Wireless earphone self-adaptive anti-noise method and related equipment

Abstract

The invention provides a wireless earphone adaptive anti-noise method and related equipment, and relates to the technical field of earphone intelligent adjustment. The method comprises the following steps: acquiring acceleration information when a user uses the earphone by virtue of an accelerometer in the earphone, acquiring head rotation information by utilizing a gyroscope, and simultaneously collecting a plurality of audio signals corresponding to the earphone. And determining a suspected use scene by using an earphone use scene model which is trained by deep learning and is generated by the audio information set with the use scene label in combination with the audio signal. Through a scene verification model trained by the acceleration with the scene label and the head rotation information set, a real scene is determined by combining the acceleration with the scene label and the head rotation information set, and finally anti-noise parameters are adjusted according to the real scene. In this way, the wireless earphone can identify the characteristics of the current use environment more accurately, so that more accurate noise processing is realized, and the audition experience of a user in various environments is effectively improved.

Description

Technical Field

[0001] The present application relates to the technical field of intelligent earphone adjustment, and in particular to a method for adaptively reducing noise in wireless earphones and related equipment. Background Art

[0002] Wireless headphones have become a widely used audio device in our daily lives. Users often use wireless headphones in different scenarios, such as indoors, outdoors, and while in transit. The acoustic environments in these scenarios vary greatly, posing diverse challenges to the noise handling capabilities of wireless headphones.

[0003] Currently, wireless headphones typically use fixed noise processing solutions to suppress ambient noise using preset noise reduction parameters. In practice, this solution often identifies and processes noise based on the audio signal collected by the microphone, and then applies the corresponding noise reduction operation according to the pre-set noise processing strategy.

[0004] However, in complex and ever-changing usage environments, noise processing solely based on audio signals presents certain limitations. Due to the varying noise characteristics and user needs in different scenarios, a fixed noise processing strategy may produce suboptimal noise reduction and even affect the user's ability to receive effective sound. This is especially true when user activity levels change. A single noise processing solution struggles to accurately identify the actual usage scenario and, consequently, to meet dynamically changing needs. Summary of the Invention

[0005] The present application provides a wireless headset adaptive anti-noise method and related equipment, which are used to solve the technical problem that existing wireless headsets cannot accurately identify usage scenarios in complex and changing environments, resulting in unsatisfactory noise reduction effects.

[0006] In the first aspect, the present application provides an adaptive noise reduction method for wireless headphones, which is applied to wireless headphones, and the method includes: obtaining acceleration information of the current user when using the headphones through an accelerometer device in the headphones; obtaining user head rotation information through a gyroscope in the headphones; obtaining multiple audio signals corresponding to the headphones; combining multiple audio signals, determining the suspected usage scenario information of the current wireless headphones through a headphone usage scenario model, and the headphone usage scenario model is pre-trained through deep learning using multiple audio information sets annotated with usage scenario information; combining the acceleration information and the user's head rotation information, determining the actual usage scenario information of the current wireless headphones through a scenario verification model, and the scenario verification model is pre-trained through deep learning using multiple acceleration information annotated with usage scenario information and user head rotation information sets; and adjusting the noise reduction parameters of the wireless headphones based on the actual usage scenario information.

[0007] By adopting this technical solution, the headphones can simultaneously obtain physical characteristic information such as the user's movement status and head posture. This information is integrated with the audio signal for multi-dimensional analysis, forming a dual information verification mechanism. This mechanism first makes an initial judgment based on audio characteristics, and then verifies it through physical characteristic information, greatly reducing the error rate of environmental judgment. Ultimately, wireless headphones can more accurately identify the characteristics of the current usage environment, thereby achieving more precise noise processing and effectively improving the user's listening experience in various environments.

[0008] In combination with some embodiments of the first aspect, in some embodiments, the step of adjusting the anti-noise parameters of the wireless headset in combination with the actual usage scenario information specifically includes: obtaining the initial anti-noise parameters corresponding to the actual usage scenario information from a preset scene-parameter mapping table; personalizing the initial anti-noise parameters based on the user's historical usage data to obtain personalized anti-noise parameters; and performing noise reduction processing on the wireless headset according to the personalized anti-noise parameters.

[0009] By adopting this technical solution, wireless headphones have established a dynamic parameter adjustment mechanism. This mechanism first obtains basic parameters as a reference standard for noise reduction processing, and then optimizes and adjusts them based on the user's actual usage records. This gradual parameter optimization process allows wireless headphones to gradually adapt to the user's personalized needs while maintaining basic noise reduction effects.

[0010] In combination with some embodiments of the first aspect, in some embodiments, after the step of obtaining the initial anti-noise parameters corresponding to the actual usage scenario information from the preset scene-parameter mapping table, it also includes: obtaining and recording the user's usage data in the current actual usage scenario, the usage data at least including the scene type identifier, the start and end time of usage, the environmental noise data and the user operation record; calculating the cumulative usage time corresponding to the scene type identifier based on the start and end time of usage, and judging whether the cumulative usage time reaches the preset time threshold; when the cumulative usage time reaches the preset time threshold, determining the noise characteristic distribution of the scene based on the environmental noise data, the noise characteristic distribution including Noise frequency distribution, noise intensity distribution and noise duration distribution; based on the noise feature distribution and the user operation record, an adaptive noise reduction parameter optimization model for the scenario is trained through a machine learning algorithm, and the adaptive noise reduction parameter optimization model is used to output the optimal noise reduction parameter configuration; the scene type identifier and the corresponding optimal noise reduction parameter configuration are stored in a user-personalized scene-parameter mapping table, and the scene-parameter mapping table dynamically records the usage frequency of each scene; the usage frequency of each scene type identifier in the scene-parameter mapping table is periodically detected, and when the usage frequency of the target scene is lower than the preset frequency threshold, the personalized configuration data of the target scene is cleared from the scene-parameter mapping table.

[0011] By employing this technical solution and continuously recording usage data, wireless headphones can analyze the distribution patterns of noise characteristics in different environments. Based on this data, an optimization model can automatically adjust the noise reduction strategy and dynamically maintain the efficient operation of the wireless headphones by dynamically maintaining the configuration table. This continuous learning and optimization mechanism enables wireless headphones to continuously improve their noise reduction effects, providing users with increasingly precise noise reduction services.

[0012] In combination with some embodiments of the first aspect, in some embodiments, the step of obtaining multiple audio signals corresponding to the headset specifically includes: simultaneously enabling the microphone arrays inside and outside the headset; performing spatial filtering on the collected audio signals; calculating the correlation index of the external microphone signal corresponding to the microphone array; identifying noise source information based on the correlation index, the noise source information including the direction and distance of the noise source; and classifying and preprocessing the audio signal according to the noise source information.

[0013] By adopting the above technical solution and establishing a stereo sound acquisition and processing mechanism, the microphone array uses spatial filtering to improve signal quality and accurately locate noise sources. This spatialized noise processing method can more effectively distinguish sound signals from different sources, providing more accurate noise signature analysis results, thereby achieving more targeted noise reduction and significantly improving the noise reduction effect of wireless headphones.

[0014] In combination with some embodiments of the first aspect, in some embodiments, after the step of determining the suspected usage scenario information of the current wireless headset through the headset usage scenario model in combination with multiple audio signals, it also includes: determining the scene matching degree of the suspected usage scenario information matching the multiple audio signals; if the scene matching degree is lower than the set threshold, then selecting the three usage scenario information with the highest scene matching degree as the suspected usage scenario information.

[0015] By adopting the above technical solution, multiple candidate scenarios are scored and the optimal scenario is selected for user confirmation. This human-machine combined judgment mechanism can not only ensure the automatic operation of wireless headphones, but also obtain user confirmation at critical moments, thereby maximizing the accuracy of wireless headphones' judgment.

[0016] In combination with some embodiments of the first aspect, in some embodiments, if the scene matching degree is lower than the set threshold, then after the step of selecting the three usage scenario information with the highest scene matching degree as suspected usage scenario information, it also includes: combining the acceleration information and the user's head rotation information, verifying the three usage scenario information in turn through the scene verification model to determine the scene scores corresponding to the three usage scenario information; sending the target usage scenario information with the highest scene score to the user for confirmation via voice to determine the actual usage scenario information of the current wireless headset.

[0017] By adopting the above technical solution, by integrating the two links of scenario verification and user confirmation, multiple candidate scenarios are scored and the optimal scenario is selected for user confirmation, which not only ensures the accuracy of the judgment but also provides the possibility of user intervention.

[0018] In combination with some embodiments of the first aspect, in some embodiments, after the step of adjusting the anti-noise parameters of the wireless headset in combination with the actual usage scenario information, it also includes: collecting the wearer's multidimensional biometric information through a biosensor, and the biometric information includes at least breathing rate and body temperature; based on the multidimensional biometric information, the wearer's real-time fatigue level is calculated through a fatigue assessment model, and the fatigue assessment model is pre-trained by a deep learning method through fatigue level annotations with different biometrics, and the real-time fatigue index is used to characterize the current fatigue level of the headset wearer; determining a fatigue adjustment coefficient based on the real-time fatigue level, wherein the fatigue adjustment coefficient is positively correlated with the real-time fatigue level; after obtaining the current basic anti-noise parameters of the wireless headset, the fatigue adjustment coefficient is adaptively fused with the basic anti-noise parameters to obtain anti-noise optimized parameters; and adjusting the anti-noise parameters based on the anti-noise optimized parameters.

[0019] By adopting this technical solution and monitoring the user's physiological state in real time, wireless headphones can promptly detect changes in fatigue. Combined with the fatigue adjustment factor, the wireless headphones can automatically adjust noise reduction parameters to better suit the user's current physiological state, effectively improving the user experience and comfort of wireless headphones.

[0020] In a second aspect, the present application provides a wireless headset, comprising: one or more processors and a memory; the memory is coupled to the one or more processors, the memory being used to store computer program code, the computer program code comprising computer instructions, the one or more processors calling the computer instructions to cause the wireless headset to execute the method described in the first aspect and any possible implementation of the first aspect.

[0021] In a third aspect, the present application provides a computer-readable storage medium comprising instructions, which, when executed on a wireless headset, causes the wireless headset to execute the method described in the first aspect and any possible implementation of the first aspect.

[0022] In a fourth aspect, the present application provides a computer program product, which, when running on a wireless headset, enables the wireless headset to execute the method described in the first aspect and any possible implementation manner of the first aspect.

[0023] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. By adopting the above technical solution, due to the use of a dual verification mechanism that coordinates audio signals and physical feature information, the technical problem of unsatisfactory environmental noise processing in the existing technology is effectively solved, thereby achieving precise noise reduction in complex and changing environments, allowing users to receive important sound information more clearly.

[0024] 2. By adopting the above technical solution, an adaptive noise reduction parameter optimization model based on usage data is used. Therefore, the technical problem that the fixed audio processing mode in the existing technology is difficult to meet the dynamic usage needs is effectively solved, thereby realizing the intelligent optimization and dynamic adjustment of the noise reduction parameters, so that the wireless headphones can continuously provide the best noise reduction effect.

[0025] 3. By adopting the above technical solution, due to the use of an intelligent adjustment mechanism based on biometrics, the technical problem in the existing technology that audio processing cannot adapt to changes in user status is effectively solved, and automatic noise reduction adjustment based on the user's fatigue level is realized, which significantly improves the adaptability and comfort of wireless headphones during long-term use. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a flow chart of a method for adaptive noise reduction for wireless headphones according to an embodiment of the present application; Figure 2 This is another flowchart of the adaptive noise reduction method for wireless headphones in an embodiment of the present application; Figure 3 This is a schematic diagram of the physical structure of a wireless headset in an embodiment of the present application. DETAILED DESCRIPTION

[0027] The terms used in the following examples of the present application are for the purpose of describing specific embodiments only and are not intended to limit the present application. As used in the specification and appended claims of the present application, the singular expressions "a," "an," "said," "above," "the," and "this" are intended to include plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in the present application refers to and encompasses any or all possible combinations of one or more of the listed items.

[0028] In the following, the terms "first" and "second" are used for descriptive purposes only and should not be understood to imply or suggest relative importance or implicitly indicate the number of the technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of this application, unless otherwise specified, "plurality" means two or more.

[0029] For ease of understanding, the following describes the process of the method provided by this implementation. Figure 1 , which is a flow chart of the adaptive anti-noise method for wireless headphones in an embodiment of the present application.

[0030] S101, obtaining user head rotation information through a gyroscope in the headset; In this step, the wireless headphones use their built-in accelerometer to monitor the user's motion in real time. Specifically, the accelerometer senses changes in the user's acceleration in three dimensions (X, Y, and Z). For example, when the user is stationary, the accelerometer primarily detects gravity. When the user walks, the accelerometer detects regular up-and-down and left-and-right vibrations. When the user runs, it detects larger, periodic acceleration changes.

[0031] The wireless earphones preprocess the raw acceleration data they collect, including smoothing, denoising, and normalizing it. By analyzing this processed acceleration data, a preliminary judgment can be made about the user's current activity status. For example, by analyzing the amplitude and frequency characteristics of the acceleration signal, it is possible to distinguish whether the user is stationary, walking, running, or riding in a vehicle. Furthermore, the acceleration information is collected continuously, and the wireless earphones maintain real-time updates of the acceleration data to ensure timely capture of changes in the user's activity status. Furthermore, the wireless earphones also incorporate a time window mechanism to conduct a comprehensive analysis of acceleration data within a certain time range to improve the stability and reliability of activity status judgment.

[0032] S102, obtaining multiple audio signals corresponding to the earphone; Wireless headphones use built-in gyroscopes to capture the user's head motion. Gyroscopes accurately measure changes in the head's angular velocity in three rotational degrees of freedom (pitch, yaw, and roll). This data reflects the user's head rotational characteristics and provides important biomechanical characteristics for scene recognition.

[0033] In practice, the gyroscope first acquires raw angular velocity data, which is then processed through digital filtering and other signal processing operations to remove noise and unstable factors. The processed data clearly reflects the characteristics of the user's head movement. For example, when walking, the head will have a slight periodic sway. The wireless headphones then perform pattern analysis on these head movement characteristics to extract discriminative feature parameters. For example, by analyzing the frequency, amplitude, and duration of head rotation, the user's current behavioral state can be inferred.

[0034] In some embodiments, wireless headphones are equipped with both internal and external microphone arrays, which operate simultaneously during operation. The internal microphone, located near the user's ear, primarily collects sounds near the user's ear, including the user's own voice and sound reflections within the ear canal, helping to better capture the intended speech signal. The external microphones, located on the headphone housing, collect sounds from the surrounding environment, capturing a wider range of ambient audio information, such as ambient noise and other voices. The collected audio signals often contain various noise and interference, and spatial filtering can improve signal quality. The headphones utilize a spatial filtering algorithm to weight the signals received by different microphones based on the geometry of the microphone array and the sound propagation characteristics. This processing enhances sound signals from specific directions while suppressing noise from other directions. For example, when someone is speaking directly in front of the user, spatial filtering can enhance the sound directly in front of them and attenuate interfering sounds from other directions, making the intended sound clearer. It also effectively reduces the effects of reverberation and echo. By analyzing the time and amplitude differences between the signals from different microphones, it removes unwanted sounds caused by reflections, improving the purity of the sound.

[0035] The correlation index of external microphone signals measures the similarity between the sound signals received by different external microphones. The headphones obtain this index by calculating the time and frequency correlation of these signals. Specifically, the waveforms of the signals received by different microphones at the same time are compared and the similarity coefficient is calculated. A high correlation indicates that the sound source is likely close to the microphones and in similar directions. Conversely, a low correlation indicates that the sound sources are likely in different directions or at significantly different distances. Based on the calculated correlation index, the headphones can identify the direction and distance of the noise source. The direction of the noise source is determined by utilizing the microphone array layout and the time difference (TDOA) or phase difference (PDOA) of the signals arriving at different microphones. If the sound from a noise source arrives at a certain microphone first, the approximate direction of the noise source relative to the headphones can be determined by analyzing the temporal order and strength differences of the signals received by different microphones, combined with the correlation index. When determining the distance of the noise source, factors such as signal strength, correlation, and the attenuation characteristics of sound propagation are comprehensively considered. Generally speaking, the farther the distance, the weaker the signal strength and the lower the correlation. After obtaining the direction and distance information of the noise source, the headphones perform classification preprocessing on the audio signal. Different processing strategies are used for noise sources coming from different directions and distances. If the noise source is from a direction the user doesn't need to pay attention to and is relatively far away, the headphones can appropriately reduce the gain of the audio signal in that direction or use a more stringent noise reduction algorithm to process it. For noise sources that are closer and may interfere with the user's listening to important sounds, the headphones prioritize noise reduction.

[0036] S103: Determine suspected usage scenario information of the current wireless headset using a headset usage scenario model based on the multiple audio signals, where the headset usage scenario model is pre-trained through deep learning using multiple audio information sets annotated with usage scenario information; After obtaining multiple audio signals, the wireless headset begins to use the headset usage scenario model to determine the current suspected usage scenario information.

[0037] In practice, the microphones of wireless headphones continuously collect sound signals from the surrounding environment. These audio signals contain a wealth of information. For example, in an office setting, the audio signals may include keyboard tapping, people chatting, and the sound of printers. On a bus, there will be the roar of the vehicle, the sound of passengers chatting, and the sound of station announcements. The training process of the headphone usage scenario model is as follows: researchers collect a large amount of audio information from different scenarios and label it with the corresponding scene type, such as indoor meetings, outdoor streets, and shopping malls. These labeled audio information sets are used as training data and input into the deep learning model. By learning from this data, the deep learning model continuously adjusts its parameters, thereby extracting scene-specific features from the audio signal.

[0038] When wireless headphones are used, their built-in processor inputs multiple acquired audio signals into a pre-trained model of headphone usage scenarios. The model then performs a series of complex analyses and processing on the audio signals. It identifies features such as frequency components, changes in sound intensity, and duration of the sound. For example, if the audio signal contains frequent keyboard tapping sounds with relatively stable intensity, accompanied by some low conversations, the model may initially determine that the current scene is an office. However, if the audio signal contains a large amount of noisy human voices and vehicle sounds, and the sound intensity and frequency changes are relatively random, the model may interpret it as an outdoor street scene.

[0039] When processing audio signals, the model utilizes deep learning algorithms such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs). For example, a CNN automatically extracts local features from the audio signal through a convolutional layer. It then compresses and filters these features through a pooling layer. Finally, a fully connected layer integrates and classifies these features, outputting probability values ​​for each scenario. Based on these probabilities, the wireless headphones select the scenario with the highest probability as the suspected usage scenario. For example, if the model output shows a probability of 0.7 for an office scene, 0.2 for an outdoor street scene, and 0.1 for all other scenarios, the wireless headphones will identify the office scene as the suspected usage scenario.

[0040] However, relying solely on audio signals to determine the scene can lead to errors. For example, the audio signal of an indoor lecture venue may be similar to the audio signal of a noisy outdoor environment, containing a large amount of human voices and some ambient background sounds. In this case, it is necessary to combine other information to further verify the accuracy of the scene, which is the purpose of the next step S104.

[0041] In some embodiments, the wireless headset can first calculate the degree of match between the audio signal and various preset scenarios. This includes extracting characteristic parameters of the current audio signal, such as spectral distribution, sound pressure level variations, and signal periodicity. These features are then compared with standard feature templates for preset scenarios. A similarity algorithm, such as cosine similarity or Euclidean distance, is used to calculate the degree of match, generating a match score between 0 and 1 for each possible scenario. If the highest match score falls below a set threshold (e.g., 0.6), indicating that the current environment does not match the standard scenario template well enough, the wireless headset will sort all scenarios in descending order of match score and select the top three scenarios with the highest match scores as candidates. These scenarios are then marked as suspected scenarios requiring further verification. The wireless headset then uses a scenario verification model to further verify the three candidate scenarios using acceleration and head rotation information, obtaining scenario scores for each candidate scenario. The wireless headset will then notify the user of the highest-scoring scenario via a voice prompt. For example, the voice prompt may be concise and clear, such as, "This appears to be an office environment. Is this correct?" If the user confirms the correctness, the noise reduction parameters for that scenario are immediately applied. If the user denies the correctness, the next highest-scoring scenario is offered as an option. This multi-level scene recognition and confirmation mechanism can handle complex or ambiguous usage scenarios, allowing wireless headphones to more accurately identify the usage environment and provide users with more precise noise reduction services.

[0042] S104: Determine, by combining the acceleration information and the user's head rotation information, actual usage scenario information of the current wireless headset using a scenario verification model, where the scenario verification model is previously trained through deep learning using multiple sets of acceleration information and user head rotation information annotated with usage scenario information; After determining the suspected usage scenario information, the wireless headset needs to further combine acceleration information and user head rotation information to determine the actual usage scenario information through a scenario verification model. First, acceleration information and user head rotation information provide important supplementary evidence for scenario verification. The accelerometer device continuously collects changes in acceleration when the user uses the headset. For example, when the user is riding a bus, the accelerometer will detect the acceleration changes caused by the vehicle starting, braking, and turning; when the user walks in the office, the acceleration changes are relatively stable and the amplitude is small. The gyroscope obtains the user's head rotation information in real time. In different scenarios, the frequency, amplitude, and method of the user's head rotation are also different.

[0043] The scenario verification model was also trained using deep learning. Researchers collected a large amount of acceleration and head rotation information, annotated with usage scenario information, as training data. Using this data, the model learned the characteristic patterns of acceleration and head rotation in different scenarios. For example, in a bus scenario, acceleration changes with a certain regularity, and head rotation may be more frequent due to the swaying of the vehicle and observation of the situation outside the window. In contrast, in a library scenario, acceleration changes are smaller, and head rotation is relatively slow and small.

[0044] When the wireless headset obtains acceleration information and user head rotation information, it will input this information into the scene verification model. The model will analyze and process this information to extract key features. For example, by analyzing the amplitude, frequency, angle and speed of the acceleration, and other features of the head rotation, it will be compared with the various scene modes that have been learned. Suppose the headset usage scene model preliminarily determines that the suspected usage scene is "outdoor street", but when the scene verification model analyzes the acceleration information, it finds that the acceleration changes are relatively smooth, and there are no obvious acceleration changes such as vehicle starting and braking. At the same time, the head rotation is relatively small, which does not conform to the usual movement characteristics of people in outdoor street scenes. On the contrary, these characteristics are more consistent with the characteristics of a relatively quiet indoor environment, so the scene verification model will correct the results of the headset usage scene model.

[0045] During the scene verification process, the scene verification model may utilize deep learning technologies such as the Long Short-Term Memory (LSTM) network. LSTM effectively processes time series data, capturing the temporal variations in acceleration and head rotation information. It remembers past information and dynamically adjusts based on current input, enabling more accurate scene judgment. For example, when a user moves from outdoors to indoors, acceleration and head rotation information change. LSTM can track these changes and, combined with previous information, accurately identify the scene transition.

[0046] Through the verification of the scene verification model, wireless headphones can more accurately determine the current real-world usage scenario information. This not only improves the accuracy of scene recognition, but also provides a reliable basis for subsequent adjustment of anti-noise parameters based on real-world scenarios.

[0047] S105: Adjust the anti-noise parameters of the wireless headset based on the actual usage scenario information.

[0048] The headphones will first obtain the initial anti-noise parameters corresponding to the real scene from the preset scene-parameter mapping table. These parameters are pre-set based on a large amount of scene data. For example, in an office scene, the initial parameters may focus on filtering keyboard tapping sounds and slight ambient noise; the bus scene is set for vehicle driving noise and crowd noise. Then, the headphones will personalize the initial parameters based on the user's historical usage data, such as volume adjustment habits and noise reduction mode preferences in a certain scene, to obtain personalized anti-noise parameters that better meet user needs. Finally, according to the personalized anti-noise parameters, the headphones start noise reduction processing, and optimize the sound effect by adjusting the noise reduction intensity, frequency range, etc., to enhance the user's listening experience in different scenarios.

[0049] In an embodiment of the present application, acceleration information is obtained by an accelerometer, and head rotation information is obtained by a gyroscope. Combined with multiple audio signals, the actual usage scenario is determined using a headphone usage scenario model and a scenario verification model. Then, the anti-noise parameters are adjusted in combination with the actual scenario information, so that the wireless headphones can accurately identify and match the usage scenarios in complex and changeable environments, and make targeted anti-noise parameter adjustments accordingly. This not only effectively solves the problem that existing wireless headphones have difficulty in accurately identifying scenes and have poor noise reduction effects in complex environments, but also greatly improves the user's listening experience in various environments, enabling people with hearing impairments to perceive external sound information more clearly.

[0050] In some embodiments, wireless headphones monitor the wearer's multi-dimensional biometric information in real time through integrated biosensors. These sensors include a respiratory sensor for measuring respiratory rate and a temperature sensor for detecting body temperature. The respiratory sensor can obtain respiratory rate data by detecting changes in the wearer's respiratory airflow or chest movement, while the temperature sensor measures changes in body temperature through direct contact with the skin. These biosensors use high-precision detection elements to ensure the accuracy of data collection. The wireless headphones pre-process the collected raw biometric data, and the pre-processed multi-dimensional biometric information is input into the fatigue assessment model. This model is trained using deep learning methods, and the training data contains a large number of biometric samples with fatigue level annotations. During the training process, researchers collected biometric data from different users in various fatigue states, and professionals labeled the data with fatigue levels, such as mild fatigue, moderate fatigue, and severe fatigue.

[0051] The fatigue assessment model utilizes a deep neural network architecture comprised of multiple hidden layers. These layers automatically learn complex correlations between biometric features and extract high-level features related to fatigue. For example, the model may learn that a decrease in breathing rate and a slight increase in body temperature typically indicate fatigue. The model output is a standardized real-time fatigue index, typically ranging from 0 to 1, with higher values ​​indicating greater fatigue. Based on the calculated real-time fatigue index, the wireless earbuds determine a corresponding fatigue adjustment factor. This factor is positively correlated with fatigue severity; higher fatigue levels increase the adjustment factor. For example, when the fatigue index is below 0.3, the adjustment factor may be close to 1, indicating that little adjustment is required. When the fatigue index exceeds 0.7, the adjustment factor may reach 1.5 or higher, indicating that significant parameter adjustments are required.

[0052] After obtaining the current baseline noise cancellation parameters of the wireless headphones, they adaptively combine the fatigue adjustment coefficient with these parameters. This fusion process is not a simple multiplication, but rather a weighted fusion approach that takes into account the sensitivity of different parameters to fatigue. Ultimately, the wireless headphones adjust the noise cancellation process based on the optimized noise cancellation parameters obtained through the fusion operation. This adjustment is gradual, preventing sudden changes in parameters from causing discomfort to the user. The wireless headphones continuously monitor changes in biometrics, updating fatigue assessments and adjusting parameters in real time to ensure that the noise cancellation effect always meets the user's physiological needs. This biometric-based intelligent adjustment mechanism enables the wireless headphones to better adapt to the user's physical condition, providing more personalized noise cancellation and preventing excessive fatigue. For example, if the wireless headphones detect that the user's fatigue level is increasing, they may adjust the noise cancellation parameters to reduce auditory strain or even provide voice reminders to encourage the user to take a break, reflecting the product's humanized design philosophy.

[0053] After combining the above content, the following is a more detailed description of the process of the method provided by this implementation. Figure 2 , is another flow chart of the adaptive anti-noise method for wireless headphones in an embodiment of the present application.

[0054] S201, obtaining initial anti-noise parameters corresponding to the actual usage scenario information from a preset scenario-parameter mapping table; After determining the actual usage scenario, the wireless earbuds quickly retrieve the corresponding initial noise reduction parameters from a pre-set scenario-parameter mapping table. This mapping table stores the correspondence between a large number of different scenarios and initial noise reduction parameters. It is based on extensive experimental data, real-world scenario testing, and in-depth analysis of various noise characteristics. When constructing the mapping table, R&D personnel conducted detailed research on many common scenarios, such as offices, buses, subways, libraries, and outdoor streets, collecting and analyzing key characteristics of noise in these scenarios, such as frequency range, intensity, and duration. For each scenario, they set a set of initial parameters that achieve good noise reduction results based on the principles and goals of noise reduction technology. These parameters cover several key aspects of the noise reduction algorithm, such as the noise reduction intensity level, the noise reduction emphasis in different frequency bands, and the suppression strategy for different types of noise.

[0055] Once the wireless earbuds identify a real-world usage scenario, their internal processor quickly locates the corresponding initial noise reduction parameters in a mapping table based on the scenario information. This retrieval process relies on efficient data storage and search algorithms to ensure accurate acquisition of the required parameters in a short period of time.

[0056] In some embodiments, after this step, personalized noise reduction parameters can be adjusted based on the user's historical usage data to obtain personalized noise reduction parameters, and noise reduction processing can be performed accordingly. Specifically, the wireless headset's storage system aggregates user usage data from different scenarios. This data includes usage scenarios, usage duration, volume adjustment operations, noise reduction mode switching records, and the user's location information in specific scenarios (which can be obtained through the built-in positioning module. If the scene is indoors, Bluetooth beacon or Wi-Fi positioning technology can also be combined to obtain more accurate indoor location information). Data analysis algorithms are used to conduct in-depth analysis of the collected historical usage data. Taking volume adjustment data as an example, the user's average volume in each scenario, the frequency of volume adjustment, and volume preferences in different time periods are calculated. Based on the analysis of user usage habits and preferences, the initial noise reduction parameters are adjusted specifically. If the user is found to frequently increase the volume in a specific scenario, it may mean that the current noise reduction effect is too strong, resulting in excessive attenuation of useful sound. In this case, the noise reduction strength in that scenario can be appropriately reduced. If the user frequently switches the noise reduction mode, it means that the initially set noise reduction mode may not fully meet their needs. The headphones can optimize the noise reduction ratio of different frequency bands according to the frequency and timing of the user's switching.

[0057] S202: Acquire and record the user's usage data in the current real usage scenario, where the usage data includes at least a scenario type identifier, usage start and end time, ambient noise data, and user operation records; After obtaining the initial noise reduction parameters, the wireless earbuds immediately begin collecting and recording the user's usage data in the current real-world scenario. First, they clearly record the current scenario type identifier, which facilitates the earbuds' classification, management, and analysis of data from different scenarios. Whether it's a quiet indoor environment, a noisy outdoor street, or inside a vehicle, each scenario has its own specific identifier.

[0058] The earphones also accurately record the start and end times of use, which can be used to calculate the cumulative usage time in that scenario, and then analyze the user's usage habits and duration distribution in different scenarios. For example, statistics can be used to calculate the total weekly usage time of users in the office scene, or the average daily usage time in the public transportation scene. These data can reflect the frequency of use and degree of dependence of users in different scenarios.

[0059] The headset's built-in microphone continuously monitors ambient noise levels, collecting detailed information such as frequency distribution and intensity variations. By analyzing this ambient noise data, we gain a deeper understanding of the characteristics and variations of noise in different scenarios, providing strong support for optimizing subsequent noise reduction strategies.

[0060] In addition, the headphones also record the user's operation history. This includes various operations performed by the user on the headphones, such as adjusting the volume, switching noise reduction modes, and using function buttons. These operation records reflect the user's subjective needs and preferences for sound effects in the current scenario. For example, if the user frequently adjusts the volume, it may mean that the current noise reduction effect or sound gain does not meet their needs; if the user switches the noise reduction mode, it means that they are trying to find a noise reduction method that is more suitable for the current scenario. By analyzing these operation records, the headphones can better understand the user's behavior patterns and provide a basis for personalized adjustment of anti-noise parameters.

[0061] By acquiring and recording these comprehensive usage data, wireless headphones can continuously accumulate user usage information in different scenarios, providing rich data resources for subsequent data analysis and noise reduction parameter optimization, thereby gradually improving the noise reduction effect and user experience.

[0062] S203: Calculate the cumulative usage time corresponding to the scenario type identifier based on the usage start and end times, and determine whether the cumulative usage time reaches a preset time threshold; After recording the start and end times of use, the wireless headset uses a built-in calculation function to calculate the cumulative usage time corresponding to the current scenario type identifier based on these time data. It uses a precise time calculation algorithm to accumulate the usage time of each scenario. For example, if the user's first usage time in the bus scenario is 15 minutes and the second usage time is 20 minutes, the cumulative usage time will be continuously updated on the original basis to accurately reflect the user's total usage time in that scenario.

[0063] After calculating the cumulative usage time, the headset will then compare this data with the preset time threshold. The preset time threshold is a time standard pre-set based on the usage habits of a large number of users, scene characteristics, and data statistical analysis. Different scene types may correspond to different preset time thresholds. For example, for bus scenes, since users may use it frequently during their daily commute, the preset time threshold may be set relatively long; while for some special scenes, such as conference room scenes, users usually use it for a shorter time, so the preset time threshold will be set accordingly. When the cumulative usage time does not reach the preset time threshold, it means that the headset has not accumulated enough data on the noise characteristics and user needs in this scene. At this time, the headset may continue to work according to the current noise reduction parameters and continue to collect data. Once the cumulative usage time reaches the preset time threshold, it means that the headset has accumulated sufficient data. At this time, the headset can further analyze the noise characteristic distribution of the scene based on this data, preparing for the subsequent training of the adaptive noise reduction parameter optimization model.

[0064] S204: When the accumulated usage time reaches the preset time threshold, determine a noise characteristic distribution of the scene based on the ambient noise data, where the noise characteristic distribution includes a noise frequency distribution, a noise intensity distribution, and a noise duration distribution; Once the wireless earbuds determine that the cumulative usage time in a given scenario has reached a preset threshold, they begin to determine the noise distribution characteristics of that scenario based on previously collected ambient noise data. The earbuds first perform an in-depth analysis of the collected ambient noise data. Using specialized signal processing algorithms, they decompose complex noise signals into frequency components to determine the noise frequency distribution. During this process, the earbuds' processor uses mathematical methods such as Fourier transforms to convert the time-domain noise signal into the frequency domain, clearly displaying the energy distribution of different frequency bands. For example, the earbuds can determine the noise energy contribution of low-frequency bands (20-200Hz), mid-frequency bands (200-2000Hz), and high-frequency bands (above 2000Hz) in that scenario. This precise analysis of noise at different frequencies allows the earbuds to understand the frequency characteristics of the dominant noise in that scenario, such as whether there is a low-frequency, continuous rumble or a high-frequency, sharp, noisy sound. This provides critical information for subsequent, targeted adjustments to the noise reduction strategy.

[0065] To determine the distribution of noise intensity, the headphones accurately measure and compile statistics on the amplitude of the noise signal. They analyze changes in noise intensity over different time periods, such as the maximum, minimum, and average noise intensity in that scenario. Using this data, the headphones can understand the range and patterns of noise intensity fluctuations. For example, they can determine whether the noise intensity in a certain scenario suddenly increases within a specific time period, or remains relatively stable for most of the time but occasionally experiences intensity peaks. This helps the headphones determine the severity of the noise and its potential impact on the user's hearing, thereby determining the strength of the noise reduction and the key frequency bands.

[0066] For noise duration distribution, the headphones record the duration of noise of varying intensities and frequencies in a given scenario. By analyzing this temporal data, they can determine which noise bursts are brief and which are persistent. Combining this information on noise frequency distribution, noise intensity distribution, and noise duration distribution, the wireless headphones construct a comprehensive and detailed noise signature map for that scenario. By continuously updating and refining this map, the headphones gain a deeper understanding of the noise characteristics of different scenarios, providing accurate and rich data support for subsequent training of the adaptive noise reduction parameter optimization model.

[0067] S205: training an adaptive noise reduction parameter optimization model in a scenario using a machine learning algorithm according to the noise feature distribution and the user operation record, wherein the adaptive noise reduction parameter optimization model is used to output an optimal noise reduction parameter configuration; After acquiring the scene's noise characteristic distribution and user operation records, the wireless headphones begin training the adaptive noise reduction parameter optimization model. First, the headphones preprocess the noise characteristic distribution data, normalizing the noise frequency distribution, intensity distribution, and duration distribution data to the same magnitude range to facilitate subsequent machine learning algorithm processing. The headphones then perform classification and encoding processing on the user operation records. Volume adjustment operations are converted into specific volume change values, noise reduction mode switching operations are encoded into different mode identifiers, and function button operations are also assigned corresponding codes. In this way, various user operations are converted into numerical forms that can be understood by the computer, providing a unified input format for the machine learning algorithm.

[0068] To train the model, the wireless earphones use neural network algorithms from deep learning, such as the Multilayer Perceptron (MLP). The MLP consists of an input layer, multiple hidden layers, and an output layer. The input layer receives preprocessed noise feature distribution data and user operation records, while the hidden layers perform complex feature extraction and transformation on the data using nonlinear activation functions. During training, the earphones divide the large amount of noise feature distribution data and user operation records into training and test sets. The training set is used to train the MLP model, continuously adjusting the model's weights and biases to enable the model to learn the relationship between noise characteristics, user operations, and optimal noise reduction parameters. For example, during training, the model attempts to predict the optimal noise reduction intensity and frequency adjustment parameters based on input data such as noise frequency distribution and intensity distribution, as well as user volume adjustment operations. If the predicted results deviate from the actual optimal parameters, the model calculates the error using a backpropagation algorithm and adjusts the weights and biases accordingly, gradually reducing the error and improving prediction accuracy.

[0069] The test set is used to evaluate model performance. During training, the headphones regularly test the model using the test set, calculating metrics such as prediction accuracy and mean squared error. When the model's performance on the test set stops improving, or the improvement is minimal, the model is considered to have achieved satisfactory training results. Training is terminated, resulting in an adaptive noise reduction parameter optimization model. This model outputs the optimal noise reduction parameter configuration for that scenario based on the input noise feature distribution and user operation history.

[0070] S206: storing the scene type identifier and the corresponding optimal noise reduction parameter configuration in a user-personalized scene-parameter mapping table, where the scene-parameter mapping table dynamically records the usage frequency of each scene; Once the adaptive noise cancellation parameter optimization model outputs the optimal noise cancellation parameter configuration, the wireless headphones store this configuration information, along with the corresponding scenario type identifier, in the user's personalized scenario-parameter mapping table. This mapping table is a dynamic data structure specifically designed to store and manage personalized noise cancellation parameter configurations for different scenarios. Each record in the table contains at least the following fields: scenario type identifier, optimal noise cancellation parameter configuration, usage frequency counter, and last usage timestamp.

[0071] During the storage process, the wireless headset first checks whether the scene type already exists in the mapping table. If it is a new scene, a new record is created, including the scene type identifier and the newly obtained optimal noise reduction parameter configuration. The usage frequency counter is initialized to 1 and the current timestamp is recorded. If the scene already exists, its optimal noise reduction parameter configuration is updated, the usage frequency counter is incremented, and the last usage timestamp is updated. For example, if the user uses the headset in an office scenario, the wireless headset will update the usage frequency count for the office scenario and update the configuration information based on the newly obtained optimal noise reduction parameters.

[0072] S207 , periodically detecting the usage frequency of each scene type identifier in the scene-parameter mapping table, and when the usage frequency of the target scene is lower than a preset frequency threshold, clearing the personalized configuration data of the target scene from the scene-parameter mapping table.

[0073] The wireless headset periodically checks and cleans the data stored in the scene-parameter mapping table (e.g., weekly or monthly). This periodic monitoring mechanism primarily involves the following steps: First, the wireless headset calculates the frequency of use of each scene within a specific time window. This frequency calculation can be performed using a variety of strategies, such as simple usage counts or weighted average frequency (with more recent use given a higher weight).

[0074] The wireless headset compares the calculated usage frequency with a preset frequency threshold. This threshold is set based on user habits and resource optimization considerations and can be dynamically adjusted based on actual usage. For example, if a scene has been used less than once a week in the past month, the wireless headset may identify it as a low-frequency usage scene.

[0075] When it is found that the usage frequency of a certain scene (target scene) is lower than the preset threshold, the wireless headset will start the cleanup process. Before officially clearing the data, the wireless headset will first back up the configuration data of these low-frequency usage scenarios to the secondary storage space so that it can be restored in the future when needed. Then, the wireless headset will delete all relevant data of the scene from the main mapping table, including scene type identification, noise reduction parameter configuration, usage frequency statistics and other information. This periodic cleanup mechanism helps to optimize the use of system resources and improve the efficiency of parameter query and update. At the same time, by promptly clearing the configuration data of low-frequency usage scenarios, it can be ensured that the mapping table retains the user's most commonly used scene configurations, enabling the wireless headset to respond more quickly to the user's usage needs in these high-frequency scenarios.

[0076] The following describes the wireless headset in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , is a schematic diagram of the physical device structure of the wireless headset in an embodiment of the present application.

[0077] It should be noted that Figure 3 The structure of the wireless headset shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0078] like Figure 3 As shown, the wireless headset includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes, such as the methods described in the above embodiments, based on programs stored in a read-only memory (ROM) 302 or programs loaded from a storage unit 308 into a random access memory (RAM) 303. RAM 303 also stores various programs and data required for system operation. CPU 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to bus 304.

[0079] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, push button switches, and the like; an output section 307 including a liquid crystal display (LCD), an audio output device, indicator lights, and the like; a storage section 308 including a hard disk and the like; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. Removable media 311, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 310 as needed, so that computer programs read from the removable media can be installed in the storage section 308 as needed.

[0080] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for executing the methods illustrated in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via the communication section 309 and / or installed from removable media 311. When executed by the central processing unit (CPU) 301, the computer program performs the various functions defined in the present invention.

[0081] It should be noted that specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0082] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings.

[0083] Specifically, the wireless headset of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, the adaptive anti-noise method for the wireless headset provided in the above embodiment is implemented.

[0084] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the wireless headset described in the above embodiments, or may exist independently and not be incorporated into the wireless headset. The storage medium carries one or more computer programs. When executed by a processor of the wireless headset, the one or more computer programs enable the wireless headset to implement the adaptive noise cancellation method for wireless headsets provided in the above embodiments.

[0085] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0086] As used in the above embodiments, the term “when” may be interpreted to mean “if” or “after” or “in response to determining that” or “in response to detecting that”, depending on the context. Similarly, the phrases “upon determining that” or “if (stated condition or event) is detected” may be interpreted to mean “if determining that” or “in response to determining that” or “upon detecting (stated condition or event)” or “in response to detecting (stated condition or event)”, depending on the context.

[0087] Those skilled in the art will appreciate that all or part of the process steps in the above-described method embodiments can be implemented by a computer program instructing the relevant hardware. The program can be stored in a computer-readable storage medium, and when executed, the program can include the process steps in the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A wireless headset adaptive anti-noise method, applied to wireless headsets, characterized in that: The method includes: obtaining acceleration information of the current user when using the headset through an accelerometer device in the headset; Obtain user head rotation information through the gyroscope in the headset; Obtain multiple audio signals corresponding to the earphones; Combining the multiple audio signals, determining suspected usage scenario information of the current wireless headset using a headset usage scenario model, wherein the headset usage scenario model is previously trained through deep learning using multiple audio information sets annotated with usage scenario information; Combining the acceleration information and the user's head rotation information, determining the actual usage scenario information of the current wireless headset through a scenario verification model, wherein the scenario verification model is previously trained through deep learning using multiple sets of acceleration information and user head rotation information annotated with usage scenario information; The anti-noise parameters of the wireless headset are adjusted based on the actual usage scenario information.

2. The method according to claim 1, characterized in that The step of adjusting the anti-noise parameters of the wireless headset based on the actual usage scenario information specifically includes: Obtaining initial anti-noise parameters corresponding to the actual usage scenario information from a preset scenario-parameter mapping table; Personalized adjustment of the initial anti-noise parameter based on the user's historical usage data to obtain personalized anti-noise parameter; Noise reduction processing is performed on the wireless headset according to the personalized anti-noise parameters.

3. The method according to claim 2, characterized in that After the step of obtaining the initial anti-noise parameters corresponding to the actual usage scenario information from the preset scenario-parameter mapping table, the method further includes: Acquire and record the user's usage data in the current real usage scenario, the usage data including at least the scenario type identifier, usage start and end time, ambient noise data, and user operation records; Calculating the cumulative usage time corresponding to the scenario type identifier based on the usage start and end times, and determining whether the cumulative usage time reaches a preset time threshold; When the accumulated usage time reaches the preset time threshold, determining a noise characteristic distribution of the scene based on the environmental noise data, the noise characteristic distribution including noise frequency distribution, noise intensity distribution, and noise duration distribution; Based on the noise feature distribution and the user operation record, an adaptive noise reduction parameter optimization model in the training scenario is trained by a machine learning algorithm, wherein the adaptive noise reduction parameter optimization model is used to output an optimal noise reduction parameter configuration; Storing the scene type identifier and the corresponding optimal noise reduction parameter configuration in a user-personalized scene-parameter mapping table, wherein the scene-parameter mapping table dynamically records the usage frequency of each scene; The usage frequency of each scene type identifier in the scene-parameter mapping table is periodically detected, and when the usage frequency of the target scene is lower than a preset frequency threshold, the personalized configuration data of the target scene is cleared from the scene-parameter mapping table.

4. The method according to claim 1, wherein The step of obtaining multiple audio signals corresponding to the earphone specifically includes: Enable both the microphone array inside and outside the headset; Performing spatial filtering on the collected audio signal; Calculating the correlation index of the external microphone signals corresponding to the microphone array; identifying noise source information based on the correlation index, the noise source information including the direction and distance of the noise source; The audio signal is classified and pre-processed according to the noise source information.

5. The method according to claim 1, wherein After the step of combining the plurality of audio signals and determining the suspected usage scenario information of the current wireless headset through the headset usage scenario model, the method further includes: determining a scene matching degree of suspected usage scene information matching the plurality of audio signals; If the scene matching degree is lower than a set threshold, the three usage scene information with the highest scene matching degrees are selected as suspected usage scene information.

6. The method according to claim 5, characterized in that If the scene matching degree is lower than a set threshold, after the step of selecting the three usage scene information with the highest scene matching degree as suspected usage scene information, the method further includes: Combining the acceleration information and the user head rotation information, sequentially verifying the three usage scenario information using a scenario verification model to determine scenario scores corresponding to the three usage scenario information; The target usage scenario information with the highest scenario score is sent to the user via voice for confirmation, so as to determine the actual usage scenario information of the current wireless headset.

7. The method according to claim 1, characterized in that After the step of adjusting the anti-noise parameters of the wireless headset in combination with the actual usage scenario information, the method further includes: Collecting multi-dimensional biometric information of the wearer through a biosensor, wherein the biometric information includes at least respiratory rate and body temperature; Based on the multi-dimensional biometric information, the wearer's real-time fatigue level is calculated using a fatigue assessment model. The fatigue assessment model is pre-trained using a deep learning method using fatigue level annotations based on different biometric characteristics. The real-time fatigue index is used to represent the current fatigue level of the headphone wearer. determining a fatigue adjustment coefficient according to the real-time fatigue degree, wherein the fatigue adjustment coefficient is positively correlated with the real-time fatigue degree; After obtaining the current basic anti-noise parameters of the wireless headset, adaptively fusing the fatigue adjustment coefficient with the basic anti-noise parameters to obtain optimized anti-noise parameters; The anti-noise parameters are adjusted according to the anti-noise optimized parameters.

8. A wireless headset, characterized in that: The wireless headset includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the wireless headset to execute the method according to any one of claims 1 to 7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instruction is executed on the wireless headset, the wireless headset is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on a wireless headset, the wireless headset is enabled to perform the method according to any one of claims 1 to 7.

Images