reduction signal at the time variable.

[0010] Optionally, calculating the frequency response matching degree between the optimized noise reduction signal and the original audio signal includes: Calculating the frequency response matching degree between the optimized noise reduction signal and the original audio signal by using the following formula: wherein represents the frequency response matching degree between the optimized noise reduction signal and the original audio signal, represents the total number of frequency components, represents the number index corresponding to the frequency component, represents the original discrete spectrum of the i-th frequency component in the original audio signal, represents the optimized discrete spectrum of the i-th frequency component in the optimized noise reduction signal.

[0011] Optionally, performing amplitude-frequency equalization processing on the corrected audio signal to obtain audio equalization parameters includes: Identifying the frequency band energy distribution corresponding to the corrected audio signal; Calculating the frequency band energy weights corresponding to each frequency band in the corrected audio signal based on the frequency band energy distribution; Dynamically matching the frequency band energy weights with a preset target frequency response curve to generate frequency band compensation coefficients; Performing multi-stage recursive filtering on the corrected audio signal based on the frequency band compensation coefficients to obtain a multi-stage filtered signal; Performing amplitude-frequency equalization processing on the multi-stage filtered signal to obtain audio equalization parameters.

[0012] Optionally, based on the audio equalization parameters, real-time monitoring of the residual noise energy in the user's ear canal includes: Extracting the sound pressure distribution characteristics in the user's ear canal based on the audio equalization parameters; Analyze the sound pressure energy threshold corresponding to the described sound pressure distribution characteristics; Filter out the ear pressure frequency bands in the user's ear canal that exceed the described sound pressure energy threshold; Analyze the energy attenuation trend corresponding to the described ear pressure frequency band; Based on the energy attenuation trend, real-time monitor the residual noise energy in the user's ear canal.

[0013] Optionally, based on the noise regulation instruction, drive the acoustic sensor array to perform adaptive sampling of the noise frequency, and obtain a noise sampling log, including: Analyze the target noise reduction frequency band during the execution of the noise regulation instruction; Based on the target noise reduction frequency band, configure the corresponding directional sound pickup mode of the acoustic sensor array; Based on the directional sound pickup mode, adjust the noise sampling frequency corresponding to the acoustic sensor array; Based on the noise sampling frequency, collect the noise frequency band and sampling label in the acoustic sensor array; Based on the noise frequency band and the sampling label, generate a structured noise sampling log.

[0014] To solve the above problems, the present invention also provides a Bluetooth headset environmental noise adaptive adjustment system, and the system includes: A spectrogram generation module, configured to obtain the headset environmental noise corresponding to the target Bluetooth headset, collect multi-band sound data corresponding to the headset environmental noise by using a preset acoustic sensor array, and generate a noise spectrogram corresponding to the target environmental noise based on the multi-band sound data; A signal acquisition module, configured to perform time-frequency joint analysis on the noise spectrogram to obtain spectrogram analysis parameters, extract the noise masking threshold in the spectrogram analysis parameters, and collect the headset audio signal in the target environmental noise based on the noise masking threshold; A matching degree calculation module, configured to perform noise suppression on the headset audio signal based on a preset multi-order digital filter bank to obtain a primary noise reduction signal, calculate the signal-to-noise ratio corresponding to the primary noise reduction signal, perform optimization processing on the primary noise reduction signal based on the signal-to-noise ratio to obtain an optimized noise reduction signal, and calculate the frequency response matching degree between the optimized noise reduction signal and the original audio signal; An audio equalization module, configured to perform phase correction on the optimized noise reduction signal based on the frequency response matching degree to obtain a corrected audio signal, and perform amplitude-frequency equalization processing on the corrected audio signal to obtain audio equalization parameters; A scheme generation module is used to, based on the audio equalization parameters, monitor in real time the residual noise energy in the user's ear canal, generate a noise regulation instruction corresponding to the residual noise energy, drive the acoustic sensor array to perform adaptive sampling of the noise frequency based on the noise regulation instruction, obtain a noise sampling log, and synchronously generate an adaptive noise reduction scheme corresponding to the target Bluetooth headset based on the noise sampling log.

[0015] Compared with the problems described in the background art, the present invention can effectively improve the noise reduction effect of the headset in different scenarios by obtaining the headset environmental noise corresponding to the target Bluetooth headset, significantly improve the call clarity and music playback quality, and also enhance the adaptability of the Bluetooth headset in a dynamic noise environment. The present invention performs a time-frequency joint analysis on the noise spectrogram to obtain spectral analysis parameters, and extracts the noise masking threshold in the spectral analysis parameters. It can intelligently distinguish useful audio signals from noise based on this threshold, avoiding over-noise reduction or under-noise reduction, thereby retaining the true details of the audio signal to the greatest extent and significantly improving the clarity and comfort of the audio output. Further, the present invention performs noise suppression on the headset audio signal based on a preset multi-order digital filter bank to obtain a primary noise reduction signal, which can perform refined processing on the signal from multiple dimensions in response to the multi-frequency characteristics of complex environmental noise, effectively reducing the noise interference intensity, and further optimizing the headset audio quality. Further, the present invention performs phase correction on the optimized noise reduction signal based on the frequency response matching degree to obtain a corrected audio signal, which can compensate for the phase deviation generated during the noise reduction process, making the phase relationship between the frequency components of the optimized noise reduction signal closer to the original audio and reducing the deterioration of the sound quality caused by phase distortion. Finally, the present invention monitors in real time the residual noise energy in the user's ear canal based on the audio equalization parameters, can accurately grasp the dynamic change of the noise in the ear canal after noise reduction, and can timely discover the noise reduction blind area or over-noise reduction area by combining the analysis of the audio equalization parameters and the real-time noise energy data, providing data support for dynamically optimizing the noise reduction strategy. Therefore, the Bluetooth headset environmental noise adaptive adjustment method and system provided by the embodiments of the present invention can improve the audio noise reduction effect of the Bluetooth headset. Description of the Drawings

[0016] Figure 1 It is a schematic flow chart of a Bluetooth headset environmental noise adaptive adjustment method provided by an embodiment of the present invention; Figure 2 It is a schematic module diagram of a system for implementing the Bluetooth headset environmental noise adaptive adjustment provided by an embodiment of the present invention.

[0017] The implementation, functional features and advantages of the object of the present invention will be further described in conjunction with the embodiments with reference to the drawings. Detailed Embodiments

[0018] It should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0019] An embodiment of the present application provides a method for adaptively adjusting the environmental noise of a Bluetooth headset. The execution subject of the method for adaptively adjusting the environmental noise of the Bluetooth headset includes, but is not limited to, at least one of electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the method for adaptively adjusting the environmental noise of the Bluetooth headset can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to: a single server, a server cluster, a cloud server, or a cloud server cluster, etc.

[0020] Embodiment 1: Refer to Figure 1 As shown, it is a schematic flowchart of a method for adaptively adjusting the environmental noise of a Bluetooth headset provided by an embodiment of the present invention. In this embodiment, the method for adaptively adjusting the environmental noise of the Bluetooth headset includes: S1. Obtain the headset environmental noise corresponding to the target Bluetooth headset, collect multi-band sound data corresponding to the headset environmental noise by using a preset acoustic sensor array, and generate a noise spectrogram corresponding to the target environmental noise based on the multi-band sound data.

[0021] By obtaining the headset environmental noise corresponding to the target Bluetooth headset, the present invention can effectively improve the noise reduction effect of the headset in different scenarios, significantly improve the call clarity and music playback quality, and at the same time enhance the adaptability of the Bluetooth headset in a dynamic noise environment.

[0022] Among them, the target Bluetooth headset refers to a specific Bluetooth headset device that the user is using and needs to perform adaptive adjustment of environmental noise. In actual application scenarios, it can be a sports Bluetooth headset worn during running, a noise-canceling Bluetooth headset used during commuting, or a sound-insulating Bluetooth headset worn during learning, etc.; the headset environmental noise refers to various interference sounds existing in the environment where the Bluetooth headset is located. For example, the noisy sound in the subway station, the conversation sound in the office, etc. Optionally, the obtaining of the headset environmental noise corresponding to the target Bluetooth headset can be achieved by the environmental sound collection method. For example: real-time collection of environmental sound wave signals through the microphone array built in the Bluetooth headset, and extraction of the pure environmental noise component by using a noise separation algorithm (such as: spectral subtraction, independent component analysis ICA), and finally obtaining the headset environmental noise.

[0023] Furthermore, by collecting multi-band sound data corresponding to the headset environmental noise by using a preset acoustic sensor array, the present invention can obtain noise information more comprehensively, enabling the Bluetooth headset to accurately identify complex and variable environmental noise, and thus achieving more efficient adaptive noise reduction.

[0024] Among them, the preset acoustic sensor array refers to a combined system of multiple acoustic sensors pre-arranged inside the Bluetooth headset. For example, multiple microphones are installed at different parts inside the headset housing to comprehensively collect various sound information in the environment; the multi-band sound data refers to the sound signals collected and the data set obtained after being divided according to different frequency ranges. Its sound frequency covers a very wide range, from low-frequency sound waves to high-frequency sound waves. Different frequency bands contain different environmental information. For example, the sound of a car engine and the sound of subway operation are mainly concentrated in the low-frequency band; conversation sounds and alarm sounds are distributed in the mid-high frequency band. The multi-band sound data decomposes and records the sound according to multiple frequency bands such as low frequency, middle frequency, and high frequency, and details the characteristics such as the intensity and duration of the sound in each frequency band. Optionally, collecting the multi-band sound data corresponding to the ambient noise of the headset can be achieved through audio signal processing methods, such as: performing frequency-domain decomposition on the original noise signal through fast Fourier transform (FFT) to extract the energy distribution in different frequency bands, so as to obtain multi-band sound data.

[0025] Furthermore, based on the multi-band sound data, the present invention generates a noise spectrogram corresponding to the target ambient noise, which can provide a clear data reference for subsequent noise analysis and processing, facilitate accurate identification of noise components and interference frequency bands, and thus formulate a more targeted noise reduction strategy for the Bluetooth headset.

[0026] Among them, the noise spectrogram refers to a visual chart that graphically displays the frequency characteristics of the target ambient noise. It usually uses frequency as the abscissa and noise intensity (such as sound pressure level) as the ordinate, and intuitively presents the intensity distribution of the noise in each frequency band and its change over time through curves, color blocks, etc. of different colors or heights.

[0027] As an embodiment of the present invention, generating the noise spectrogram corresponding to the target ambient noise based on the multi-band sound data includes: determining the noise analysis requirements corresponding to the multi-band sound data; extracting the sound pressure signal in the target ambient noise based on the noise analysis requirements; performing time-frequency transformation on the sound pressure signal to obtain an initial spectrum matrix; removing the background interference components in the initial spectrum matrix to obtain optimized spectrum data; generating the noise spectrogram corresponding to the target ambient noise based on the optimized spectrum data.

[0028] Among them, the noise analysis requirement refers to the specific information or goal expected to be obtained when analyzing environmental noise according to different usage scenarios and functional requirements of a Bluetooth headset. For example, in a call scenario, the noise analysis requirement can be to accurately identify and separate the noise frequency bands that interfere with the call; in a music playback scenario, the requirement can be to analyze the degree of influence of noise on the sounds of different musical instrument frequency bands, etc.; the sound pressure signal refers to the signal generated by the change in the pressure of a medium (such as air) caused by the propagation of sound. When the noise sound wave in the environment propagates to the acoustic sensor of the Bluetooth headset, the sensor converts the change in air pressure caused by the sound wave into an electrical signal, and this electrical signal is the sound pressure signal, which contains all the information of the environmental noise, such as the intensity, frequency, waveform, etc. of the noise; the initial spectrum matrix refers to the two-dimensional data matrix obtained after performing time-frequency transformation on the sound pressure signal. The time-frequency transformation converts the originally time-varying sound pressure signal into a representation form that contains both time and frequency information. Each element in the matrix corresponds to the signal intensity at a specific time point and frequency point. Through the initial spectrum matrix, the distribution of noise at different times and frequencies can be initially observed; the optimized spectrum data refers to the data obtained by removing the background interference components from the initial spectrum matrix. The background interference components can come from the background noise of the sensor itself, the interference signals generated by the device circuit, or some stable environmental sounds that are irrelevant to the target environmental noise.

[0029] Furthermore, determining the noise analysis requirement corresponding to the multi-band sound data can be achieved through a requirement classification algorithm. For example, classifying the noise application scenarios through K-means clustering or a decision tree model to obtain the noise analysis requirement; extracting the sound pressure signal in the target environmental noise can be achieved through a sound pressure level calculation tool. For example, standardizing the original sound wave signal through an A-weighting network or a sound level meter calibration module to obtain the sound pressure signal; performing time-frequency transformation on the sound pressure signal can be achieved through a time-frequency analysis algorithm. For example, converting the time-domain signal into a frequency-domain representation through the short-time Fourier transform (STFT) or the Wigner-Ville distribution to obtain the initial spectrum matrix; removing the background interference components from the initial spectrum matrix can be achieved through a noise suppression algorithm. For example, separating the noise and signal components through spectral subtraction or non-negative matrix factorization (NMF) to obtain the optimized spectrum data; generating the noise spectrogram corresponding to the target environmental noise can be achieved through a data visualization tool. For example, plotting the frequency-domain energy distribution through the specgram function of Matplotlib or the heat map of Seaborn to obtain the noise spectrogram.

[0030] S2. Perform time-frequency joint analysis on the noise spectrogram to obtain spectral analysis parameters, extract the noise masking threshold from the spectral analysis parameters, and collect the headphone audio signal in the target ambient noise based on the noise masking threshold.

[0031] By performing time-frequency joint analysis on the noise spectrogram, obtaining spectral analysis parameters, and extracting the noise masking threshold from the spectral analysis parameters, the present invention can intelligently distinguish useful audio signals from noise based on this threshold, avoid over-denoising or under-denoising, thereby retaining the true details of the audio signal to the greatest extent and significantly improving the clarity and comfort of the audio output.

[0032] Among them, the spectral analysis parameters refer to a series of quantitative indicators that can describe the characteristics of noise obtained after performing time-frequency joint analysis on the noise spectrogram, which cover information such as the energy distribution of noise in different frequency bands, the intensity of frequency components, and the change trend of each frequency component over time. For example, data such as the proportion of noise energy in a specific frequency band, the central position of the noise frequency, and the bandwidth of the frequency band; the noise masking threshold refers to the sound pressure level value corresponding to a sound when the human ear can just not perceive the existence of the sound in a specific noise environment. It is based on the auditory masking effect of the human ear, that is, when there is strong noise in the environment, weaker sounds will be masked and difficult to detect, and the noise masking threshold can be used as a standard for the Bluetooth headset to judge which environmental sounds belong to negligible noise. Optionally, the time-frequency joint analysis of the noise spectrogram can be implemented through a feature extraction algorithm, such as calculating the energy aggregation characteristics of the spectrogram through Gabor transform or Cohen-class time-frequency distribution to obtain spectral analysis parameters; the extraction of the noise masking threshold from the spectral analysis parameters can be implemented through an acoustic model, such as calculating the critical masking level of each frequency band through the MPEG acoustic model or Johnston masking threshold algorithm to obtain the noise masking threshold.

[0033] Furthermore, based on the noise masking threshold, the present invention collects the headphone audio signal in the target ambient noise, which can make the audio signal collected by the Bluetooth headset more in line with the human ear's perception requirements, effectively improve the purity and usefulness of the audio signal, lay a solid foundation for subsequent noise reduction processing and high-quality audio output, and significantly optimize the user's auditory experience.

[0034] Among them, the headphone audio signal refers to the electrical signal containing the sound content required by the user obtained after a series of collection and processing operations by the target Bluetooth headset in a specific environment. It can be either the user's voice call signal, the played music signal, or the environmental sound signal retained after screening (such as necessary prompt sounds, etc.).

[0035] As an embodiment of the present invention, collecting the headphone audio signal in the target ambient noise based on the noise masking threshold includes: setting an audio collection range corresponding to the target Bluetooth headset according to the noise masking threshold; configuring filtering parameters corresponding to the audio collection range; extracting valid noise frames in the target ambient noise based on the filtering parameters; performing superposition processing on the valid noise frames to obtain a superposed frequency band sample; and collecting the headphone audio signal in the target ambient noise based on the superposed frequency band sample.

[0036] Among them, the audio collection range refers to the collection interval of sound frequency and intensity set for the target Bluetooth headset according to the noise masking threshold. This range clearly stipulates which frequency bands and what intensity of sound signals the headset focuses on collecting, avoiding collecting meaningless weak noises or sounds beyond the human ear's perception range. For example, when there is a large amount of low-frequency background noise in the environment but it does not affect human ear hearing, the collection range can avoid this low-frequency band and focus on voice, music and other signals in the mid-high frequency bands; the filtering parameters refer to a set of values used to adjust the signal filtering characteristics in the audio collection process. By configuring the filtering parameters, the collection system of the Bluetooth headset can more accurately screen out the required audio signals. For example, setting appropriate cut-off frequency parameters can effectively filter out useless noise signals below the noise masking threshold while retaining useful audio signals; the valid noise frames refer to the noise data segments that meet the requirements of the audio collection range and contain useful information after being screened by the filtering parameters in the target ambient noise. For example, in a noisy environment, the valid noise frames can include interfering sounds such as machine sounds and conversations, or can also include call sound segments; the superposed frequency band sample refers to the audio data set obtained by performing superposition processing on multiple valid noise frames, which integrates the information of the valid noise frames at different times and frequency bands. For example, after superposing multiple valid noise frames containing different voice frequency components, the spectral characteristics of the voice signal can be restored more completely.

[0037] Further, the audio acquisition range corresponding to the target Bluetooth headset can be achieved through a frequency band analysis tool. For example, the audible frequency band of 20 Hz - 20 kHz for the human ear can be determined through FFT spectrum analysis to obtain the audio acquisition range. The filtering parameters corresponding to the configured audio acquisition range can be achieved through a digital filter design tool. For example, the parameters of an IIR band-pass filter can be designed through the FDAtool of MATLAB to obtain the filtering parameters. The extraction of the effective noise frames from the target ambient noise can be achieved through a voice activity detection algorithm. For example, non-silent frames can be screened through an energy-based VAD algorithm to obtain the effective noise frames. The superposition processing of the effective noise frames can be achieved through a signal averaging algorithm. For example, the power spectra of multiple frames of noise can be superimposed through the Welch method to obtain the superimposed frequency band samples. The acquisition of the headphone audio signal from the target ambient noise can be achieved through a dual-microphone beamforming technique. For example, the sound source in the target direction can be enhanced through the MVDR algorithm to obtain the headphone audio signal.

[0038] S3. Based on a preset multi-order digital filter bank, perform noise suppression on the headphone audio signal to obtain a primary noise reduction signal, calculate the signal-to-noise ratio corresponding to the primary noise reduction signal, and based on the signal-to-noise ratio, perform optimization processing on the primary noise reduction signal to obtain an optimized noise reduction signal, and calculate the frequency response matching degree between the optimized noise reduction signal and the original audio signal.

[0039] Based on a preset multi-order digital filter bank, the present invention performs noise suppression on the headphone audio signal to obtain a primary noise reduction signal, which can target the multi-frequency characteristics of complex ambient noise and perform refined processing on the signal from multiple dimensions, effectively reducing the noise interference intensity, thereby further optimizing the headphone audio quality.

[0040] Among them, the preset multi-order digital filter bank refers to a system composed of multiple digital filters with different orders and different filtering characteristics. These filters are arranged and work together according to preset parameters and structures. Each digital filter has its unique frequency response characteristic. For example, a low-pass filter can allow low-frequency signals to pass through and suppress high-frequency signals, while a high-pass filter does the opposite, and a band-pass filter can allow signals in a specific frequency band to pass through. The primary noise reduction signal refers to the audio signal obtained after further noise suppression processing on the frequency-domain filtered signal. Based on the frequency-domain filtered signal, through more refined algorithms and processing means, the residual noise components are further weakened, providing a better basic signal for subsequent audio optimization processing.

[0041] As an embodiment of the present invention, suppressing the noise of the headphone audio signal based on a preset multi-stage digital filter bank to obtain a primary noise reduction signal includes: extracting the time-frequency features in the headphone audio signal and constructing a time-frequency distribution matrix corresponding to the time-frequency features; identifying the noise-dominated frequency band in the headphone audio signal based on the time-frequency distribution matrix; dynamically adjusting the cut-off frequency parameter of the preset multi-stage digital filter bank according to the noise-dominated frequency band; filtering the headphone audio signal in sub-bands by using the adjusted multi-stage digital filter bank to obtain a frequency-domain filtered signal; and suppressing the noise of the frequency-domain filtered signal to obtain the primary noise reduction signal.

[0042] Among them, the time-frequency feature refers to the characteristics and variation laws presented by the headphone audio signal in two dimensions of time and frequency, which covers information such as the intensity and phase of each frequency component contained in the signal at different times. For example, a speech signal may contain different frequency formants at different time points, and the changes of these formants and the energy distribution of each frequency component constitute the time-frequency feature of the speech signal; the time-frequency distribution matrix refers to representing the time-frequency feature of the headphone audio signal in the form of a matrix, which is a visual and structured data presentation method. The rows of the matrix correspond to different time points, the columns correspond to different frequency points, and the values of the matrix elements represent the intensity or energy magnitude of the signal at the corresponding time-frequency coordinates; the noise-dominated frequency band refers to the noise frequency range with the largest energy proportion and the most significant interference to the useful audio signal in the headphone audio signal. For example, in a factory workshop environment, the low-frequency rumbling sound and high-frequency friction sound generated by machine operation are mixed together, and the low-frequency rumbling sound becomes the noise-dominated frequency band due to its large energy; the cut-off frequency parameter refers to a key parameter in the preset multi-stage digital filter bank, which determines the boundary of the frequency range allowed to pass through the filter. For a low-pass filter, the low-frequency signal below the cut-off frequency parameter can pass through smoothly, and the high-frequency signal above this parameter will be suppressed; for a high-pass filter, it is the opposite. By dynamically adjusting the cut-off frequency parameter, the filter bank can more accurately adapt to different noise frequency bands; the frequency-domain filtered signal refers to the signal obtained by filtering the headphone audio signal in sub-bands by using the adjusted multi-stage digital filter bank. During the filtering process, different filters process the signals of their respective responsible frequency bands according to the set cut-off frequency parameter, filter out the noise signals that do not meet the requirements, and retain the useful audio signal components. The final set of these signals processed in sub-bands is the frequency-domain filtered signal.

[0043] Further, the extraction of time-frequency features from the headphone audio signal can be achieved through time-frequency analysis methods. For example, voice features can be extracted through Mel Frequency Cepstral Coefficients (MFCC) or Gammatone filter banks to obtain time-frequency features. The construction of the time-frequency distribution matrix corresponding to the time-frequency features can be realized through matrix transformation algorithms. For example, the time-domain signal can be converted into a frequency-domain energy distribution matrix through Short-Time Fourier Transform (STFT) to obtain the time-frequency distribution matrix. The identification of the noise-dominated frequency band in the headphone audio signal can be achieved through spectrum analysis tools. For example, the energy-concentrated frequency band can be determined through Power Spectral Density (PSD) analysis to obtain the noise-dominated frequency band. The dynamic adjustment of the cut-off frequency parameters of the preset multi-order digital filter bank can be realized through adaptive filtering algorithms. For example, the passband boundaries of each filter can be optimized in real time through the LMS algorithm to obtain the cut-off frequency parameters. The filtering of the headphone audio signal in sub-bands can be achieved through a digital filter bank. For example, multi-band signal separation can be performed through a Butterworth filter bank to obtain the frequency-domain filtered signal. The noise suppression of the frequency-domain filtered signal can be achieved through spectral subtraction. For example, the noise components in each frequency band can be eliminated through an improved MMSE spectral subtraction method to obtain the primary noise reduction signal.

[0044] By calculating the signal-to-noise ratio corresponding to the primary noise reduction signal, the present invention can quantitatively evaluate the audio quality after noise reduction processing, intuitively show the proportional relationship between the useful components and the noise components in the signal, and through this value, it can be clearly judged whether the current noise reduction effect meets the expectation.

[0045] Among them, the signal-to-noise ratio is an important indicator to measure the signal quality. By comparing the ratio of the energy of the useful audio signal to the energy of the residual noise signal in the primary noise reduction signal, taking the logarithm (base 10) and then multiplying by 10 to convert it into the form of decibels (dB), it intuitively reflects the strength of the useful components relative to the noise components in the signal. The higher the value, the more prominent the useful audio signal, the smaller the noise interference, and the better the audio quality.

[0046] As an embodiment of the present invention, the calculation of the signal-to-noise ratio corresponding to the primary noise reduction signal includes: Calculating the signal-to-noise ratio corresponding to the primary noise reduction signal using the following formula: Among them, represents the signal-to-noise ratio corresponding to the primary noise reduction signal, and respectively represent the start time and end time of the defined time interval, represents the time variable, represents the useful audio signal component in the primary noise reduction signal at the time variable, Represents the residual noise signal component in the primary noise reduction signal when the time variable is considered.

[0047] Specifically, the defined time interval refers to a period of time range determined by the start time and the end time Since the audio signal changes continuously over time, in order to accurately calculate the signal-to-noise ratio of a certain segment of the signal, it is necessary to select such a specific time interval for analysis. For example, when analyzing a segment of a voice call or a music playback clip, selecting this interval can ensure that the calculated signal-to-noise ratio corresponds to the effective audio of this segment and avoid interference from signals in other time periods; the time variable refers to a continuous variable representing the dimension of time, which is used to describe the state of the audio signal at different moments; the useful audio signal component refers to the part that the user expects to hear in the primary noise reduction signal, such as the voice content of a call, the played music, etc. It carries the main information and is a key component in the audio signal. In the noise reduction process, this part of the signal should be retained and highlighted as much as possible; the residual noise signal component refers to the remaining noise part in the primary noise reduction signal after noise reduction processing (such as operations like filtering with a multi-stage digital filter bank).

[0048] Furthermore, the principle of the above formula is to measure the quality of the primary noise reduction signal based on the energy ratio. It first calculates the energy of the useful audio signal component and the residual noise signal component within the defined time interval through integration, then calculates the energy ratio of the two, and finally takes the logarithm with base 10 and multiplies by 10 to convert it into decibel form to obtain the signal-to-noise ratio SNR. The decibel form conforms to the logarithmic perception characteristics of human hearing and can intuitively reflect the strength of the useful signal relative to the noise signal, evaluating the noise reduction effect.

[0049] Based on the signal-to-noise ratio, the present invention performs optimization processing on the primary noise reduction signal to obtain an optimized noise reduction signal. By specifically adjusting the noise reduction algorithm or filter parameters, it can effectively suppress noise and improve the audio purity.

[0050] Among them, the optimized noise reduction signal refers to an audio signal obtained by further processing the residual noise in the signal based on the signal-to-noise ratio, a quantitative index, on the basis of the primary noise reduction signal. By analyzing the signal-to-noise ratio, it specifically adjusts the noise reduction algorithm or filter parameters to more precisely suppress the residual noise and increase the proportion of the useful audio signal. Optionally, the optimization processing of the primary noise reduction signal can be achieved through an adaptive filtering algorithm. For example, the RLS algorithm is used to track and eliminate the residual environmental noise, thereby obtaining the optimized noise reduction signal.

[0051] Furthermore, by calculating the frequency response matching degree between the optimized noise reduction signal and the original audio signal, the present invention can evaluate the degree to which the noise reduction process retains the original frequency characteristics of the audio, and accordingly determine whether the noise reduction process excessively changes the audio timbre, and timely detect and correct potential distortion problems.

[0052] Among them, the frequency response matching degree is an index used to measure the similarity degree of the optimized noise reduction signal and the original audio signal in terms of frequency response characteristics, and its value range is between 0 and 1. The closer it is to 1, the more similar the optimized noise reduction signal is to the original audio signal in terms of frequency response, which means that the noise reduction process changes the frequency characteristics of the original audio signal less, and can better retain the timbre, sound quality and other characteristics of the original audio.

[0053] As an embodiment of the present invention, calculating the frequency response matching degree between the optimized noise reduction signal and the original audio signal includes: Calculating the frequency response matching degree between the optimized noise reduction signal and the original audio signal by using the following formula: Among them, represents the frequency response matching degree between the optimized noise reduction signal and the original audio signal, represents the total number of frequency components, represents the corresponding number index of the frequency component, represents the original discrete spectrum of the i-th frequency component in the original audio signal, represents the optimized discrete spectrum of the i-th frequency component in the optimized noise reduction signal.

[0054] Specifically, the closer it is to 0, the greater the frequency response difference between the two; the frequency component refers to the superposition of sine waves with different frequencies, and these sine waves with different frequencies are frequency components. Each frequency component has its corresponding amplitude and phase, which jointly determine the characteristics of the audio signal. In discrete spectrum analysis, the audio signal is decomposed into a series of discrete frequency points, and each frequency point represents a frequency component; the original discrete spectrum refers to the discrete representation obtained after converting the original audio signal to the frequency domain through methods such as discrete Fourier transform. It consists of discrete frequency points, and each point represents the amplitude of the original audio signal at the -th frequency component (in some cases, phase information may also be included, but the formula here mainly focuses on the amplitude), which reflects the energy size of the original audio signal at this frequency. The complete original discrete spectrum shows the energy distribution of the original audio signal in the entire frequency range; the optimized discrete spectrum refers to the discrete representation obtained after performing frequency domain conversion operations such as discrete Fourier transform on the optimized noise reduction signal, and it also consists of It consists of discrete frequency points, where represents the amplitude of the optimized noise-reduced signal at the th frequency component, reflecting the energy state of the signal at this frequency component after noise reduction optimization.

[0055] Furthermore, the principle of the above formula is based on calculating the frequency response matching degree by comparing the amplitude differences of the optimized noise-reduced signal and the original audio signal at each frequency component. First, it calculates the square root of the sum of the squares of the amplitude differences of the corresponding frequency components of the two to obtain the total difference amount, then normalizes it by dividing by the square root of the sum of the squares of the amplitudes of each frequency component of the original audio signal, and finally subtracts the normalized difference value from 1 to obtain the frequency response matching degree. The closer the value is to 1, the more similar the frequency response characteristics of the two are.

[0056] S4. Based on the frequency response matching degree, perform phase correction on the optimized noise-reduced signal to obtain a corrected audio signal, and perform amplitude-frequency equalization processing on the corrected audio signal to obtain audio equalization parameters.

[0057] Based on the frequency response matching degree, the present invention performs phase correction on the optimized noise-reduced signal to obtain a corrected audio signal, which can compensate for the phase deviation generated during the noise reduction process, making the phase relationship between the frequency components of the optimized noise-reduced signal closer to that of the original audio and reducing the deterioration of sound quality caused by phase distortion.

[0058] Among them, the corrected audio signal refers to the audio signal obtained by performing phase correction based on the frequency response matching degree on the basis of the optimized noise-reduced signal. During the noise reduction process, the phase of the signal may change, which will affect the sense of space, stereo effect, and timbre of the sound. Phase correction is to adjust these offset phases to make them as close as possible to the state that the original audio signal should have. Optionally, the phase correction of the optimized noise-reduced signal can be achieved through complex domain processing methods, such as using a complex neural network (DCCRN) to optimize both amplitude and phase characteristics simultaneously to obtain the corrected audio signal.

[0059] Furthermore, the present invention performs amplitude-frequency equalization processing on the corrected audio signal to obtain audio equalization parameters, which can specifically compensate for or suppress specific frequency bands, correct the frequency response deviation caused by noise reduction, transmission, etc., and can provide a basis for further optimization of the audio system, making the energy distribution of each frequency band of the sound more reasonable.

[0060] Among them, the audio equalization parameters refer to a set of key parameters obtained by performing amplitude-frequency equalization processing on the multi-stage filtered signal, which contains detailed adjustment information for each frequency band of the audio signal, such as the gain value, attenuation value, filtering slope, etc. of each frequency band. For example, in the equalizer settings of headphones or speakers, configuring according to the audio equalization parameters can make the output audio achieve the best effect in terms of sound quality, timbre, and listening experience.

[0061] As an embodiment of the present invention, the amplitude-frequency equalization processing of the corrected audio signal to obtain audio equalization parameters includes: identifying the frequency band energy distribution corresponding to the corrected audio signal; calculating the frequency band energy weights corresponding to each frequency band in the corrected audio signal based on the frequency band energy distribution; dynamically matching the frequency band energy weights with a preset target frequency response curve to generate a frequency band compensation coefficient; performing multi-stage recursive filtering on the corrected audio signal based on the frequency band compensation coefficient to obtain a multi-stage filtered signal; and performing amplitude-frequency equalization processing on the multi-stage filtered signal to obtain audio equalization parameters.

[0062] Among them, the frequency band energy distribution refers to the energy distribution of the corrected audio signal in different frequency intervals. For example, in a corrected audio signal containing music and speech, the low frequency band concentrates the energy of instruments such as drums, the middle frequency band contains the energy of human voices and most stringed instruments, and the high frequency band is the energy manifestation of instruments such as cymbals; the frequency band energy weight refers to a value used to measure the relative importance of the energy of each frequency band in the corrected audio signal in the overall energy, which reflects the contribution ratio of different frequency bands in the audio signal. For example, if the energy proportion of the low frequency band is large, its frequency band energy weight is high, indicating that this frequency band plays a key role in the overall listening experience of the audio; conversely, the weight of the frequency band with low energy is relatively low; the frequency band compensation coefficient refers to the parameter generated after dynamically matching the frequency band energy weight with a preset target frequency response curve. The target frequency response curve represents the desired ideal audio frequency response state. The frequency band compensation coefficient is used to indicate the degree of increase or attenuation required for each frequency band. For example, if there is a difference between the energy weight of a certain frequency band and the corresponding frequency band of the target frequency response curve, the frequency band compensation coefficient will determine the compensation amplitude for this frequency band according to the gap between the two. A positive number indicates that the energy needs to be increased, and a negative number indicates that the energy needs to be attenuated, so as to make the frequency response of the corrected audio signal closer to the ideal state. The multi-stage filtered signal refers to the signal obtained by performing multi-stage recursive filtering on the corrected audio signal based on the frequency band compensation coefficient. For example, first perform a filtering adjustment on the low frequency band, and then process the middle and high frequency bands in sequence, gradually correcting the energy distribution of each frequency band of the audio signal. Finally, the obtained multi-stage filtered signal is more in line with the target requirements in terms of frequency response.

[0063] Further, the identification of the frequency band energy distribution corresponding to the corrected audio signal can be achieved through a spectral energy analysis tool. For example, the energy integral of each critical band is calculated through a Bark scale filter bank to obtain the frequency band energy distribution. The calculation of the frequency band energy weight corresponding to each frequency band in the corrected audio signal can be achieved through an energy normalization algorithm. For example, weighting is performed by combining the frequency band energy ratio with an auditory perception model (such as the ERB scale) to obtain the frequency band energy weight. The dynamic matching of the frequency band energy weight with a preset target frequency response curve can be achieved through an adaptive equalization algorithm. For example, an LMS adaptive filter is used to adjust the gain deviation of each frequency band in real time to obtain a frequency band compensation coefficient. The multi-stage recursive filtering of the corrected audio signal can be achieved through digital signal processing technology. For example, cascaded recursive filtering processing is achieved through an IIR filter bank to obtain a multi-stage filtered signal. The amplitude-frequency equalization processing of the multi-stage filtered signal can be achieved through a parametric equalizer. For example, a PEQ parametric equalizer is used to adjust the amplitude response of each frequency band according to the auditory equal loudness curve to obtain audio equalization parameters.

[0064] S5. Based on the audio equalization parameters, the residual noise energy in the user's ear canal is monitored in real time, a noise regulation instruction corresponding to the residual noise energy is generated. Based on the noise regulation instruction, the acoustic sensor array is driven to perform adaptive sampling of the noise frequency to obtain a noise sampling log. Based on the noise sampling log, an adaptive noise reduction scheme corresponding to the target Bluetooth headset is synchronously generated.

[0065] Based on the audio equalization parameters of the present invention, the residual noise energy in the user's ear canal is monitored in real time, and the dynamic change of the noise in the ear canal after noise reduction can be accurately grasped. By combining the analysis of the audio equalization parameters and the real-time noise energy data, the noise reduction blind area or the over-noise reduction area can be timely discovered, providing data support for dynamically optimizing the noise reduction strategy.

[0066] Among them, the residual noise energy refers to the energy of the noise still existing in the ear canal after processing the sound in the ear canal based on the audio equalization parameters. Although most of the noise is reduced through audio equalization and noise reduction processing, due to the complexity of environmental noise and the limitations of noise reduction technology, there will always be some residual noise. The total energy of these residual noises is the residual noise energy.

[0067] As an embodiment of the present invention, the real-time monitoring of the residual noise energy in the user's ear canal based on the audio equalization parameters includes: extracting the sound pressure distribution characteristics in the user's ear canal based on the audio equalization parameters; analyzing the sound pressure energy threshold corresponding to the sound pressure distribution characteristics; screening the ear pressure frequency bands in the user's ear canal that exceed the sound pressure energy threshold; analyzing the energy attenuation trend corresponding to the ear pressure frequency bands; and real-time monitoring of the residual noise energy in the user's ear canal based on the energy attenuation trend.

[0068] Among them, the sound pressure distribution characteristics refer to the set of characteristics such as the magnitude, direction, and change law of the sound pressure at different positions in the user's ear canal. For example, at the entrance of the ear canal and near the eardrum, the sound pressure magnitudes may be different, and under the action of sounds of different frequencies, the sound pressure distributions at each position will also present different patterns. These characteristics together constitute the sound pressure distribution characteristics; the sound pressure energy threshold refers to the critical value determined according to the sound pressure distribution characteristics and used to measure the magnitude of the sound pressure energy. It represents a standard or boundary for distinguishing between normal and abnormal sound pressure energy levels. When the sound pressure energy in a certain area of the ear canal exceeds this threshold, it indicates that there may be strong noise interference or other acoustic abnormalities here; the ear pressure frequency band refers to the specific frequency range in the user's ear canal where the sound pressure energy exceeds the sound pressure energy threshold. For example, in a noisy industrial environment, the low-frequency noise generated by the operation of machines will cause the sound pressure energy in the low-frequency band of the ear canal to exceed the threshold, and at this time, this low-frequency band is the ear pressure frequency band, which reflects the frequency range in the ear canal that is significantly affected by noise; the energy attenuation trend refers to a description of the change of the sound pressure energy in the ear pressure frequency band over time or frequency, reflecting the trend and law of the weakening or reduction of the sound pressure energy in this frequency band. For example, during the operation of the noise reduction system, observing the energy attenuation trend of the ear pressure frequency band, if it is found that the energy attenuation is slow, it indicates that the current noise reduction measures have poor noise suppression effect on this frequency band and need to be further adjusted; if the energy decays rapidly, it indicates that the noise reduction measures are effective and can be continued or appropriately optimized.

[0069] Further, the extraction of the sound pressure distribution characteristics in the user's ear canal can be achieved through ear canal acoustic measurement techniques. For example, by collecting multi-point sound pressure data in the ear canal using a miniature probe microphone array, the sound pressure distribution characteristics can be obtained. The analysis of the sound pressure energy threshold corresponding to the sound pressure distribution characteristics can be achieved through a psychoacoustic model. For example, by using the ISO 226 equal-loudness curve combined with an individual hearing sensitivity test to determine the comfort threshold, the sound pressure energy threshold can be obtained. The screening of the ear pressure frequency bands exceeding the sound pressure energy threshold in the user's ear canal can be achieved through a frequency band detection algorithm. For example, by using a sliding window energy detector to identify the exceeding standard frequency bands, the ear pressure frequency bands can be obtained. The analysis of the energy attenuation trend corresponding to the ear pressure frequency bands can be achieved through a time-frequency analysis tool. For example, by using an exponential decay model to fit the energy attenuation curves of each frequency band, the energy attenuation trend can be obtained. The real-time monitoring of the residual noise energy in the user's ear canal can be achieved through a noise energy estimation algorithm. For example, by using the minimum statistics (MS) method to track the change of the background noise energy, the residual noise energy can be obtained.

[0070] By generating a noise control instruction corresponding to the residual noise energy, the present invention can accurately respond to the change of the residual noise in the ear canal, and this instruction can directly drive the noise reduction system to dynamically adjust the working parameters, timely and effectively suppressing the residual noise.

[0071] Among them, the noise control instruction refers to a series of control signals or parameter configuration information generated for actively adjusting the working state of the noise reduction system, which contains specific operation instructions for the noise reduction device. For example, when it is detected that the residual noise energy increases in the low frequency band, the noise control instruction will instruct the noise reduction system to increase the filtering intensity of the low frequency noise and quickly reduce the noise interference in this frequency band. Optionally, the generation of the noise control instruction corresponding to the residual noise energy can be achieved through an intelligent decision-making algorithm. For example, by using a fuzzy logic controller to map the residual noise energy to a dynamic noise reduction intensity parameter, the noise control instruction can be obtained.

[0072] Further, based on the noise control instruction, the present invention drives the acoustic sensor array to perform adaptive sampling of the noise frequency, obtains a noise sampling log, can accurately acquire complex and variable noise characteristics, and through instruction guidance, can focus on the key noise frequency bands, greatly improving the accurate efficiency of audio data sampling.

[0073] Among them, the noise sampling log refers to a record document generated by structurally integrating the collected noise frequency band data with the corresponding sampling tags, which details the complete information of each noise sampling in a standardized format, forming a system's noise data resource library. For example, engineers can analyze the noise sampling log to find the deficiencies of the noise reduction function in certain specific environments, and thus improve the algorithm targeted to enhance the noise reduction performance.

[0074] As an embodiment of the present invention, based on the noise regulation instruction, driving the acoustic sensor array to perform adaptive sampling of the noise frequency to obtain a noise sampling log, including: parsing the target noise reduction frequency band during the execution of the noise regulation instruction; configuring the corresponding directional sound pickup mode of the acoustic sensor array based on the target noise reduction frequency band; adjusting the noise sampling frequency corresponding to the acoustic sensor array based on the directional sound pickup mode; collecting the noise frequency band and sampling tags in the acoustic sensor array based on the noise sampling frequency; and generating a structured noise sampling log based on the noise frequency band and the sampling tags.

[0075] Among them, the target noise reduction frequency band refers to the noise frequency range that needs to be focused on by the noise reduction system currently parsed from the noise regulation instruction. For example, in a subway environment, if the low-frequency noise generated by wheel-rail friction has a greater impact on users, this low-frequency band will become the target noise reduction frequency band; the directional sound pickup mode refers to the working mode formed after parameter configuration of the acoustic sensor array based on the target noise reduction frequency band. In this mode, the sensor array will adjust its own sound pickup sensitivity and direction to make it more focused on collecting sound signals in the target noise reduction frequency band, enhancing the ability to capture noise in a specific direction and specific frequency band, and at the same time reducing interference from other irrelevant frequency band sounds; the noise sampling frequency refers to the time interval for the acoustic sensor array to collect noise signals or the number of sampling times per unit time, which is adjusted according to the directional sound pickup mode, and a suitable sampling frequency is set according to the characteristics of the target noise reduction frequency band. For high-frequency noise with rapid changes, a higher sampling frequency is required to completely capture its waveform details; while for low-frequency noise with slow changes, a lower sampling frequency can also meet the analysis requirements; the noise frequency band refers to the specific frequency range included in the noise signal collected by the acoustic sensor array. For example, the noise generated by mechanical equipment can be concentrated in a certain mid-high frequency band, while environmental noises such as wind noise can be distributed in a wider frequency band range; the sampling tag refers to the identification information added to each noise sampling data, including metadata such as sampling time, sampling location, sampling environment description, and device state during sampling. For example, through the time information in the sampling tag, the change law of noise at different times can be analyzed; the location information helps to understand the noise characteristics in different environments.

[0076] Further, the target noise reduction frequency band during the execution of the parsed noise regulation instruction can be achieved through an instruction decoding algorithm. For example, the frequency band parameter configuration in the instruction is extracted through a JSON parser to obtain the target noise reduction frequency band. The configured directional sound pickup mode corresponding to the acoustic sensor array can be achieved through beamforming technology. For example, the MVDR algorithm is used to optimize the directivity parameters of the microphone array to obtain the directional sound pickup mode. The adjustment of the noise sampling frequency corresponding to the acoustic sensor array can be achieved through dynamic sampling technology. For example, based on the Nyquist theorem, the sampling clock is dynamically adjusted through a PLL phase-locked loop to obtain the noise sampling frequency. The acquisition of the noise frequency band in the acoustic sensor array can be achieved through real-time spectrum analysis. For example, the Goertzel algorithm is used for selective sampling of specific frequency bands to obtain the noise frequency band. The acquisition of the sampling tag in the acoustic sensor array can be achieved through metadata tagging technology. For example, sampling context tags are automatically generated through timestamps and spatial coordinates to obtain the sampling tag. The generation of a structured noise sampling log can be achieved through a log formatting tool. For example, Apache Log4j is used to organize the sampling data according to a predefined template to obtain the noise sampling log.

[0077] Based on the noise sampling log, the present invention synchronously generates an adaptive noise reduction scheme corresponding to the target Bluetooth headset, which can accurately match the noise characteristics in the current environment according to real and detailed noise data, flexibly adjust the noise reduction parameters of the headset, and dynamically adapt to complex and changing noise scenarios.

[0078] Among them, the adaptive noise reduction scheme refers to a comprehensive strategy based on noise sampling log data that can dynamically adjust the noise reduction function of the Bluetooth headset according to the real-time changes in the user's environment. By analyzing information such as the noise frequency band, intensity, and change trend recorded in the log, it intelligently adjusts the noise reduction algorithm parameters inside the headset. For example, it adjusts the frequency response characteristics of the filter to accurately suppress noise in specific frequency bands and changes the noise reduction gain coefficient to control the noise reduction intensity. For example, in a noisy subway environment, the scheme will enhance the suppression of low-frequency wheel rail noise; while in an office environment, it will reduce the noise reduction intensity to avoid overly weakening human voice communication. This scheme will also cooperate to optimize the audio equalization settings to ensure the sound quality of useful audio signals such as music and voice while effectively reducing noise. Optionally, the synchronous generation of the adaptive noise reduction scheme corresponding to the target Bluetooth headset can be achieved through a rule inference engine. For example, based on the noise pattern-noise reduction strategy mapping rules in the expert knowledge base, the optimal scheme for the current environment is automatically matched to obtain the adaptive noise reduction scheme.

[0079] Compared with the problems described in the background art, the present invention can effectively improve the noise reduction effect of the earphone in different scenarios by obtaining the earphone ambient noise corresponding to the target Bluetooth earphone, significantly improve the call clarity and music playback quality, and also enhance the adaptability of the Bluetooth earphone in a dynamic noise environment. The present invention performs time-frequency joint analysis on the noise spectrogram to obtain spectral analysis parameters, and extracts the noise masking threshold in the spectral analysis parameters. It can intelligently distinguish useful audio signals from noise based on this threshold, avoiding excessive or insufficient noise reduction, so as to retain the true details of the audio signal to the greatest extent and significantly improve the clarity and comfort of the audio output. Further, the present invention performs noise suppression on the earphone audio signal based on a preset multi-order digital filter bank to obtain a primary noise reduction signal, which can perform refined processing on the signal from multiple dimensions according to the multi-frequency characteristics of complex environmental noise, effectively reducing the noise interference intensity, thereby further optimizing the earphone audio quality. Further, the present invention performs phase correction on the optimized noise reduction signal based on the frequency response matching degree to obtain a corrected audio signal, which can compensate for the phase deviation generated during the noise reduction process, making the phase relationship between the frequency components of the optimized noise reduction signal closer to the original audio and reducing the deterioration of the sound quality caused by phase distortion. Finally, the present invention monitors the residual noise energy in the user's ear canal in real time based on the audio equalization parameter, and can accurately grasp the dynamic change of the noise in the ear canal after noise reduction. By combining the analysis of the audio equalization parameter and the real-time noise energy data, it can timely detect the noise reduction blind area or the over-noise reduction area, providing data support for dynamically optimizing the noise reduction strategy. Therefore, the Bluetooth earphone ambient noise adaptive adjustment method and system provided by the embodiments of the present invention can improve the audio noise reduction effect of the Bluetooth earphone.

[0080] Embodiment 2: As Figure 2 shown, it is a functional module diagram of a Bluetooth earphone ambient noise adaptive adjustment system of the present invention.

[0081] The Bluetooth earphone ambient noise adaptive adjustment system 200 described in the present invention can be installed in an electronic device. According to the implemented functions, the Bluetooth earphone ambient noise adaptive adjustment system may include a spectrogram generation module 201, a signal acquisition module 202, a matching degree calculation module 203, an audio equalization module 204, and a solution generation module 205. The modules described in the present invention can also be referred to as units, which refer to a series of computer program segments that can be executed by the processor of the electronic device and can complete fixed functions, and are stored in the memory of the electronic device.

[0082] In the embodiments of the present invention, the functions of each module / unit are as follows: The spectrum diagram generation module 201 is configured to obtain the headphone ambient noise corresponding to the target Bluetooth headset, collect multi-band sound data corresponding to the headphone ambient noise by using a preset acoustic sensor array, and generate a noise spectrum diagram corresponding to the target ambient noise based on the multi-band sound data; The signal acquisition module 202 is configured to perform time-frequency joint analysis on the noise spectrum diagram to obtain spectrum analysis parameters, extract the noise masking threshold in the spectrum analysis parameters, and collect the headphone audio signal in the target ambient noise based on the noise masking threshold; The matching degree calculation module 203 is configured to perform noise suppression on the headphone audio signal based on a preset multi-order digital filter bank to obtain a primary noise reduction signal, calculate the signal-to-noise ratio corresponding to the primary noise reduction signal, perform optimization processing on the primary noise reduction signal based on the signal-to-noise ratio to obtain an optimized noise reduction signal, and calculate the frequency response matching degree between the optimized noise reduction signal and the original audio signal; The audio equalization module 204 is configured to perform phase correction on the optimized noise reduction signal based on the frequency response matching degree to obtain a corrected audio signal, and perform amplitude-frequency equalization processing on the corrected audio signal to obtain audio equalization parameters; The solution generation module 205 is configured to, based on the audio equalization parameters, monitor the residual noise energy in the user's ear canal in real time, generate a noise regulation instruction corresponding to the residual noise energy, drive the acoustic sensor array to perform adaptive sampling of the noise frequency based on the noise regulation instruction to obtain a noise sampling log, and synchronously generate an adaptive noise reduction solution corresponding to the target Bluetooth headset based on the noise sampling log.

[0083] Specifically, each module in the Bluetooth headset ambient noise adaptive adjustment system 200 in the embodiment of the present invention adopts the same technical means as those Figure 1 in the Bluetooth headset ambient noise adaptive adjustment method described above, and can produce the same technical effects, which will not be elaborated here.

[0084] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for adaptively adjusting ambient noise of a Bluetooth headset, characterized in that: The method comprises: Acquire the earphone environmental noise corresponding to the target Bluetooth earphone, collect multi-band sound data corresponding to the earphone environmental noise using a preset acoustic sensor array, and generate a noise spectrum corresponding to the target environmental noise based on the multi-band sound data; Performing a joint time-frequency analysis on the noise spectrum to obtain spectrum analysis parameters, extracting a noise masking threshold in the spectrum analysis parameters, and collecting an earphone audio signal in the target environmental noise based on the noise masking threshold; Based on a preset multi-order digital filter group, the headphone audio signal is subjected to noise suppression to obtain a primary noise reduction signal, a signal-to-noise ratio corresponding to the primary noise reduction signal is calculated, based on the signal-to-noise ratio, the primary noise reduction signal is subjected to optimization processing to obtain an optimized noise reduction signal, and a frequency response matching degree between the optimized noise reduction signal and the original audio signal is calculated; Based on the frequency response matching degree, performing phase correction on the optimized noise reduction signal to obtain a corrected audio signal, and performing amplitude-frequency equalization processing on the corrected audio signal to obtain an audio equalization parameter; Based on the audio equalization parameters, the residual noise energy in the user's ear canal is monitored in real time, and a noise control instruction corresponding to the residual noise energy is generated. Based on the noise control instruction, the acoustic sensor array is driven to perform adaptive sampling of the noise frequency to obtain a noise sampling log. Based on the noise sampling log, an adaptive noise reduction solution corresponding to the target Bluetooth headset is synchronously generated.

2. The method for adaptively adjusting environmental noise of a Bluetooth headset according to claim 1, characterized in that: The generating a noise spectrum diagram corresponding to the target environmental noise based on the multi-band sound data comprises: Determining noise analysis requirements corresponding to the multi-band sound data; Based on the noise analysis requirement, extracting a sound pressure signal from the target environmental noise; Performing time-frequency transformation on the sound pressure signal to obtain an initial spectrum matrix; Removing background interference components from the initial spectrum matrix to obtain optimized spectrum data; Based on the optimized spectrum data, a noise spectrum diagram corresponding to the target environmental noise is generated.

3. The method for adaptively adjusting environmental noise of a Bluetooth headset according to claim 1, characterized in that: The collecting the headphone audio signal in the target environmental noise based on the noise masking threshold comprises: According to the noise masking threshold, setting the audio collection range corresponding to the target Bluetooth headset; Configure the filtering parameters corresponding to the audio acquisition range; Based on the filtering parameters, extracting effective noise frames from the target environmental noise; Performing a superposition process on the effective noise frame to obtain a superposition frequency band sample; Based on the superimposed frequency band samples, an earphone audio signal in the target environmental noise is collected.

4. The method for adaptively adjusting environmental noise of a Bluetooth headset according to claim 1, characterized in that: The method of performing noise suppression on the headphone audio signal based on a preset multi-order digital filter group to obtain a primary noise reduction signal includes: Extracting time-frequency features from the headphone audio signal and constructing a time-frequency distribution matrix corresponding to the time-frequency features; Based on the time-frequency distribution matrix, identifying the noise-dominant frequency band in the headphone audio signal; Dynamically adjust the cutoff frequency parameters of the preset multi-order digital filter group according to the noise dominant frequency band; Using the adjusted multi-order digital filter group, the headphone audio signal is subjected to frequency band filtering to obtain a frequency domain filtered signal; Noise suppression is performed on the frequency domain filtered signal to obtain the primary noise reduction signal.

5. The method for adaptively adjusting environmental noise of a Bluetooth headset according to claim 1, characterized in that: The calculating the signal-to-noise ratio corresponding to the primary noise reduction signal includes: The signal-to-noise ratio corresponding to the primary noise reduction signal is calculated using the following formula: in, represents the signal-to-noise ratio corresponding to the primary noise reduction signal, and Respectively represent the start time and end time of the limited time interval, represents the time variable, represents the useful audio signal component in the primary noise reduction signal at the time variable, represents the residual noise signal component in the primary noise reduction signal as a time variable.

6. The method for adaptively adjusting environmental noise of a Bluetooth headset according to claim 1, characterized in that: The calculating the frequency response matching degree between the optimized noise reduction signal and the original audio signal comprises: The frequency response matching degree between the optimized noise reduction signal and the original audio signal is calculated using the following formula: in, represents the frequency response matching degree between the optimized noise reduction signal and the original audio signal, represents the total number of frequency components, Indicates the quantity index corresponding to the frequency component, represents the original discrete spectrum of the i-th frequency component in the original audio signal, represents the optimized discrete spectrum of the i-th frequency component in the optimized denoised signal.

7. The method for adaptively adjusting environmental noise of a Bluetooth headset according to claim 1, characterized in that: The step of performing amplitude-frequency equalization processing on the corrected audio signal to obtain audio equalization parameters includes: Identifying frequency band energy distribution corresponding to the corrected audio signal; Based on the frequency band energy distribution, calculating the frequency band energy weight corresponding to each frequency band in the corrected audio signal; Dynamically matching the frequency band energy weight with a preset target frequency response curve to generate a frequency band compensation coefficient; Based on the frequency band compensation coefficient, performing multi-stage recursive filtering on the corrected audio signal to obtain a multi-stage filtered signal; Perform amplitude-frequency equalization processing on the multi-stage filtered signal to obtain audio equalization parameters.

8. The method for adaptively adjusting environmental noise of a Bluetooth headset according to claim 1, characterized in that: The real-time monitoring of residual noise energy in the ear canal of the user based on the audio equalization parameter includes: Based on the audio equalization parameters, extracting sound pressure distribution characteristics in the user's ear canal; Analyzing the sound pressure energy threshold corresponding to the sound pressure distribution characteristics; Screening the ear pressure frequency band in the user's ear canal that exceeds the sound pressure energy threshold; Analyzing the energy attenuation trend corresponding to the ear pressure frequency band; Based on the energy attenuation trend, the residual noise energy in the ear canal of the user is monitored in real time.

9. The method for adaptively adjusting environmental noise of a Bluetooth headset according to claim 1, characterized in that: The step of driving the acoustic sensor array to perform adaptive sampling of noise frequency based on the noise control instruction to obtain a noise sampling log includes: Analyzing the target noise reduction frequency band of the noise control instruction during execution; Based on the target noise reduction frequency band, configuring a directional sound pickup pattern corresponding to the acoustic sensor array; Based on the directional sound pickup pattern, adjusting the noise sampling frequency corresponding to the acoustic sensor array; Based on the noise sampling frequency, collecting the noise frequency band and sampling labels in the acoustic sensor array; A structured noise sampling log is generated based on the noise frequency band and the sampling label.

10. A Bluetooth headset environmental noise adaptive adjustment system, characterized in that: The system comprises: A spectrum graph generation module is used to obtain the earphone environmental noise corresponding to the target Bluetooth earphone, collect multi-band sound data corresponding to the earphone environmental noise using a preset acoustic sensor array, and generate a noise spectrum graph corresponding to the target environmental noise based on the multi-band sound data; A signal acquisition module, configured to perform a time-frequency joint analysis on the noise spectrum to obtain spectrum analysis parameters, extract a noise masking threshold from the spectrum analysis parameters, and acquire an earphone audio signal from the target environmental noise based on the noise masking threshold; a matching degree calculation module, configured to perform noise suppression on the headphone audio signal based on a preset multi-order digital filter group to obtain a primary noise reduction signal, calculate a signal-to-noise ratio corresponding to the primary noise reduction signal, optimize the primary noise reduction signal based on the signal-to-noise ratio to obtain an optimized noise reduction signal, and calculate a frequency response matching degree between the optimized noise reduction signal and the original audio signal; An audio equalization module, configured to perform phase correction on the optimized noise reduction signal based on the frequency response matching degree to obtain a corrected audio signal, and perform amplitude-frequency equalization processing on the corrected audio signal to obtain an audio equalization parameter; A scheme generation module is used to monitor the residual noise energy in the user's ear canal in real time based on the audio equalization parameters, generate noise control instructions corresponding to the residual noise energy, drive the acoustic sensor array to perform adaptive sampling of noise frequency based on the noise control instructions, obtain a noise sampling log, and synchronously generate an adaptive noise reduction scheme corresponding to the target Bluetooth headset based on the noise sampling log.

Images