An ANC and talk algorithm cooperative balance control method and device

Abstract

This invention discloses a method and device for coordinated balanced control of active noise cancellation (ANC) and call audio pickup algorithms. The method includes: simultaneously acquiring multiple raw acoustic signals from inside and outside the ear canal, and decoupling and identifying environmental noise separately; performing self-calibration of the head-related transfer function (HRTF); establishing a bidirectional closed-loop parameter feedback channel between the ANC module and the call audio pickup module, enabling bidirectional closed-loop parameter feedback between the two data streams; the ANC module determining the interference intensity of environmental noise based on the feedback results and performing layered constraint filtering; acquiring independent weights and corresponding parameters for multiple modes, and calculating the overall control parameters based on the weights and parameters of each mode; and synchronously executing noise reduction in the ANC module and call audio enhancement in the call audio pickup module based on the overall control parameters. Through the implementation of the above scheme, this solution can achieve bidirectional coordinated balance between active noise cancellation and call audio pickup under high wind noise and various wearing conditions.

Description

Technical Field

[0001] This invention relates to the field of audio signal processing technology, and in particular to a method, device, headset, and readable storage medium for coordinated balance control of ANC and call algorithm. Background Technology

[0002] Currently, mid-to-high-end TWS (True Wireless Stereo) earbuds generally use a feedforward microphone (FF) + feedback microphone (FB) to achieve an adaptive ANC (Active Noise Cancellation) module, coupled with an independent ENC (Environment Noise Cancellation) call noise reduction algorithm to complete voice calls. The mainstream control logic on the market can be divided into the following three categories: 1. Qualcomm Integrated Platform Solution In this type of solution, the ANC module sends noise reduction level markers unidirectionally to the cVc (Clear Voice Capture) call algorithm, without a mechanism for the call algorithm to control the ANC in reverse. In scenarios with strong winds or air leakage in the ear canal, the low-frequency gain of the ANC is only reduced by a fixed level, and a table is used to adapt to human differences by looking up the static three-level wearing parameters. In addition, this type of solution only distinguishes between environmental noise and wind noise, and cannot identify the low-frequency harmonic interference generated by the ANC itself.

[0003] 2. Sony and Apple's self-developed chip solutions In this type of solution, unidirectional linkage is achieved by sharing audio data streams with hardware, and the ANC output amplitude is limited by a fixed curve; the fit depends on factory preset parameters and cannot be corrected in real time for the user's personalized ear canal transfer function; in extreme wind scenarios, the transparent noise cancellation mode is forcibly switched, sacrificing noise cancellation capabilities to ensure call quality.

[0004] 3. Other chip platforms or third-party call algorithm solutions In this type of solution, the ANC module and the call algorithm operate completely independently, with only the upper-layer firmware transmitting the ANC level signal unidirectionally; there is no bidirectional parameter interaction channel; the algorithm cannot distinguish between three types of interference: ANC harmonics, airflow noise, and ear canal leakage resonance; under different wearing tightness, ear shape, and air leakage levels, the ANC reverse sound wave enters the call microphone, causing voice suppression, swallowing, and low-frequency buzzing; in high wind noise scenarios, the computing power of the two sets of adaptive filters competes and converges, resulting in a call signal-to-noise ratio fluctuation of 8~15dB, leading to a significant difference in user experience.

[0005] Therefore, it is necessary to improve the current collaborative control method of ANC and call algorithm. Summary of the Invention

[0006] To address at least one of the aforementioned technical problems, the main objective of this invention is to provide a method, apparatus, and readable storage medium for coordinated balance control of ANC and call algorithms.

[0007] To achieve the above objectives, one technical solution adopted by the present invention is: providing a method for coordinated balance control of ANC and call algorithm, comprising: Simultaneously acquire multiple raw acoustic signals from inside and outside the ear canal, and decouple and identify environmental noise in the multiple raw acoustic signals respectively; The Head-Related Transfer Function (HRTF) is self-calibrated. The HRTF is calculated in real time based on the frequency sweep test signal. The HRTF is then used to synchronously correct the filtering convergence coefficient of the ANC module and the voice fusion weight of the call ENC module. A bidirectional closed-loop parameter feedback channel is established between the ANC module and the call ENC module to enable bidirectional closed-loop parameter feedback between the two data streams of the ANC module and the call ENC module. The ANC module determines the interference intensity of environmental noise based on the feedback results and performs hierarchical constraint filtering. Obtain the independent weights and corresponding parameters of multiple modes, and calculate the total control parameters based on the weights and parameters of each mode; The noise reduction of the ANC module and the voice enhancement of the call ENC module are executed synchronously according to the overall control parameters.

[0008] The process of simultaneously acquiring multiple raw acoustic signals from inside and outside the ear canal, and decoupling and identifying environmental noise in these signals, specifically includes: Simultaneously acquire three raw acoustic signals collected by the FF feedforward microphone inside and outside the ear canal, the FB feedback microphone, and the COM call microphone; A lightweight decision tree classification model is constructed based on the time-domain energy fluctuation, frequency-domain coherence, and resonance peak characteristics of the original acoustic signal to separate the three types of noise interference and the energy proportion of the interference noise in the original acoustic signal. Independent frequency band suppression commands are generated for each type of interference and sent to the call ENC module.

[0009] The three types of interference include Type A interference, Type B interference, and Type C interference. Type A interference is ANC periodic low-frequency harmonics, Type B interference is airflow turbulence noise, and Type C interference is ear canal leakage resonance noise.

[0010] Before calculating the Head-Related Transfer Function (HRTF) in real time based on the frequency sweep test signal, the method further includes automatically triggering micro-calibration of the HRTF when any one of the following four conditions is met: completion of headphone wearing recognition, establishment of call link, sudden change in wind noise level, or ANC filter convergence deviation exceeding the threshold.

[0011] The step of calculating the Head-Related Transfer Function (HRTF) in real time based on the frequency sweep test signal, and using the HRTF to synchronously correct the filtering convergence coefficient of the ANC module and the voice fusion weight of the call ENC module, further includes: Head-related transfer function (HRTF) deviation detection is triggered every set number of frames. If the head-related transfer function deviation exceeds the threshold, micro-calibration is initiated.

[0012] The ANC and call algorithm collaborative balance control method as described in claim 2 is characterized in that the bidirectional data flow between the ANC module and the call ENC module realizes bidirectional closed-loop parameter feedback, and the ANC module performs hierarchical constraint filtering based on the interference intensity fed back by the call ENC module, specifically including: The ANC module outputs a forward data stream to the call ENC module in real time. The forward data stream includes five sets of dynamic characteristic quantities: FF real-time filter gain, FB loop residual, ear canal leakage coefficient, reverse sound wave output amplitude, and filter adaptive step size. The call ENC algorithm adjusts the voice fusion weight of the microphone based on the forward data stream and outputs the reverse data stream to the ANC module in real time. The reverse data stream includes four call quality evaluation indicators: uplink voice SNR, ANC harmonic interference energy, airflow noise level and voice distortion. The ANC module determines the interference intensity of environmental noise based on four call quality assessment indicators and the energy proportion of interference noise, and performs layered constraint filtering based on the interference intensity.

[0013] The interference intensity includes mild interference, moderate interference, and severe interference. The hierarchical constraint filtering based on interference intensity specifically includes: When the interference intensity is determined to be mild, only the maximum output amplitude of the low-frequency ANC in the 50-600Hz range is compressed, while the mid-to-high frequency noise reduction is fully preserved. When the interference intensity is determined to be moderate interference, the fixed coefficients of the FB feedback filter are locked, and the frequency of the FF feedforward adaptive iterative update is reduced. When the interference intensity is determined to be severe interference, the low-frequency noise reduction gain is smoothly reduced to maintain narrowband low-frequency boom suppression, without switching to transparency mode or turning off the ANC module.

[0014] To achieve the above objectives, another technical solution adopted by the present invention is: providing an ANC and call algorithm collaborative balance control device, comprising: The acquisition module is used to simultaneously acquire multiple raw acoustic signals from inside and outside the ear canal, and to decouple and identify environmental noise in the multiple raw acoustic signals respectively. The self-calibration module is used to self-calibrate the Head-Related Transfer Function (HRTF). It calculates the HRTF in real time based on the frequency sweep test signal and uses the HRTF to synchronously correct the filtering convergence coefficient of the ANC module and the voice fusion weight of the call ENC module. The bidirectional closed-loop feedback module is used to establish a bidirectional closed-loop parameter feedback channel between the ANC module and the call ENC module, so that the bidirectional data streams of the ANC module and the call ENC module can achieve bidirectional closed-loop parameter feedback. The ANC module determines the interference intensity of environmental noise based on the feedback results and performs hierarchical constraint filtering. The weight fusion module is used to obtain the independent weights and corresponding parameters of multiple modes, and calculate the total control parameters based on the weights and parameters of each mode. The processing module is used to synchronously execute the noise reduction of the ANC module and the call voice enhancement of the ENC module according to the overall control parameters.

[0015] To achieve the above objectives, another technical solution adopted by the present invention is as follows: providing an earphone, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the steps in the above method.

[0016] To achieve the above objectives, another technical solution adopted by the present invention is to provide a readable storage medium on which a computer program is stored, wherein when the computer program is executed by a processor, the steps in the above method are implemented.

[0017] The technical solution of this invention acquires multiple original acoustic signals from inside and outside the ear canal simultaneously and decouples and identifies environmental noise in these signals. It performs self-calibration of the Head Correlation Transfer Function (HRTF), calculates the HRTF in real time based on the frequency sweep test signal, and uses the HRTF to synchronously correct the filtering convergence coefficient of the ANC module and the voice fusion weight of the call ENC module. A bidirectional closed-loop parameter feedback channel is established between the ANC module and the call ENC module, enabling bidirectional data flow between them to achieve bidirectional closed-loop parameter feedback. The ANC module determines the interference intensity of environmental noise based on the feedback results and performs layered constraint filtering. Independent weights and corresponding parameters for multiple modes are acquired, and the overall control parameters are calculated based on the weights and parameters of each mode. The overall control parameters are used to synchronously execute noise reduction in the ANC module and call voice enhancement in the call ENC module. This achieves a bidirectional coordinated balance between active noise reduction and call pickup under various wearing conditions, including high wind noise and different user wearing conditions. The system does not need to automatically shut down the ANC throughout the process, significantly reducing the difference in call signal-to-noise ratio caused by different user wearing conditions, thereby improving the user experience. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a method for ANC and call algorithm collaborative balance control according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating an example of a collaborative balancing control method for ANC and call algorithm according to an embodiment of the present invention; Figure 3 This is a block diagram of a collaborative balance control device for ANC and call algorithm according to an embodiment of the present invention; Figure 4 This is a block diagram of a headphone module according to an embodiment of the present invention.

[0020] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] 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 a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0022] It should be noted that the descriptions involving "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Furthermore, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.

[0023] Unlike existing TWS earphones that suffer from drawbacks such as one-way linkage between the ANC module and call algorithm, coarse noise recognition, static wear adaptation, and abrupt mode switching, this solution proposes a collaborative and balanced control method for ANC and call algorithm. This method achieves a two-way collaborative balance between active noise cancellation and call pickup under various wearing conditions, including high wind noise and different user experiences, without adding new hardware sensors or significantly increasing DSP computing power. The system does not require automatic ANC shutdown throughout the process, greatly reducing the difference in call signal-to-noise ratio caused by different user wearing styles, thereby improving the user experience. The specific method flow of this collaborative and balanced control method for ANC and call algorithm is described in the following embodiment.

[0024] Please refer to Figure 1 and Figure 2 , Figure 1 This is a flowchart of a method for ANC and call algorithm collaborative balance control according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating an example of an ANC and call algorithm collaborative balancing control method according to an embodiment of the present invention. In this embodiment, the ANC and call algorithm collaborative balancing control method includes steps S110-S150: S110: Simultaneously acquire multiple raw acoustic signals from inside and outside the ear canal, and decouple and identify environmental noise in each of these signals. Multiple raw acoustic signals can be acquired through a multi-microphone acquisition module in the earphone. Specifically, the multi-microphone acquisition module includes an FF feedforward microphone, an FB feedback microphone, and a COM call microphone, which are used to acquire the three raw acoustic signals from inside and outside the ear canal in real time and in parallel. After acquiring the raw acoustic signals, they can be received by a three-dimensional noise decoupling and identification module. A lightweight decision tree is used to identify environmental noise, obtaining various types of environmental noise and their energy percentages. Corresponding suppression commands can be generated based on the types of environmental noise for further processing by the call ENC module.

[0025] S120. Self-calibrate the Head-Related Transfer Function (HRTF). Calculate the HRTF in real-time based on the sweep test signal, and use the HRTF to synchronously correct the filtering convergence coefficient of the ANC module and the voice fusion weights of the call ENC module. Specifically, the HRTF online self-calibration module multiplexes the headphone speaker output calibration sweep signal, receives multi-microphone acquisition response data, calculates and updates the personalized HRTF, and outputs correction parameters to the ANC module and the call ENC module to synchronously correct the filtering convergence coefficient of the ANC module and the internal microphone voice fusion weights of the call ENC module, thus achieving HRTF self-calibration. The updated HRTF represents the ear canal's response characteristics to various frequencies in the form of a frequency domain complex transfer function. The correction is divided into two steps: first, correcting the ANC filter by substituting the new HRTF into the secondary path model of the ANC feedback loop and recalculating the convergence coefficient of the FF / FB filter, making the reverse sound wave generated by the ANC more accurately match the current ear canal acoustic characteristics. Secondly, the ENC voice fusion weights have been corrected. HRTF reflects the actual proportion of the wearer's own voice entering the ear canal via bone conduction and air conduction. The ENC dynamically adjusts the fusion weights of the signals collected by the FB internal microphone (feedback microphone) based on this proportion—when HRTF high-frequency attenuation is severe, the gain compensation of the internal microphone's voice signal is increased. The call ENC module has a built-in call algorithm. In addition, the headset locally stores multiple sets of user-historical HRTF templates, enabling quick matching on the next wear and shortening the calibration time to less than 8ms.

[0026] S130. Establish a bidirectional closed-loop parameter feedback channel between the ANC and the call ENC algorithms to enable bidirectional closed-loop parameter feedback between the ANC module and the call ENC algorithm. The ANC module determines the interference intensity of environmental noise based on the feedback results and performs layered constraint filtering. Establishing a bidirectional closed-loop parameter feedback channel between the ANC module and the call ENC module through the ANC-ENC bidirectional closed-loop feedback engine enables bidirectional closed-loop parameter feedback between the data of the ANC module and the ENC module. Specifically, firstly, the ANC module outputs a positive data stream to the call ENC module in real time; secondly, the ENC module dynamically adjusts the AEC echo cancellation threshold, NS noise suppression floor, and voice protection gain based on the positive data stream to distinguish between ANC harmonics and far-end call echoes, avoiding excessive suppression of local voices. Finally, the ENC module generates a reverse data stream and outputs it to the ANC module in real time. The ANC module determines the interference intensity of environmental noise based on the reverse data stream and performs layered fine-grained constraint filtering based on the interference intensity. The interference intensity can be divided into multiple levels to facilitate fine-grained control operations based on the corresponding levels.

[0027] S140. Obtain the independent weights and corresponding parameters of multiple modes, and calculate the total control parameters based on the weights and parameters of each mode. Through the multi-mode weight fusion interaction module, the MCU in the earphone can connect with the APP of the smart terminal to obtain the independent weights and corresponding parameters of multiple modes sent by the APP. The multiple modes include three types: intelligent balance mode, outdoor call mode, and ultra-clear mode. The corresponding weights for the three are: W1 intelligent balance weight, W2 outdoor call weight, and W3 ultra-clear weight, respectively. The total control parameters are calculated using a weighted fusion formula, specifically: Total parameters = W1 × Smart strategy parameters + W2 × Outdoor strategy parameters + W3 × Ultra-clear parameters.

[0028] All three modes share the same core logic of bidirectional closed-loop feedback, three-dimensional noise decoupling, and HRTF self-calibration, eliminating the need for hard reset of mode switching parameters. Only when the user manually increases the W3 Extreme Clarity weight to a preset threshold will the system smoothly reduce ANC output until noise reduction is disabled. As a manual fallback strategy, the system will not actively disable ANC noise reduction in automatic operation. The intelligent strategy parameters, outdoor strategy parameters, and Extreme Clarity parameters are not arbitrary but rather parameterized configuration sets for the three core logics of three-dimensional noise decoupling, bidirectional closed-loop feedback, and HRTF self-calibration under different target weights. In other words, the same underlying algorithm presets three different initial values ​​for filter coefficients, gain limiting values, update frequencies, etc., for the three targets of "intelligent balance," "outdoor call," and "extreme clarity." The W1 / W2 / W3 weights issued by the APP are weighted fusion coefficients of these three preset parameters, not the parameters themselves. The above three strategy parameters are preset configuration sets for the underlying core algorithm under different optimization targets.

[0029] The three modes—Intelligent Balance, Outdoor Call, and Ultimate Clarity—are not three independent algorithms, but rather share the same underlying core engine (i.e., three-dimensional noise decoupling, bidirectional closed-loop feedback, and HRTF self-calibration). When switching modes, no filter coefficients or algorithm states are reset; only the weighted values ​​of W1 / W2 / W3 in the weighted fusion formula are adjusted to change the system's preference between "noise reduction depth" and "voice clarity," resulting in a smooth and abrupt transition.

[0030] S150. Based on the overall control parameters, the noise reduction of the ANC module and the voice enhancement of the call ENC module are executed synchronously. The weighted and fused joint control parameters are loaded through the DSP module, and FF-FB hybrid ANC and ENC voice enhancement are run synchronously. Specifically, the DSP executes two processing links in parallel within the same clock cycle: one is the ANC link: FF / FB microphone signal → adaptive filter → generation of reverse acoustic wave → speaker output; the other is the ENC link: COM call microphone signal → AEC echo cancellation → multi-band NS noise suppression → beamforming voice enhancement → uplink transmission. The two links share the parameters transmitted by the bidirectional feedback engine, but the filtering operations are independent and parallel, with no computational power contention. The acoustic indicators of the three microphones, call SNR, and HRTF deviation values ​​are continuously collected and closed-loop correction parameters are applied. The cycle period of this closed-loop correction is once per frame (5~10ms). 1. Acquisition: Acquire the acoustic signal of the current frame using three microphones; 2. Evaluation: ENC calculates indicators such as call SNR and harmonic interference energy; 3. Feedback: The indicator is sent to the ANC control unit via a reverse data stream; 4. Adjustment: The ANC adjusts the filtering parameters according to the interference intensity; 5. Positive: ANC will resend the adjusted feature values ​​to ENC; 6. Iteration: Enter the next frame loop. After the Bluetooth headset call ends, it smoothly restores the user's original ANC noise cancellation level before the call, without any abrupt auditory changes.

[0031] The aforementioned DSP module carries out all computational tasks, including ANC adaptive filtering, bidirectional feedback engine, noise decoupling, and HRTF calibration. The call ENC module receives noise suppression commands, ANC positive feature quantities, and HRTF human voice fusion weights to complete uplink voice noise reduction and enhancement.

[0032] Compared with existing technologies, this solution has at least the following advantages: 1. Pioneering bidirectional closed-loop parameter feedback architecture for ANC and ENC Breaking away from the industry's inherent logic of "one-way ANC output and passive ENC adaptation," this solution adds an ENC reverse constraint ANC filtering mechanism. It finely adjusts ANC low-frequency gain, FB coefficient, and iteration step size based on real-time voice quality, rather than abruptly switching gears or disabling noise reduction. No similar two-way closed-loop interactive solution exists on the market. 2. The three-dimensional noise decoupling and classification recognition mechanism differs from existing binary noise recognition. It independently distinguishes three types of superimposed interference: ANC harmonics, airflow noise, and ear canal leakage resonance, suppressing them independently for each channel. This avoids uniform attenuation of human voice when multiple interferences are superimposed, representing a completely new noise classification and processing logic. 3. The head-related transfer function (HRTF) features seamless online real-time self-calibration, abandoning the factory static multi-level lookup table scheme. It automatically micro-calibrates the user's exclusive acoustic HRTF during wear, calls, and high-wind scenarios, dynamically and synchronously correcting ANC and human voice fusion parameters, eliminating differences in wearing conditions among different individuals. 4. The multi-mode weighted dynamic fusion interaction strategy eliminates the mutually exclusive control strategies of intelligent, outdoor, and ultra-clear modes. Weighted fusion calculation is used to unify control parameters, with the underlying core collaborative algorithm running throughout the process. Specifically, this includes: a three-dimensional noise decoupling module that continuously classifies the signals from the three microphones in real time from the moment the headset is powered on, regardless of whether a call is in progress; and a two-way closed-loop feedback engine that continuously performs ANC during a call. ENC parameters are transmitted and adjusted bidirectionally; the HRTF self-calibration module automatically starts calibration and updates parameters under trigger conditions such as wear recognition and call establishment. These three modules do not stop or reset due to mode switching. When the APP adjusts the weights of W1 / W2 / W3, only the output of the weighted fusion formula (i.e., the preference parameters of each module) is changed; the modules themselves continue to run in the background. The system only gradually reduces ANC output until it is turned off when the user manually increases W3 to a preset threshold. As a manual fallback, ANC is only turned off when the user actively triggers it, balancing intelligent and customizable fallback needs. Existing products all use mutually exclusive level switching.

[0033] In one specific embodiment, the simultaneous acquisition of multiple raw acoustic signals from inside and outside the ear canal, and the decoupling and identification of environmental noise in the multiple raw acoustic signals, specifically includes: Simultaneously acquire three raw acoustic signals from the FF (Feedforward) microphone (inside and outside the ear canal), the FB (Feedback) microphone, and the COM (Command) microphone. The feedforward microphone is located on the outside of the earphone and is used to pick up ambient noise. The FB microphone is located on the inside of the earphone and is used to pick up the mixed signal inside the ear canal. The COM microphone is located on the outside of the earphone and is used to pick up the user's real-time voice during conversations.

[0034] A lightweight decision tree classification model is constructed based on the temporal energy fluctuation, frequency domain coherence, and resonance peak characteristics of the original acoustic signal to separate three types of noise interference and their energy proportions. Based on the three-dimensional characteristics of the original acoustic signal—temporal energy fluctuation, frequency domain coherence, and resonance peak characteristics—a lightweight decision tree classification model is built to achieve three-dimensional noise decoupling and identification. Specifically, for each frame of the original acoustic signal, three key features are extracted: temporal energy fluctuation, used to distinguish between wind noise (intense and irregular energy fluctuations) and ANC harmonics (stable and periodic energy); frequency domain coherence, calculating the correlation between FF and FB microphone signals, where the reverse sound wave actively emitted by the ANC has high coherence between the two paths, while external wind noise has low coherence; and resonance peak characteristics, where ear canal leakage resonance produces stable resonance peaks at specific frequency points (usually 200~500Hz), which can be identified through spectral peak detection. The decision tree classifies noise level by level according to the priority of "first determine if it is wind noise → then determine if it is leakage resonance → finally classify it as ANC harmonics", and outputs three types of noise labels and their respective energy proportions.

[0035] Independent frequency band suppression commands are generated for each type of interference and sent to the call ENC module. These commands are isolated from each other and do not attenuate the core human voice frequency band of 200-3500Hz. Based on the frequency domain distribution characteristics of the three types of noise (ANC harmonics concentrated in 50-300Hz, wind noise in a wideband 200Hz-2kHz range, and leakage resonance in a narrowband spike 200-500Hz), the aforementioned frequency band suppression commands generate corresponding notch or attenuation filter coefficients for each type of noise. These coefficients are independently superimposed onto the noise suppression module of the ENC, but the 200-3500Hz human voice frequency band is forcibly bypassed to ensure that human voice is not attenuated. The call ENC module contains a multi-band noise suppressor. After receiving the aforementioned frequency band suppression commands, it dynamically adjusts the suppression depth parameters (such as the attenuation gain of NS) of the corresponding frequency band to the target value. Simultaneously, the AEC echo cancellation threshold is adaptively raised according to the ANC harmonic energy to avoid misinterpreting ANC reverse sound waves as far-end echoes and incorrectly canceling them.

[0036] Specifically, the three types of interference include Type A interference, Type B interference, and Type C interference. Type A interference is ANC periodic low-frequency harmonics, Type B interference is airflow turbulence noise, and Type C interference is ear canal leakage resonance noise. Among these, ANC periodic low-frequency harmonics have relatively small interference, ear canal leakage resonance noise is moderate, and airflow turbulence noise has the greatest impact. The interference level of each noise can be further distinguished by the proportion of noise energy.

[0037] In one specific embodiment, before calculating the Head-Related Transfer Function (HRTF) in real time based on the frequency sweep test signal, the method further includes: automatically triggering micro-calibration of the HRTF when any one of the following four conditions is met: earphone wearing recognition completed, call link established, sudden change in wind noise level, or ANC filter convergence deviation exceeding a threshold. By automatically triggering micro-calibration of the HRTF, imperceptible self-calibration of the HRTF can be achieved. The micro-calibration process includes four stages: 1. Trigger Judgment: One of the four trigger conditions is detected (wearing recognition completed, call link established, sudden change in wind noise level, ANC filter convergence deviation); 2. Signal transmission: The headphone speaker plays a low-amplitude (<70dB SPL) narrowband sweep signal (20Hz~4kHz, duration about 5ms); 3. Acoustic response acquisition: Simultaneous acquisition of acoustic response inside and outside the ear canal using FF / FB / COM microphones; 4. Parameter calculation: Perform deconvolution operation on the acquired response and the transmitted signal to obtain the current ear canal transfer function, and compare it with the HRTF of the previous frame. If the deviation exceeds the threshold, update it.

[0038] The aforementioned frequency sweep test signal is generated in real time by the signal generator built into the headset's DSP and multiplexed from the headset's built-in speaker output. Specifically, a set of log-sweep waveform data is pre-stored in the DSP and output to the speaker driver unit according to a preset timing sequence when calibration is triggered. The signal amplitude is controlled by a limiter to a range that does not cause auditory discomfort or affect normal calls (usually more than 15dB below normal playback volume).

[0039] In one specific embodiment, the step of calculating the Head-Related Transfer Function (HRTF) in real time based on the frequency sweep test signal, and using the HRTF to synchronously correct the filtering convergence coefficient of the ANC module and the voice fusion weight of the call ENC module, further includes: The Head Related Transfer Function (HRTF) deviation is triggered every set number of frames. If the HRTF deviation exceeds the threshold, micro-calibration is initiated. This set number of frames can be flexibly set according to actual requirements; for example, the set number of frames can be 50-200 frames.

[0040] In one specific embodiment, the bidirectional data stream between the ANC module and the call ENC module achieves bidirectional closed-loop parameter feedback. The ANC module performs hierarchical constraint filtering based on the interference intensity feedback from the call ENC module, specifically including: The ANC module outputs a forward data stream to the call ENC module in real time. The forward data stream includes five sets of dynamic characteristic quantities: FF real-time filter gain, FB loop residual, ear canal leakage coefficient, reverse acoustic wave output amplitude, and filter adaptive step size.

[0041] The call ENC algorithm adjusts the microphone voice fusion weights based on the forward data stream and outputs a reverse data stream to the ANC module in real time. The reverse data stream includes four call quality evaluation indicators: uplink voice SNR, ANC harmonic interference energy, airflow noise level, and voice distortion. These four indicators are calculated in real time by the ENC algorithm after processing each frame (typically 5-10ms). Specifically: Uplink voice SNR: Voice frames and noise frames are distinguished by speech activity detection (VAD), and the power of each frame is statistically analyzed to calculate the ratio. ANC harmonic interference energy: Perform FFT spectrum analysis on the FB microphone signal to count the total energy of harmonic components with the same frequency as the ANC drive signal in the 50~300Hz frequency band; Airflow noise level: Calculate the time-domain energy fluctuation variance of the FF microphone signal in the 200Hz~2kHz frequency band. The larger the variance, the higher the airflow noise level. Speech distortion: The degree of distortion is assessed by comparing the spectral envelope similarity (such as cepstral distance) between the voice captured by the FB microphone and the local reference voice template.

[0042] The four indicators are packaged into a control frame and sent to the ANC module via a bidirectional feedback bus.

[0043] The ANC module determines the interference intensity of environmental noise based on four call quality assessment indicators and the energy proportion of interfering noise, and performs layered constraint filtering based on the interference intensity. The interference intensity is comprehensively determined by the energy proportion of the three types of noise output by the three-dimensional noise decoupling module combined with the call quality assessment indicators returned by ENC. The determination result is output to the ANC module in the form of a level label (1 / 2 / 3).

[0044] Specifically, the interference intensity includes mild interference, moderate interference, and severe interference, and the hierarchical constraint filtering based on interference intensity specifically includes: When the interference intensity is determined to be mild, only the maximum output amplitude of the 50-600Hz low-frequency ANC is compressed, while the mid-to-high frequency noise reduction is fully preserved. Mild interference: The energy of each of the three types of noise is lower than the preset threshold, and the call SNR decreases by <3dB. Specifically, in the ANC's output limiter, a dynamic upper limit threshold is set for the 50-600Hz frequency band. Under normal operation, this threshold is relatively high (allowing full-amplitude output). When a mild interference command is received, this threshold is linearly reduced to 60%~80% of the normal value (the specific ratio is continuously adjusted according to the interference intensity). Low-frequency signals exceeding this threshold will be clipped or compressed, thereby limiting the output energy of low-frequency reverse sound waves and reducing interference to the microphone.

[0045] When the interference intensity is determined to be moderate, the fixed coefficients of the FB feedback filter are locked, and the FF feedforward adaptive iteration update frequency is reduced. Moderate interference: Wind noise energy or leakage resonance energy exceeds the threshold, and the ANC harmonic interference energy and the decrease in call SNR are between 3 and 6 dB. The coefficient update period of the FF adaptive filter is adjusted from once per frame (usually once every 5 ms) to once every N frames. The specific standards are as follows: Under normal circumstances, it updates once every 1 frame; Moderate interference, updated every 4 frames (update frequency reduced by 75%); In cases of severe interference, the filter is updated every 8 frames (update frequency reduced by 87.5%), and the FF filter enters freeze mode, keeping only the current coefficients unchanged and pausing adaptive convergence.

[0046] It is understandable that the update frequency can be flexibly adjusted according to the actual situation, and no restrictions are imposed here.

[0047] When the interference intensity is determined to be severe interference, smoothly reduce the low-frequency noise reduction gain, maintain narrowband low-frequency boom suppression, and do not switch to transparency mode or turn off the ANC module. Severe interference: Wind noise energy continuously exceeds the threshold (e.g., for more than 3 frames), and the call SNR drops by ≥6dB.

[0048] In this solution, all bidirectional data streams use 10-20ms gradual interpolation to smoothly update parameters, eliminating ear pressure impact and voice stuttering. Understandably, other gradual interpolation methods can also be used to smoothly update parameters in bidirectional data streams; this is not a limitation here.

[0049] This solution has at least the following technical advantages: 1. Significantly reduces differences in wearability across all population groups, resulting in a marked improvement in call consistency. Under different ear shapes, wearing tightness, and ear canal air leakage conditions, the SNR fluctuation of calls in a strong wind environment of 6~12m / s has been reduced from the existing 8~15dB to ≤4dB; there are no low-frequency buzzing, swallowing, or muffled voice phenomena, and the call performance of the three types of samples (closed / semi-leaking / shallow wearing) tends to be consistent. 2. ANC noise reduction capability is maintained throughout the process in strong wind and noisy environments. In extreme continuous strong winds of 12m / s, only the low-frequency ANC gain is compressed in layers to maintain the suppression of mid-to-high frequency environmental noise. The system will not automatically switch to transparency mode or forcibly turn off ANC; it solves the pain points of noise reduction loss in commuting outdoor scenarios and huge background noise for both parties in the call, and retains the core selling point of the product, ANC.

[0050] 3. The algorithm's computing power utilization is improved by 30%, reducing the chip's operating load. The bidirectional closed-loop feedback mechanism avoids the two sets of adaptive filters diverging and iterating repeatedly; the three-dimensional noise-lightweight decision tree does not require a neural network and can run stably on a low-computing-power DSP, without the computing power contention and lag that occurs during calls in windy conditions. 4. Seamless human-computer interaction, balancing general user experience with professional backup needs. User-customizable modes and underlying intelligent collaborative algorithms run in tandem, with no sudden auditory changes during switching; the ultimate clarity noise reduction off function is only an optional backup, and the factory default intelligent balance strategy is in line with the mid-to-high-end product market positioning, with no negative user feedback. 5. High hardware compatibility and extremely low mass production modification costs. It is fully compatible with existing FF+FB+COM three-microphone TWS hardware, without the need for additional microphones, sensors, or cavity structures; it can be implemented simply through firmware algorithm iteration and secondary development of the call algorithm interface, without the need for mold opening or hardware material changes, resulting in low mass production risk and a short debugging cycle.

[0051] Please refer to Figure 3 , Figure 3 This is a block diagram of a collaborative balancing control device for ANC and call algorithm according to an embodiment of the present invention. In this embodiment, the collaborative balancing control device for ANC and call algorithm is applied to headphones, particularly Bluetooth headphones. The device includes an acquisition module 110, a self-calibration module 120, a bidirectional closed-loop feedback module 130, a weight fusion module 140, and a processing module 150. Specifically, The acquisition module 110 is used to simultaneously acquire multiple original acoustic signals from inside and outside the ear canal, and to decouple and identify environmental noise in the multiple original acoustic signals respectively. The self-calibration module 120 is used to self-calibrate the head-related transfer function (HRTF). It calculates the HRTF in real time based on the frequency sweep test signal and uses the HRTF to synchronously correct the filtering convergence coefficient of the ANC module and the voice fusion weight of the call ENC module. The bidirectional closed-loop feedback module 130 is used to establish a bidirectional closed-loop parameter feedback channel between the ANC module and the call ENC module, so that the bidirectional data stream between the ANC module and the call ENC module can achieve bidirectional closed-loop parameter feedback. The ANC module determines the interference intensity of environmental noise based on the feedback result and performs hierarchical constraint filtering. The weight fusion module 140 is used to obtain the independent weights and corresponding parameters of multiple modes, and calculate the total control parameters based on the weights and parameters of each mode. Processing module 150 is used to synchronously execute the noise reduction of the ANC module and the call voice enhancement of the call ENC module according to the overall control parameters.

[0052] Please see Figure 4 , Figure 4 This is a block diagram of a headset according to an embodiment of the present invention. This headset can be used to implement the ANC and call algorithm collaborative balance control method in the aforementioned embodiments.Figure 4 As shown, the headset includes a memory 301, a processor 302, a bus 303, and a computer program stored in the memory 301 and executable on the processor 302. The memory 301 and the processor 302 are connected via the bus 303. When the processor 302 executes the computer program, it implements the ANC and call algorithm collaborative balance control method described in the foregoing embodiments. The number of processors can be one or more.

[0053] The memory 301 can be a high-speed random access memory (RAM) or a non-volatile memory, such as a disk storage device. The memory 301 is used to store executable program code, and the processor 302 is coupled to the memory 301.

[0054] Furthermore, embodiments of this application also provide a computer-readable storage medium, which may be disposed in the ANC and call algorithm collaborative balance control method in the above embodiments. This computer-readable storage medium may be the aforementioned... Figure 4 The memory in the illustrated embodiment.

[0055] The computer-readable storage medium stores a computer program that, when executed by a processor, implements the ANC and call algorithm collaborative balance control method described in the foregoing embodiments. Furthermore, the computer-readable storage medium can also be any medium capable of storing program code, such as a USB flash drive, external hard drive, read-only memory (ROM), RAM, magnetic disk, or optical disk.

[0056] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0057] The modules described as separate components may or may not be physically separate. Similarly, the components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0058] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0059] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned readable storage medium includes various media capable of storing program code, such as USB flash drives, external hard drives, ROM, RAM, magnetic disks, or optical disks.

[0060] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0061] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0062] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural transformations made using the contents of the present invention's specification and drawings under the inventive concept of the present invention, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

Claims

1. A method for coordinated balance control of ANC and call algorithm, characterized in that, The ANC and call algorithm collaborative balance control method includes: Simultaneously acquire multiple raw acoustic signals from inside and outside the ear canal, and decouple and identify environmental noise in the multiple raw acoustic signals respectively; The Head-Related Transfer Function (HRTF) is self-calibrated. The HRTF is calculated in real time based on the frequency sweep test signal. The HRTF is then used to synchronously correct the filtering convergence coefficient of the ANC module and the voice fusion weight of the call ENC module. A bidirectional closed-loop parameter feedback channel is established between the ANC module and the call ENC module to enable bidirectional closed-loop parameter feedback between the two data streams of the ANC module and the call ENC module. The ANC module determines the interference intensity of environmental noise based on the feedback results and performs hierarchical constraint filtering. Obtain the independent weights and corresponding parameters of multiple modes, and calculate the total control parameters based on the weights and parameters of each mode; The noise reduction of the ANC module and the voice enhancement of the call ENC module are executed synchronously according to the overall control parameters.

2. The ANC and call algorithm collaborative balance control method as described in claim 1, characterized in that, The process of simultaneously acquiring multiple raw acoustic signals from inside and outside the ear canal, and decoupling and identifying environmental noise in these signals, specifically includes: Simultaneously acquire three raw acoustic signals collected by the FF feedforward microphone inside and outside the ear canal, the FB feedback microphone, and the COM call microphone; A lightweight decision tree classification model is constructed based on the time-domain energy fluctuation, frequency-domain coherence, and resonance peak characteristics of the original acoustic signal to separate the three types of noise interference and the energy proportion of the interference noise in the original acoustic signal. Independent frequency band suppression commands are generated for each type of interference and sent to the call ENC module.

3. The ANC and call algorithm collaborative balance control method as described in claim 2, characterized in that, The three types of interference include Type A interference, Type B interference, and Type C interference. Type A interference is ANC periodic low-frequency harmonics, Type B interference is airflow turbulence noise, and Type C interference is ear canal leakage resonance noise.

4. The ANC and call algorithm collaborative balance control method as described in claim 2, characterized in that, Before calculating the Head-Related Transfer Function (HRTF) in real time based on the frequency sweep test signal, the method further includes automatically triggering micro-calibration of the HRTF when any one of the following four conditions is met: completion of headphone wearing recognition, establishment of call link, sudden change in wind noise level, or ANC filter convergence deviation exceeding the threshold.

5. The ANC and call algorithm collaborative balance control method as described in claim 4, characterized in that, The step of calculating the Head-Related Transfer Function (HRTF) in real time based on the frequency sweep test signal, and using the HRTF to synchronously correct the filtering convergence coefficient of the ANC module and the voice fusion weights of the call ENC module, also includes: Head-related transfer function (HRTF) deviation detection is triggered every set number of frames. If the head-related transfer function deviation exceeds the threshold, micro-calibration is initiated.

6. The ANC and call algorithm collaborative balance control method as described in claim 2, characterized in that, The bidirectional data stream between the ANC module and the call ENC module enables bidirectional closed-loop parameter feedback. The ANC module performs hierarchical constraint filtering based on the interference intensity feedback from the call ENC module, specifically including: The ANC module outputs a forward data stream to the call ENC module in real time. The forward data stream includes five sets of dynamic characteristic quantities: FF real-time filter gain, FB loop residual, ear canal leakage coefficient, reverse sound wave output amplitude, and filter adaptive step size. The call ENC algorithm adjusts the voice fusion weight of the microphone based on the forward data stream and outputs the reverse data stream to the ANC module in real time. The reverse data stream includes four call quality evaluation indicators: uplink voice SNR, ANC harmonic interference energy, airflow noise level and voice distortion. The ANC module determines the interference intensity of environmental noise based on four call quality assessment indicators and the energy proportion of interference noise, and performs layered constraint filtering based on the interference intensity.

7. The ANC and call algorithm collaborative balance control method as described in claim 6, characterized in that, The interference intensity includes mild interference, moderate interference, and severe interference. The hierarchical constraint filtering based on interference intensity specifically includes: When the interference intensity is determined to be mild, only the maximum output amplitude of the low-frequency ANC in the 50-600Hz range is compressed, while the mid-to-high frequency noise reduction is fully preserved. When the interference intensity is determined to be moderate interference, the fixed coefficients of the FB feedback filter are locked, and the frequency of the FF feedforward adaptive iterative update is reduced. When the interference intensity is determined to be severe interference, the low-frequency noise reduction gain is smoothly reduced to maintain narrowband low-frequency boom suppression, without switching to transparency mode or turning off the ANC module.

8. A collaborative balancing control device for ANC and call algorithm, characterized in that, The ANC and call algorithm coordinated balance control device includes: The acquisition module is used to simultaneously acquire multiple raw acoustic signals from inside and outside the ear canal, and to decouple and identify environmental noise in the multiple raw acoustic signals respectively. The self-calibration module is used to self-calibrate the Head-Related Transfer Function (HRTF). It calculates the HRTF in real time based on the frequency sweep test signal and uses the HRTF to synchronously correct the filtering convergence coefficient of the ANC module and the voice fusion weight of the call ENC module. The bidirectional closed-loop feedback module is used to establish a bidirectional closed-loop parameter feedback channel between the ANC module and the call ENC module, so that the bidirectional data streams of the ANC module and the call ENC module can achieve bidirectional closed-loop parameter feedback. The ANC module determines the interference intensity of environmental noise based on the feedback results and performs hierarchical constraint filtering. The weight fusion module is used to obtain the independent weights and corresponding parameters of multiple modes, and calculate the total control parameters based on the weights and parameters of each mode. The processing module is used to synchronously execute the noise reduction of the ANC module and the call voice enhancement of the ENC module according to the overall control parameters.

9. An earphone, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements the steps of the method according to any one of claims 1-7.

10. A readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-7.

Images