Adaptive noise reduction parameter adjustment method and system for a bluetooth headset

Abstract

The application discloses a kind of adaptive noise reduction parameter adjustment method and system of bluetooth headset, the method includes: during audio play, real-time audio content is carried out psychoacoustic masking threshold analysis, environmental noise is carried out spectrum analysis, determines joint spectrum gap, emits probe signal in spectrum gap, obtains the acoustic response signal of the feedforward microphone collection, obtains the acoustic response signal of the feedforward microphone collection;Acoustic response signal is compared with reference acoustic model, and acoustic path variation feature vector is extracted;Acoustic path variation feature vector is matched with the preset degradation type template library, and diagnostic conclusion is generated according to matching result, and diagnostic conclusion at least includes dominant degradation type and quantitative contribution;According to diagnostic conclusion, the coefficient of earphone active noise reduction filter is reconstructed;If matching result satisfies preset suspicious condition, then reduce noise reduction filter gain, to improve the accuracy and reliability of adaptive noise reduction.

Description

Technical Field

[0001] This invention relates to the field of Bluetooth headset noise reduction technology, specifically to a method and system for adjusting adaptive noise reduction parameters in Bluetooth headsets. Background Technology

[0002] During the use of active noise-canceling headphones, the feedforward microphone mesh may become progressively blocked due to the adhesion of foreign objects such as sweat and sebum, or the transfer function of the feedforward acoustic path may drift due to factors such as speaker diaphragm aging and temperature and humidity changes. This causes the preset noise reduction parameters to mismatch with the actual acoustic path, resulting in a decrease in noise reduction performance and making it difficult to achieve accurate adaptive noise reduction compensation. Summary of the Invention

[0003] To overcome the shortcomings of the prior art, the present invention provides an adaptive noise reduction parameter adjustment method and system for Bluetooth headsets to solve the problems in the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: An adaptive noise cancellation parameter adjustment method for Bluetooth earphones, applied to a Bluetooth earphone system, the Bluetooth earphone system including Bluetooth earphones and a charging case for storing the Bluetooth earphones, the charging case being provided with a reference speaker and a reference microphone; the method includes: During audio playback in the Bluetooth headset, the speaker of the Bluetooth headset is controlled to emit a detection signal, and the acoustic response signal collected by the feedforward microphone of the Bluetooth headset in response to the detection signal is acquired; wherein, the frequency band of the detection signal is dynamically configured within a spectral gap determined by the psychoacoustic masking threshold of the real-time audio content and the energy distribution of the ambient noise, and the emission frequency of the detection signal is dynamically adjusted according to the rate of change of physical state characterized by data collected by the physical environment sensor; Based on the acoustic response signal, the current preliminary degradation type is determined from a preset degradation type set and matched with the acoustic feature template corresponding to the preliminary degradation type to obtain the current matching degree. When the current matching degree does not fall into the preset signal interval, the current preliminary degradation type and its corresponding acoustic feature template are excluded from the degradation type set, and a new preliminary degradation type is re-determined from the remaining candidate degradation types and the matching is performed until the current matching degree falls into the preset signal interval. If there are no remaining candidate types in the degradation type set, it is determined that the preset doubt condition is met, the gain of the active noise cancellation filter in the Bluetooth headset is reduced, and a doubt prompt is sent to the user terminal communicating with the Bluetooth headset. When the charging case detects that the Bluetooth earphone has been stored, the charging case controls the Bluetooth earphone to perform a verification measurement and to review the diagnostic event corresponding to the question prompt. When the verification confirms the diagnostic conclusion is credible, the reduction of gain in the Bluetooth headset is lifted, and the noise reduction parameters of the Bluetooth headset are reconstructed according to the diagnostic conclusion.

[0005] In one embodiment, based on the acoustic response signal, the current preliminary degradation type is determined from a preset set of degradation types, and matched with the acoustic feature template corresponding to the preliminary degradation type to obtain the current matching degree, including: The acoustic response signal is compared with the reference acoustic model to extract the acoustic path variation feature vector; Calculate the weighted Euclidean distance between the acoustic path variation feature vector and the acoustic feature templates corresponding to each candidate degradation type in the degradation type set, and determine the candidate degradation type with the smallest distance as the initial degradation type; The calculated minimum weighted Euclidean distance or the correlation coefficient normalized based on that distance is used as the current matching degree.

[0006] In one embodiment, excluding the current initially determined degradation type and its corresponding acoustic feature template from the degradation type set, and re-determining a new initially determined degradation type from the remaining candidate degradation types, includes: Remove the acoustic feature templates corresponding to the candidate degradation types that have been identified as the preliminary degradation types from the degradation type set; In the set of degradation types after removal, the weighted Euclidean distance between the acoustic path variation feature vector and the acoustic feature template corresponding to each remaining candidate degradation type is recalculated; The remaining candidate degradation type with the smallest distance is determined as the new initial degradation type.

[0007] In one embodiment, reducing the gain of the active noise cancellation filter in the Bluetooth headset includes: The attenuation ratio is determined based on at least one of the following when the preset doubt condition is triggered: the total number of candidate degradation types excluded, the order of exclusion, and the current matching degree corresponding to the most recently determined initial degradation type. The higher the total number of excluded candidate degradation types, or the worse the matching degree of the first excluded initial degradation type, the higher the attenuation ratio. According to the attenuation ratio, the gain value of the active noise cancellation filter of the Bluetooth headset in the current noise cancellation frequency band is reduced.

[0008] In one embodiment, the charging case controls the Bluetooth headset to perform a calibration measurement, and reviews the diagnostic conclusion based on the results of the calibration measurement, including: The reference speaker inside the charging case is controlled to emit a verification signal, and the response of the verification signal is synchronously acquired through the feedforward microphone of the Bluetooth headset and the reference microphone inside the charging case; Based on the response acquired by the reference microphone, differential calibration is performed on the response acquired by the feedforward microphone to obtain the absolute acoustic path characteristics of the Bluetooth headset. The absolute acoustic path features are compared with the acoustic response features stored in the diagnostic event corresponding to the question mark; If the comparison results are inconsistent, the reference acoustic baseline of the Bluetooth headset will be updated based on the absolute acoustic path characteristics.

[0009] In one embodiment, the noise reduction parameters of the Bluetooth headset are reconstructed based on the diagnostic conclusion, including: Based on the quantitative contribution in the diagnostic conclusion, update the parameter values ​​of the corresponding physical components in the acoustic digital twin of the Bluetooth headset, wherein the acoustic digital twin is a model constructed using a lumped parameter equivalent circuit model that reflects the current acoustic state of the Bluetooth headset; Based on the updated acoustic digital twin, new noise reduction filter coefficients are generated under a hierarchical constraint optimization framework; The hierarchical constraint optimization framework consists of a first layer that generates a full-band gain upper limit envelope under closed-loop stability constraints, a second layer that determines the anchor point of the low-frequency noise reduction coefficient within the gain upper limit envelope with the objective of minimizing the psychoacoustic roughness of the predicted residual noise in the ear canal, and a third layer that maximizes the noise reduction depth of the target frequency band with the gain upper limit envelope and the anchor point as constraints.

[0010] In one embodiment, controlling the speaker of the Bluetooth headset to emit the detection signal includes: During audio playback in the Bluetooth headset, psychoacoustic masking threshold analysis is performed on the real-time audio content to determine the masking sound pressure level of the audio content in different frequency bands. Perform real-time spectrum analysis on environmental noise to determine the current noise energy distribution; Based on the masking sound pressure level and the noise energy distribution, a joint time spectrum is constructed, and spectral gaps in the joint time spectrum where the audio masking energy is lower than a first preset threshold and the noise energy is lower than a second preset threshold are located. The frequency band of the detection signal is dynamically configured within the spectral gap; The physical environment data collected by the physical environment sensor of the Bluetooth headset is acquired, and the transmission frequency of the detection signal is dynamically adjusted according to the rate of change of physical state represented by the physical environment data.

[0011] In one embodiment, it further includes: When the reviewed diagnostic conclusion indicates that the feedforward microphone is in a progressive semi-blocked state and its quantitative contribution exceeds a preset threshold, the hearing enhancement filter coefficients are calculated using the identified blockage equivalent acoustic model. Switch the Bluetooth headset from active noise cancellation mode to transparency enhancement mode configured by the transparency enhancement filter coefficients.

[0012] A Bluetooth headset includes a speaker, a feedforward microphone, a headset processor, and a memory; The memory stores instructions that, when executed by the earphone processor, cause the Bluetooth earphone to perform the steps described above.

[0013] A Bluetooth headset system includes Bluetooth headsets as described above, and a charging case for storing the Bluetooth headsets; The charging case is equipped with a reference speaker and a reference microphone. The charging case is configured to perform the steps in the above method by the charging case after detecting that the Bluetooth headset has been stored.

[0014] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention deploys a reference speaker and a reference microphone within the charging case. After the earphones are stored, the closed cavity, free from environmental interference, controls the reference speaker to emit a verification signal. The feedforward and reference microphones simultaneously acquire the response and perform differential calibration to obtain the absolute acoustic path characteristics of the earphones. This absolute acoustic path characteristic is compared with the acoustic response characteristics in online diagnostics of questionable events, allowing for high-precision verification of the online diagnostic conclusions. Based on the diagnostic conclusions, the coefficients of the earphones' active noise cancellation filter are reconstructed; if the matching result meets preset questionable conditions, the noise cancellation filter gain is reduced, thereby improving the accuracy and reliability of adaptive noise cancellation. Attached Figure Description

[0015] Figure 1 A flowchart illustrating an adaptive noise reduction parameter adjustment method for Bluetooth headphones provided by the present invention; Figure 2 A schematic diagram illustrating the working principle of a Bluetooth headset system provided by this invention; Figure 3 A schematic diagram illustrating the working principle of one of the Bluetooth earphones and charging cases provided by this invention; Figure 4 This is a schematic diagram illustrating the working principle of one of the Bluetooth headsets and smartphones provided by the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] One embodiment of the present invention provides a Bluetooth headset system, including Bluetooth headsets and a charging case. The Bluetooth headset is an in-ear or semi-in-ear active noise-canceling headset, which internally houses a speaker, a feedforward microphone, a headset processor, and a memory. In addition to conventional charging and storage functions, the charging case internally houses a reference speaker and a reference microphone for performing high-precision acoustic calibration measurements after the headset is stored.

[0018] The adaptive noise reduction parameter adjustment method performed by the above Bluetooth headset system will be explained in detail below.

[0019] like Figures 1 to 4 As shown, the present invention provides a method for adjusting adaptive noise reduction parameters of Bluetooth headphones, comprising the following steps: Step S100: During audio playback in the Bluetooth headset, control the speaker of the Bluetooth headset to emit a detection signal and acquire the acoustic response signal collected by the feedforward microphone of the Bluetooth headset in response to the detection signal.

[0020] During audio playback, the Bluetooth headset's speaker emits a detection signal. This signal propagates through the air in the ear canal or external space and is collected by a feedforward microphone to form an acoustic response signal. By dynamically determining the optimal detection signal frequency band and emission frequency under the current audio and environmental noise background, a high signal-to-noise ratio detection that is imperceptible to the human ear can be achieved.

[0021] The detection signal can be a Chirp signal with a logarithmic frequency sweep from 22kHz to 23.5kHz, a duration of 4ms, and a peak amplitude calibrated to 30dBSPL.

[0022] Step S200: Perform iterative matching and exclusion diagnosis based on the acoustic response signal to initially determine the degradation type.

[0023] Specifically, the headphone processor has a pre-defined set of degradation types, including various candidate degradation types, such as: sebum-induced semi-permeable membrane blockage, speaker diaphragm aging and collapse, dust adhesion to the pressure relief hole, and temperature and humidity drift. Each degradation type corresponds to a laboratory-calibrated acoustic feature template, denoted as the template vector candidate template. The diagnostic process is as follows: S210. Compare the acoustic response signal obtained in step S100 with the reference acoustic model to extract the acoustic path variation feature vector. The reference acoustic model is the initial acoustic transfer function model established when the headphones leave the factory. The comparison method is to compare the amplitude and phase of the acoustic response signal with the reference acoustic model at each frequency point, and extract features such as amplitude deviation, phase deviation, and high-frequency roll-off slope change at each frequency point to form the acoustic path variation feature vector.

[0024] In one embodiment, the reference acoustic model is collaboratively corrected using dual-source calibration data, which includes: First calibration data (hard benchmark): After the Bluetooth earbuds are placed in the charging case, the charging case controls its built-in reference speaker to emit an exponentially swept frequency signal from 20Hz to 24kHz as a calibration signal. This exponentially swept frequency signal is synchronously acquired by the reference microphone in the charging case and the feedforward microphone of the earbuds, obtaining the reference channel response and the earbud feedforward channel response, respectively. Using the absolute acoustic characteristics of the reference microphone calibrated in an anechoic chamber at the factory as a benchmark, the ratio of the transfer function of the earbud feedforward channel response to the reference channel response is calculated. After deconvolution operation to eliminate the influence of the inconsistency in the frequency response of the reference speaker itself, the absolute acoustic transfer function of the earbud feedforward path is obtained, which serves as the first calibration data.

[0025] Second calibration data (soft compensation): Through the communication connection between the headphones and the user terminal, common aging and erosion model data is received from the common aging and erosion model database on the cloud server. This model is established based on long-term acoustic drift data of a group of headphones of the same model and is performed using statistical regression methods. The model takes the cumulative usage time of the headphones, the historical ambient temperature and humidity, and the number of charging cycles as input parameters, and outputs the expected acoustic drift compensation curve as the second calibration data.

[0026] The initial reference acoustic model is corrected by weighted fusion of the first and second calibration data to obtain the current reference acoustic model. The weighting coefficients are dynamically determined based on the signal-to-noise ratio of the first calibration data and the cumulative usage time of the headphones. A hard baseline provides a precise absolute acoustic path reference, while soft compensation provides statistically significant aging drift compensation. Together, they eliminate reference baseline drift caused by non-blocking factors such as system aging.

[0027] S220. Calculate the weighted Euclidean distance between the acoustic path variation feature vector and all candidate templates in the degradation type set. The weights of each feature component are pre-calibrated based on the discriminative power of its corresponding physical degradation type. Select the candidate degradation type with the smallest weighted Euclidean distance as the initial degradation type.

[0028] S230. Determine whether the minimum weighted Euclidean distance falls within a preset confidence interval (e.g., minimum weighted Euclidean distance < threshold). If it falls within the interval, the matching confidence level is met; if it does not fall within the interval, the matching confidence level is insufficient.

[0029] S240. If the minimum weighted Euclidean distance does not fall within the confidence interval, it indicates that the confidence level of this match is insufficient. In this case, the currently selected candidate type template will be excluded from the set of degenerate types in this round of diagnosis.

[0030] S250. Repeat steps S220-S230 from the remaining candidate types to find the type with the smallest distance and determine the confidence level, forming an iterative closed loop of matching-exclusion-rematching.

[0031] S260. When the minimum weighted Euclidean distance of a certain match falls into the preset confidence interval for the first time, the current preliminary degradation type is determined as the dominant degradation type in the diagnostic conclusion, and the quantitative contribution in the diagnostic conclusion is updated with reference to the quantitative information reflected by the matching degree.

[0032] S270. If, during the iteration process, all candidate types in the degradation type set are eliminated one by one, and no candidate types remain, the diagnosis is terminated, and the preset suspicious condition is determined to be met. At this time, all acoustic variation feature vectors and other data during the iteration process are encapsulated as suspicious records, directly triggering the safety retreat processing in step S300.

[0033] Taking the degradation type of "sebum-type semipermeable membrane blockage" as an example, the process of establishing its acoustic feature template is as follows: Calibration Phase: In a laboratory environment, simulated sebum of varying thicknesses (equivalent thicknesses of 5μm, 10μm, 15μm, 20μm, 25μm, and 30μm) was quantitatively coated onto the mesh of a standard feedforward microphone. After each coating, a logarithmic sweep signal from 22kHz to 23.5kHz was emitted by the speaker of the Bluetooth headset, and the acoustic response was acquired by the feedforward microphone. The "high-frequency roll-off slope" (unit: dB / oct) and the "phase change" of a specific frequency band were extracted to construct a multi-dimensional acoustic variation feature vector at that thickness. The feature vectors of multiple sets of different thicknesses were statistically averaged to obtain the acoustic feature template corresponding to the degradation type of "sebum-type semi-permeable membrane blockage".

[0034] During the online usage phase: the measured acoustic variation feature vector extracted from the current acoustic response signal is used to calculate the distance and match it with all templates, including acoustic feature templates.

[0035] It's important to note that candidate degradation types do not directly correspond to physical causes like "seal blockage" or "speaker aging," but rather to clustering categories of "equivalent acoustic transfer function change patterns." Specifically, by pre-collecting a large amount of real or simulated acoustic degradation data, unsupervised clustering (such as K-Means clustering) is performed on the acoustic path variation feature vectors (including amplitude deviation, phase deviation, and high-frequency roll-off slope changes at each frequency point). This divides the degradation "patterns" into several clusters, with the center of each cluster serving as an acoustic feature template for a candidate degradation type. During diagnosis, only the most similar degradation pattern cluster needs to be matched, without needing to analyze its physical causes, to estimate the quantitative impact of the degradation pattern on noise reduction performance for subsequent parameter reconstruction. This significantly reduces the complexity of the diagnostic algorithm and its reliance on prior physical knowledge.

[0036] This invention constructs a pre-defined set of acoustic degradation types and employs an iterative elimination method for diagnosis. When the confidence of the currently best-matching degradation type is insufficient, it is excluded and a rematch is initiated until a high-confidence type is found or a doubtful criterion is triggered due to the exclusion of all types. This method abandons the reliance on a single threshold judgment and approaches the truth through continuous falsification, significantly improving the robustness and credibility of diagnostic conclusions in complex environments.

[0037] Step S300: When there are no remaining candidate types in the degradation type set and the preset doubt condition is met, the gain of the active noise cancellation filter in the Bluetooth headset is reduced, and a doubt prompt is sent to the user terminal that is connected to the Bluetooth headset.

[0038] At this point, actively reducing the gain of the noise reduction filter allows the system to retreat to a safe and conservative state, fundamentally eliminating the risk of performance degradation and system instability caused by diagnostic uncertainties.

[0039] Simultaneously, a potential issue warning is sent to the user's terminal (such as a smartphone) connected to the earphones. The warning might read, "The noise cancellation system has detected a possible microphone malfunction; it is recommended to return the earphones to the charging case for automatic calibration," guiding the user to return the earphones to the charging case for offline verification.

[0040] In one embodiment, reducing the gain of the active noise cancellation filter in the Bluetooth headset includes: determining an attenuation ratio based on the minimum weighted Euclidean distance corresponding to the most recently determined initial degradation type; the worse the matching degree (i.e., the larger the minimum weighted Euclidean distance), the higher the attenuation ratio; and reducing the gain value of the active noise cancellation filter in the current noise cancellation frequency band of the Bluetooth headset according to the attenuation ratio.

[0041] For example, an exemplary mapping relationship is as follows: When the matching degree is slightly above the threshold to 1.5 times the threshold, the decay rate is 10%; When the matching degree is between 1.5 and 2.5 times the threshold, the attenuation rate is 25%. When the matching degree is between 2.5 times and 4 times the threshold, the attenuation rate is 50%. When the matching degree is greater than or equal to 4 times the threshold, the attenuation ratio is 75%.

[0042] For example, in the low-to-mid frequency band of 100Hz to 2kHz, the gain value of each frequency point of the filter is multiplied by (1 - attenuation ratio). For high-frequency bands above 2kHz, since the high-frequency gain is inherently low and the risk of high-frequency howling is small, the original gain value can be kept unchanged, and attenuation is only applied to the low-to-mid frequency band. This attenuation concession process, by actively reducing the noise reduction depth, gains an increase in the open-loop transfer function gain margin of the system, thereby effectively preventing positive feedback self-oscillation (howling) introduced due to unknown and drastic changes in the feedforward path transfer function, ensuring the hearing safety of the wearer during periods of uncertain diagnosis.

[0043] When all candidate degradation types are excluded and the diagnostic conclusion has high uncertainty, this invention does not force a low-confidence diagnostic result. Instead, the Bluetooth headset actively reduces the gain of the active noise cancellation filter, yielding to a safe and conservative state with sufficient gain margin. This attenuation yielding process sacrifices the noise cancellation depth for an increase in the open-loop transfer function gain margin, effectively preventing positive feedback self-oscillation and howling caused by unknown and drastic changes in the feedforward path transfer function. Simultaneously, by sending a questioning prompt to the user terminal, the user is guided to return the headset to the charging case for offline verification, thus achieving proactive safety handling when the diagnosis is questionable. The exclusion of all candidate degradation types serves as a clear and verifiable condition for triggering the safety yielding, avoiding the forced output of incorrect conclusions under uncertainty and ensuring system safety from the source of diagnosis.

[0044] Step S400: When the charging case detects that the Bluetooth earphone has been stored, the charging case controls the Bluetooth earphone to perform a calibration measurement, and reviews the diagnostic conclusion based on the result of the calibration measurement.

[0045] Specifically, the charging case utilizes its internally deployed reference speaker and microphone to perform calibration measurements in a stored state free from environmental interference. This process upgrades the charging case from a simple charging storage device into a diagnostic verification and calibration station, allowing users to trigger high-precision offline verification through routine storage actions without additional intervention.

[0046] In one embodiment, the charging case controls the Bluetooth headset to perform a calibration measurement, and reviews the diagnostic conclusion based on the results of the calibration measurement, including: S410 controls the reference speaker inside the charging case to emit a verification signal, and synchronously acquires the response of the verification signal through the feedforward microphone of the Bluetooth earbuds and the reference microphone inside the charging case. When the earbuds are placed in the charging case, the charging communication contacts or proximity sensor of the charging case detect the earbud insertion event. After detection, the charging case sends a verification trigger command to the earbuds via the charging communication contacts or Bluetooth communication. The verification signal is, for example, an exponential sweep signal from 20Hz to 24kHz, with a duration of 200ms and a sound pressure level of 60dB SPL.

[0047] S420: Based on the response acquired by the reference microphone, differential calibration is performed on the response acquired by the feedforward microphone to obtain the absolute acoustic path characteristics of the Bluetooth headset. The differential calibration is calculated by multiplying the ratio of the feedforward microphone transfer function to the reference microphone transfer function by the standard transfer function from the reference microphone to the feedforward microphone calibrated at the factory of the charging case, thereby eliminating the influence of coupling of the charging case shell and individual differences of the speakers.

[0048] S430. Compare the absolute acoustic path features with the acoustic response features in the diagnostic events corresponding to the questionable prompts. The comparison method can be frequency domain correlation coefficient calculation, with a correlation coefficient threshold of 0.85. When the correlation coefficient is greater than or equal to the threshold, the verification result is considered consistent with the online diagnostic conclusion, and the diagnostic conclusion is reliable; when the correlation coefficient is less than the threshold, the conclusion is considered inconsistent.

[0049] If the comparison results are consistent, the diagnostic conclusion is confirmed to be reliable, the gain reduction is removed, and the parameter reconstruction in step S500 is performed.

[0050] S440. If the comparison results are inconsistent, the reference acoustic baseline of the Bluetooth headset will be updated with weights based on the absolute acoustic path features. The update method is to weight and fuse the absolute acoustic path features with the original reference acoustic baseline. The weight coefficients are determined based on the signal-to-noise ratio measured by the charging case calibration (the higher the signal-to-noise ratio, the greater the weight of the new measurement). The updated reference acoustic baseline will serve as the comparison benchmark for subsequent online diagnostics.

[0051] This invention deploys a reference speaker and a reference microphone within the charging case. After the earphones are stored, the closed cavity, free from environmental interference, controls the reference speaker to emit a verification signal. The feedforward and reference microphones simultaneously acquire the response and perform differential calibration to obtain the absolute acoustic path characteristics of the earphones. This absolute acoustic path characteristic is compared with the acoustic response characteristics in questionable events during online diagnostics, allowing for high-precision verification of the online diagnostic conclusions. When the comparison is consistent, the diagnostic conclusion is confirmed as reliable, and the safety margin is lifted. When the comparison is inconsistent, the online reference acoustic baseline is updated with offline high-precision measurement results, ensuring the accuracy of subsequent online diagnostics. This not only provides a high-confidence truth verification method for online diagnostics but also enables the reference acoustic baseline to be continuously corrected throughout the earphones' lifespan, overcoming the problem of reference baseline drift during long-term use. Users can trigger this calibration process imperceptibly through daily storage, requiring no additional operation, significantly improving the system's long-term reliability and ease of use.

[0052] Step S500: When the diagnostic conclusion is verified to be reliable, the gain reduction of the Bluetooth headset is removed, and the noise reduction parameters of the Bluetooth headset are reconstructed according to the diagnostic conclusion. Specifically, this includes: S510. Based on the quantitative contribution in the diagnostic conclusion, update the parameter values ​​of the corresponding physical components in the acoustic digital twin of the Bluetooth headset. The Bluetooth headset maintains an acoustic digital twin. This digital twin uses a lumped-parameter equivalent circuit model, where the blocked portion of the feedforward microphone mesh is represented by a T-type two-port network. The variable resistor in the network corresponds to the equivalent acoustic impedance, and the variable capacitor corresponds to the equivalent acoustic reactance (capacitive). Map the quantitative contribution in the diagnostic conclusion to the corresponding parameters of the digital twin. For example, if the diagnostic conclusion is "sebum-type semi-permeable membrane blockage, equivalent blockage layer thickness 15μm, equivalent acoustic impedance 380MKS Rayls, equivalent acoustic reactance -55MKS Rayls", then update the variable resistor to the resistance value corresponding to 380MKS Rayls, and update the variable capacitor to the capacitance value corresponding to -55MKS Rayls.

[0053] S520, based on the updated acoustic digital twin, generates new noise reduction filter coefficients under a hierarchical constraint optimization framework. The hierarchical constraint optimization framework consists of three layers: The first layer generates the full-band gain upper limit envelope under closed-loop stability constraints. Specifically, a closed-loop system state-space model is constructed based on the updated digital twin. By analyzing the eigenvalue trajectories of the system matrix and using structured singular value analysis, the maximum permissible gain envelope across the entire frequency band is generated. This envelope represents the upper limit of the amplification factor that ensures the system does not experience self-oscillation at the corresponding frequency points.

[0054] The second layer determines the anchor point for the low-frequency noise reduction coefficients within the upper gain envelope, aiming to minimize the psychoacoustic roughness of the predicted residual noise in the ear canal. Specifically, the psychoacoustic roughness is calculated using the Zwicker psychoacoustic model, and the specific process is as follows: In the updated acoustic digital twin, a typical urban traffic noise (sound pressure level 65 dBSPL) is simulated as input, and the residual noise spectrum at the ear canal reference point is predicted after the noise is processed by the current noise reduction filter.

[0055] The residual noise spectrum is mapped to the Bark domain, and critical frequency bands are divided with a step size of 1 Bark. The time envelope of the residual noise signal in each critical frequency band is extracted, its modulation spectrum is analyzed, and the modulation depth in the modulation frequency range of 15Hz to 300Hz is calculated.

[0056] The modulation depths of each critical frequency band are weighted and summed according to the Zwicker roughness formula to obtain the psychoacoustic roughness prediction value R, in asper. For example: in, For modulation frequency, Let be the change in masking depth within the z-th critical frequency band. Let be the step size for integrating the Bark domain.

[0057] The optimization objective is to minimize the predicted psychoacoustic roughness R, and the center frequency of the noise reduction filter in the low-frequency range (20Hz to 500Hz) is used as the optimization target. The gain G and quality factor Q are the optimization variables, and gradient descent is used for iterative search. When the value of R converges to its minimum, at this point... G and Q are the anchor points for the low-frequency noise reduction coefficient.

[0058] The third layer uses the upper gain envelope and the low-frequency anchor point determined by the second layer as joint constraints to maximize the noise reduction depth within the target noise reduction frequency band (500Hz to 2kHz). Specifically, the maximum gain envelope and anchor point parameters are transformed into frequency domain inequality constraints to construct a constrained optimization problem, which is solved using the primordial-dual interior-point method. After frequency domain gradient iterative convergence, the new noise reduction filter coefficients for the target frequency band are obtained.

[0059] After confirming the diagnostic conclusion, this invention first updates the parameter values ​​of the corresponding physical components in the acoustic digital twin. Then, it generates new noise reduction filter coefficients under a hierarchical constraint optimization framework: the first layer generates a full-band gain upper limit envelope under closed-loop stability constraints to ensure that the system does not experience self-excited oscillations; the second layer determines the low-frequency noise reduction coefficient anchor point within the gain upper limit envelope with the goal of minimizing the psychoacoustic roughness of the predicted residual noise in the ear canal, ensuring auditory comfort of the residual noise; the third layer uses the outputs of the first two layers as joint constraints to maximize the noise reduction depth within the target frequency band. This hierarchical progressive optimization framework effectively balances the relationship between stability, comfort, and noise reduction depth, enabling the reconstructed noise reduction parameters to approach the optimal performance under the current physical state while ensuring safety and comfort.

[0060] When reconstructing noise reduction parameters based on diagnostic conclusions, this invention employs a three-layer progressive optimization architecture: The first layer generates a full-band gain upper limit envelope under closed-loop stability constraints based on the updated acoustic digital twin, physically ensuring that the reconstructed filter will not induce self-oscillation; the second layer determines low-frequency noise reduction coefficient anchor points within the gain envelope with the goal of minimizing the psychoacoustic roughness of residual noise in the ear canal, ensuring the subjective listening comfort of residual noise; the third layer maximizes the noise reduction depth of the target frequency band under the joint constraints of the gain envelope and the low-frequency anchor points. This hierarchical framework unfolds stability constraints, psychoacoustic comfort optimization, and noise reduction performance maximization in a logical order, allowing each optimization objective to be fully solved within its respective level. This avoids the problem of sacrificing stability and noise reduction depth that may occur in traditional single-optimizer schemes where multiple objectives are directly weighted, ensuring that the final generated filter coefficients achieve an optimal balance between safety, comfort, and noise reduction effect.

[0061] In one embodiment, the frequency band of the detection signal emitted by the speaker of the Bluetooth headset in step S100 can be determined using the dynamic configuration method of this embodiment. Before acquiring the acoustic response signal, the method further includes: S110. During audio playback in the Bluetooth headset, perform psychoacoustic masking threshold analysis on the real-time audio content to determine the masking sound pressure level of the audio content in different frequency bands. Specifically, by using a standard psychoacoustic model, perform psychoacoustic masking threshold analysis continuously at 1ms frame lengths to calculate the masking sound pressure level of the audio content in different critical frequency bands. The masking threshold indicates that signals with a sound pressure level lower than the threshold in that frequency band will be masked by the audio content and become inaudible to the human ear.

[0062] S120. Perform real-time spectrum analysis on the ambient noise to determine the current noise energy distribution. Specifically, obtain the energy distribution of the current ambient noise in each frequency band by performing real-time spectrum analysis on the feedforward microphone signal.

[0063] S130. Based on the masking sound pressure level and noise energy distribution, a joint time-spectrum is constructed, whereby the joint time-spectrum is used to identify frequency bands that are imperceptible to the human ear and where noise interference is below a threshold. Within the joint time-spectrum, spectral gaps are located where the audio masking energy is below a first preset threshold (e.g., -55 dBFS) and the noise energy is below a second preset threshold (e.g., -45 dBFS). When the bandwidth of the connected frequency band that satisfies both conditions is greater than the bandwidth required by the detection signal (e.g., 1.5 kHz), the frequency band is determined as a usable spectral gap.

[0064] S140. The frequency band of the probe signal used to acquire the acoustic response signal is dynamically configured within the spectral slot. For example, the default probe frequency band is 22-23.5kHz. When this band is occupied, it can be shifted to an adjacent spectral slot such as 20-21.5kHz or 23.5-25kHz. The start and end points of the probe signal use a cosine square fade-in / fade-out envelope to reduce spectral leakage.

[0065] In this embodiment, the spectrum gap is located by dual constraints of audio masking threshold and noise energy, and the detection signal is dynamically configured in a frequency band that is imperceptible to the user and has a high signal-to-noise ratio, thus achieving high signal-to-noise ratio detection that is imperceptible to the human ear.

[0066] In one embodiment, prior to acquiring the acoustic response signal, the method further includes: S150: Acquire physical environment data collected by the physical environment sensors of the Bluetooth headset, including at least one of a humidity sensor and a motion sensor. The humidity sensor is used to detect changes in relative humidity near the ear canal, and the motion sensor (such as an inertial measurement unit, IMU) is used to detect the user's step frequency and exercise intensity.

[0067] The transmission frequency of the detection signal is dynamically adjusted based on the rate of change of the physical state as represented by the physical environment data.

[0068] One exemplary mapping scheme is as follows: When the humidity change rate is less than 2%RH / min and the user is in a static / low-intensity state, the detection frequency is once every 2000ms; When the humidity change rate is 2%-5%RH / min and the user is under moderate intensity, the detection frequency is once every 1000ms; When the humidity change rate is 5%-10%RH / min and the user is in a medium-to-high intensity state, the detection frequency is once every 500ms; When the humidity change rate exceeds 10%RH / min or the user is under high intensity, the detection frequency is once every 300ms.

[0069] The more drastic the change in physical state, the higher the detection frequency, in order to capture the rapid evolution of physical state in real time; the more stable the state, the sparser the detection, in order to reduce system power consumption.

[0070] In one embodiment, the present invention also provides at least one of the following layered collaborative implementation methods to replace complex real-time psychoacoustic calculations performed directly in the Bluetooth headset. For example: On-device scene lookup table method: A multi-dimensional lookup table is pre-stored in the Bluetooth headset's memory. This lookup table is indexed by "audio content type" and "ambient noise scene," with each index storing the offline pre-calculated optimal detection signal frequency band configuration and the corresponding initial gradient of noise reduction parameters. The Bluetooth headset only needs to run a lightweight scene classifier to identify the current scene in real time, and then directly look up the configuration parameters in the table, avoiding the need for online calculation of the complete psychoacoustic masking threshold.

[0071] like Figure 3 Mobile-side assisted computation: Utilizing the processing power of the user terminal connected to the Bluetooth headset. The Bluetooth headset transmits real-time collected audio content feature packets and environmental noise feature packets to the user terminal via Bluetooth Low Energy. The accompanying application on the user terminal is responsible for performing the complete calculations of psychoacoustic masking threshold analysis, environmental noise spectrum analysis, joint temporal spectrum construction, and spectrum gap localization, and sends back the calculated optimal detection signal frequency band, bandwidth, and transmission frequency, etc., to the Bluetooth headset. The headset only receives and applies the configuration parameters.

[0072] Lightweight Neural Network Inference on the End-User Side: A pruned and quantized lightweight convolutional recurrent neural network is deployed on the Bluetooth headset. This network directly takes the Mel spectrograms of the original audio and noise as input and outputs in parallel the recommended center frequency and bandwidth of the probe signal, the estimated psychoacoustic roughness of the predicted residual noise in the ear, and the recommended adjustment of the noise reduction filter coefficients. The inference process is only the forward propagation of the neural network, and the computational cost is far lower than that of traditional regular psychoacoustic models and gradient descent iterative optimization.

[0073] Before acquiring the acoustic response signal, this invention performs psychoacoustic masking threshold analysis and real-time environmental noise spectrum analysis on the real-time audio content. It then emits a detection signal within a spectral gap where both audio masking energy and noise energy are low, making the detection signal imperceptible to the user's subjective hearing. Simultaneously, based on the rate of change of physical states represented by data such as humidity and motion status collected by physical environment sensors, the transmission frequency of the detection signal is dynamically adjusted. The detection frequency is increased when physical states change drastically to capture rapid evolution in real time, and decreased when the state is stable to save power. While ensuring detection quality, this invention balances real-time monitoring and system energy efficiency, significantly improving the user experience and engineering practicality of online diagnostics.

[0074] In one embodiment, it further includes: S610. When the reviewed diagnostic conclusion indicates that the feedforward microphone is in a progressively semi-blocked state and its quantitative contribution exceeds a preset threshold, the hearing enhancement filter coefficients are calculated using the identified blockage equivalent acoustic model. A progressively semi-blocked state refers to the physical process in which the feedforward microphone mesh is gradually adhered to by external materials, leading to a progressive degradation of its acoustic transmission characteristics. Preset thresholds include, for example, a blockage level reaching severe blockage (e.g., level 4 or above), an equivalent blockage layer thickness exceeding 35 μm, and a noise reduction depth attenuation below 3 dB.

[0075] The equivalent acoustic model of the obstruction describes its acoustic transmission characteristics, primarily exhibiting a low-pass filtering effect, with significantly greater attenuation of high-frequency ambient sounds than low-frequency sounds. The specific steps for calculating the hearing-enhancing filter coefficients are as follows: S611. Establish the transfer function of the blockage based on the acoustic impedance and acoustic impedance values ​​of the blockage equivalent acoustic model; S612. Calculate the inverse filter of the transfer function to compensate for high-frequency attenuation caused by blockage; S613. Cascade the inverse filter with the preset ambient sound transparency filter to obtain a transparency enhancement filter.

[0076] S620 switches the Bluetooth headphones from active noise cancellation mode to transparency enhancement mode, configured by the transparency enhancement filter coefficients. The mode switching can be performed by gradually changing the critical band interpolation in the Bark sensing frequency domain, with a transition time of approximately 100ms, to avoid audible noise impact and sudden changes in ear pressure during the switching process.

[0077] In transparency enhancement mode, the headphones use a feedforward microphone to collect ambient sound signals, which are then processed by a transparency enhancement filter and played to the user through the speaker. The headphones also compensate for the attenuation of high-frequency ambient sounds caused by blockage.

[0078] In this embodiment, when the noise reduction function reaches its physical limit, it actively switches from a noise reducer to an environmental perception enhancer, ensuring that users still have the necessary environmental sound perception ability in extreme congestion conditions, thus achieving graceful degradation when the function fails.

[0079] When the charging case confirms offline that the feedforward microphone is in a progressively semi-blocked state and the blockage exceeds the physical limit of electroacoustic compensation, this invention uses the identified blockage equivalent acoustic model to calculate the transparency enhancement filter coefficients, switching the headphones from active noise cancellation mode to transparency enhancement mode. This mode compensates for the attenuation of high-frequency ambient sounds caused by blockage, allowing users to still perceive high-frequency warning signals such as vehicle horns and alarms. This mechanism avoids the ear-blocking effect and loss of environmental perception caused by continuing to operate malfunctioning noise cancellation when the blockage is severe, providing a safety net from the perspective of functional failure and ensuring user safety during extreme usage scenarios.

[0080] Please see Figure 1-4 A Bluetooth headset includes: a speaker, a feedforward microphone, a headset processor, and a memory. The headset processor is a digital signal processor or an ARM processor, which runs a diagnostic model, executes diagnostic algorithms, and calculates noise reduction filter coefficients. The memory stores instruction programs, an acoustic degradation mode library, acoustic feature template parameters, a reference acoustic baseline, acoustic digital twin parameters, and previous diagnostic records.

[0081] In one embodiment, the device further includes: a user terminal communication module, which may be a low-power Bluetooth communication module, for exchanging data with the user terminal, sending question prompts and receiving cloud calibration data; a feedback microphone for collecting secondary acoustic path response and residual noise in the ear canal; and a physical environment sensor, including at least one of a humidity sensor and a motion sensor, for collecting physical environment data.

[0082] When the instructions stored in the memory are executed by the headphone processor, the Bluetooth headset performs the steps described above that are executed by the Bluetooth headset.

[0083] Please see Figure 1-4 A Bluetooth headset system includes Bluetooth headsets as described above, and a charging case for storing the Bluetooth headsets.

[0084] In addition to the standard charging circuitry and storage cavity, the charging case also includes a reference speaker and a reference microphone. The reference speaker, which can be a miniature balanced armature speaker, is positioned at the left and right ear positions within the charging case's storage cavity, with a transmission frequency range of 20Hz to 24kHz. The reference microphone, which can be a MEMS microphone, is positioned corresponding to the reference speaker and is used to receive verification signal responses.

[0085] The charging case is equipped with an in-case processor, which is configured to perform any of the steps in the above method performed by the charging case after detecting that the Bluetooth headset has been stored, including: controlling the reference speaker to emit a verification signal, synchronously acquiring the response through the headset's feedforward microphone and the reference microphone, performing differential calibration calculation to obtain absolute acoustic path features, comparing the absolute acoustic path features with the acoustic response features in the suspicious diagnostic events uploaded by the Bluetooth headset, and updating the reference acoustic baseline stored on the headset based on the comparison results.

[0086] In one embodiment, the charging case is also provided with a first communication interface, which is a charging communication contact or a short-range communication module, used to establish data communication with the Bluetooth headset after the Bluetooth headset is placed in the case, read suspicious records, and write back model correction parameters.

[0087] Correspondingly, the Bluetooth headset is equipped with a second communication interface that is electrically or verbally connected to the first communication interface of the charging case. The second communication interface can be a corresponding charging communication contact or a short-range communication module. The second communication interface of the Bluetooth headset is connected to the headset processor of the Bluetooth headset for establishing data communication with the headset processor.

[0088] In this embodiment, by configuring a reference acoustic system and verification processing capability for the charging case, the charging case is upgraded from a single-function device for charging and storage to a multi-functional acoustic maintenance station that integrates charging and storage, offline acoustic calibration, and diagnostic verification and correction. Together with the Bluetooth headset, it achieves full lifecycle acoustic health management and adaptive noise reduction parameter adjustment.

[0089] The foregoing description of specific exemplary embodiments of the present invention is for illustrative and explanatory purposes. These descriptions are not intended to limit the invention to the precise forms disclosed, and it is obvious that many changes and variations can be made based on the above teachings. Although embodiments of the invention have been shown and described, these specific embodiments are merely explanations of the invention and are not intended to limit it. The specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. The purpose of selecting and describing exemplary embodiments is to explain the specific principles of the invention and its practical application, so that those skilled in the art, after reading this specification, can make modifications, substitutions, variations, and various choices and changes to the embodiments as needed without departing from the principles and spirit of the invention, provided that such modifications, substitutions, variations, and choices and changes are within the scope of the claims of the invention and are protected by patent law.

Claims

1. A method for adjusting adaptive noise reduction parameters of a Bluetooth headset, applied to a Bluetooth headset system, the Bluetooth headset system comprising a Bluetooth headset and a charging case for storing the Bluetooth headset, characterized in that, The charging compartment is equipped with a reference speaker and a reference microphone; the method includes: During audio playback in the Bluetooth headset, the speaker of the Bluetooth headset is controlled to emit a detection signal, and the acoustic response signal collected by the feedforward microphone of the Bluetooth headset in response to the detection signal is acquired; wherein, the frequency band of the detection signal is dynamically configured within a spectral gap determined by the psychoacoustic masking threshold of the real-time audio content and the energy distribution of the ambient noise, and the emission frequency of the detection signal is dynamically adjusted according to the rate of change of physical state characterized by data collected by the physical environment sensor; Based on the acoustic response signal, the current preliminary degradation type is determined from a preset degradation type set and matched with the acoustic feature template corresponding to the preliminary degradation type to obtain the current matching degree. When the current matching degree does not fall into the preset signal interval, the current preliminary degradation type and its corresponding acoustic feature template are excluded from the degradation type set, and a new preliminary degradation type is re-determined from the remaining candidate degradation types and the matching is performed until the current matching degree falls into the preset signal interval. If there are no remaining candidate types in the degradation type set, it is determined that the preset doubt condition is met, the gain of the active noise cancellation filter in the Bluetooth headset is reduced, and a doubt prompt is sent to the user terminal communicating with the Bluetooth headset. When the charging case detects that the Bluetooth earphone has been stored, the charging case controls the Bluetooth earphone to perform a verification measurement and to review the diagnostic event corresponding to the question prompt. When the verification confirms the diagnostic conclusion is credible, the reduction of gain in the Bluetooth headset is lifted, and the noise reduction parameters of the Bluetooth headset are reconstructed according to the diagnostic conclusion.

2. The adaptive noise reduction parameter adjustment method for Bluetooth headphones according to claim 1, characterized in that, Based on the acoustic response signal, the current preliminary degradation type is determined from a preset set of degradation types, and matched with the acoustic feature template corresponding to the preliminary degradation type to obtain the current matching degree, including: The acoustic response signal is compared with the reference acoustic model to extract the acoustic path variation feature vector; Calculate the weighted Euclidean distance between the acoustic path variation feature vector and the acoustic feature templates corresponding to each candidate degradation type in the degradation type set, and determine the candidate degradation type with the smallest distance as the initial degradation type; The calculated minimum weighted Euclidean distance or the correlation coefficient normalized based on that distance is used as the current matching degree.

3. The adaptive noise reduction parameter adjustment method for Bluetooth headphones according to claim 2, characterized in that, Excluding the current preliminary degradation type and its corresponding acoustic feature template from the set of degradation types, and re-determining a new preliminary degradation type from the remaining candidate degradation types, including: Remove the acoustic feature templates corresponding to the candidate degradation types that have been identified as the preliminary degradation types from the degradation type set; In the set of degradation types after removal, the weighted Euclidean distance between the acoustic path variation feature vector and the acoustic feature template corresponding to each remaining candidate degradation type is recalculated; The remaining candidate degradation type with the smallest distance is determined as the new initial degradation type.

4. The adaptive noise reduction parameter adjustment method for Bluetooth headphones according to claim 1, characterized in that, Reducing the gain of the active noise cancellation filter in the Bluetooth headset includes: The attenuation ratio is determined based on at least one of the following when the preset doubt condition is triggered: the total number of candidate degradation types excluded, the order of exclusion, and the current matching degree corresponding to the most recently determined initial degradation type. The higher the total number of excluded candidate degradation types, or the worse the matching degree of the first excluded initial degradation type, the higher the attenuation ratio. According to the attenuation ratio, the gain value of the active noise cancellation filter of the Bluetooth headset in the current noise cancellation frequency band is reduced.

5. The adaptive noise reduction parameter adjustment method for Bluetooth headphones according to claim 1, characterized in that, The charging case controls the Bluetooth headset to perform a calibration measurement, and the diagnostic conclusion is reviewed based on the results of the calibration measurement, including: The reference speaker inside the charging case is controlled to emit a verification signal, and the response of the verification signal is synchronously acquired through the feedforward microphone of the Bluetooth headset and the reference microphone inside the charging case; Based on the response acquired by the reference microphone, differential calibration is performed on the response acquired by the feedforward microphone to obtain the absolute acoustic path characteristics of the Bluetooth headset. The absolute acoustic path features are compared with the acoustic response features stored in the diagnostic event corresponding to the question mark; If the comparison results are inconsistent, the reference acoustic baseline of the Bluetooth headset will be updated based on the absolute acoustic path characteristics.

6. The adaptive noise reduction parameter adjustment method for Bluetooth headphones according to claim 1, characterized in that, Based on the diagnostic conclusion, the noise reduction parameters of the Bluetooth headset are reconstructed, including: Based on the quantitative contribution in the diagnostic conclusion, update the parameter values ​​of the corresponding physical components in the acoustic digital twin of the Bluetooth headset, wherein the acoustic digital twin is a model constructed using a lumped parameter equivalent circuit model that reflects the current acoustic state of the Bluetooth headset; Based on the updated acoustic digital twin, new noise reduction filter coefficients are generated under a hierarchical constraint optimization framework; The hierarchical constraint optimization framework consists of a first layer that generates a full-band gain upper limit envelope under closed-loop stability constraints, a second layer that determines the anchor point of the low-frequency noise reduction coefficient within the gain upper limit envelope with the objective of minimizing the psychoacoustic roughness of the predicted residual noise in the ear canal, and a third layer that maximizes the noise reduction depth of the target frequency band with the gain upper limit envelope and the anchor point as constraints.

7. The adaptive noise reduction parameter adjustment method for Bluetooth headphones according to claim 1, characterized in that, Controlling the speaker of the Bluetooth headset to emit the detection signal includes: During audio playback in the Bluetooth headset, psychoacoustic masking threshold analysis is performed on the real-time audio content to determine the masking sound pressure level of the audio content in different frequency bands. Perform real-time spectrum analysis on environmental noise to determine the current noise energy distribution; Based on the masking sound pressure level and the noise energy distribution, a joint time spectrum is constructed, and spectral gaps in the joint time spectrum where the audio masking energy is lower than a first preset threshold and the noise energy is lower than a second preset threshold are located. The frequency band of the detection signal is dynamically configured within the spectral gap; The physical environment data collected by the physical environment sensor of the Bluetooth headset is acquired, and the transmission frequency of the detection signal is dynamically adjusted according to the rate of change of physical state represented by the physical environment data.

8. The adaptive noise reduction parameter adjustment method for Bluetooth headphones according to claim 5, characterized in that, Also includes: When the reviewed diagnostic conclusion indicates that the feedforward microphone is in a progressive semi-blocked state and its quantitative contribution exceeds a preset threshold, the hearing enhancement filter coefficients are calculated using the identified blockage equivalent acoustic model. Switch the Bluetooth headset from active noise cancellation mode to transparency enhancement mode configured by the transparency enhancement filter coefficients.

9. A Bluetooth headset, characterized in that, Includes speakers, feedforward microphones, headphone processors, and memory; The memory stores instructions that, when executed by the earphone processor, cause the Bluetooth earphone to perform the steps of the method as described in any one of claims 1 to 8.

10. A Bluetooth headset system, characterized in that, Includes the Bluetooth headset as described in claim 9, and a charging case for storing the Bluetooth headset; The charging case is provided with a reference speaker and a reference microphone, and the charging case is configured to perform the steps performed by the charging case in any one of claims 1 to 8 after detecting that the Bluetooth headset has been stored.

Images