Detailed Description
The embodiment of the application provides a noise reduction calibration method and system for a wireless earphone. The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.
For easy understanding, a specific flow of an embodiment of the present application is described below, referring to fig. 1, and an embodiment of a noise reduction calibration method for a wireless earphone in an embodiment of the present application includes:
Step S101, initializing an acoustic model inside a wireless earphone to obtain a first initial earphone acoustic model and a second initial earphone acoustic model;
It can be understood that the execution body of the present application may be a noise reduction calibration system of a wireless earphone, and may also be a terminal or a server, which is not limited herein. The embodiment of the application is described by taking a server as an execution main body as an example.
Specifically, the internal acoustic characteristics of the wireless earphone are simulated and analyzed through ANSYS physics software, and the acoustic characteristic parameter set of the wireless earphone under different frequencies is calculated based on the acoustic simulation capability of the software. Based on acoustic characteristic parameters obtained through simulation, corresponding internal acoustic processing models are respectively created for a left earphone and a right earphone in the wireless earphone, and acoustic data of the left earphone and acoustic data of the right earphone are respectively processed by the two models so as to ensure that the noise reduction effects of the two ears are consistent in cooperation. The internal acoustic processing model mainly comprises a noise identification network, an inverse sound wave synthesis network and an inverse sound wave fusion network, and is used for effectively identifying the environmental noise, accurately synthesizing the inverse sound wave and finally carrying out fusion processing. And respectively generating an initialization network super-parameter set of the left earphone and the right earphone through the acoustic characteristic parameter set, wherein the super-parameter set directly influences the performance and the effect of the internal acoustic processing model. And performing super-parameter initialization by applying respective initialization network super-parameter sets to the left and right internal acoustic processing models to obtain a first initial earphone acoustic model and a second initial earphone acoustic model. These two models represent the initial states of the left and right headphones in the noise reduction calibration method, respectively.
Step S102, collecting a test environment noise signal and carrying out noise filtering processing on the test environment noise signal to generate a target noise signal;
Specifically, noise signals are collected from the test environment, and a microphone or other sound collection device with high sensitivity is used to ensure that various sounds in the environment can be captured. A noise reduction frequency range of the wireless earphone is defined, and the range indicates a specific frequency range to be focused on by noise reduction, which is usually the frequency range most sensitive to human ears, such as a middle-low frequency region, so that the noise reduction effect is more effectively improved and the sound quality is ensured. And amplifying the acquired environmental noise signals to increase the strength of the signals. And (3) carrying out signal standardization processing on the amplified environmental noise signals to ensure that the amplitude of the signals reaches a standard level. And removing high-frequency noise in the standard environmental noise signal through a filter, and only keeping the signal lower than a certain specific frequency according to a predefined noise reduction frequency range to obtain a first noise signal. And removing low-frequency noise in the first noise signal through a filter, ensuring that the finally obtained second noise signal is concentrated in the most critical frequency range according to the noise frequency range, and improving the noise reduction effect. And carrying out noise enhancement processing on the second noise signal, strengthening noise components important to noise reduction performance, removing or reducing the influence of noise components which are not important, and finally generating a target noise signal which is used as a reference in the noise reduction calibration process and is used for adjusting and optimizing a noise reduction algorithm and an acoustic model of the wireless earphone.
Step S103, performing NKP decomposition on the target noise signals to obtain a plurality of sub-noise signals;
Specifically, the target noise signal is subjected to fast fourier transform, and the target noise signal is converted into a frequency domain, so that the distribution characteristic of the noise signal in the frequency domain is obtained, and the frequency domain noise signal contains the frequency component of noise. And carrying out feature recognition on the frequency domain noise signals, and obtaining a noise feature set by recognizing key feature frequencies or modes in the frequency domain signals to reflect the main frequency components and the energy distribution condition of the target noise signals. And predicting decomposition feature points according to the noise feature set, matching the identified noise features with a predefined decomposition model, and predicting the decomposition feature points capable of representing key features of the noise signals. And predicting a signal decomposition section for the signal decomposition characteristic points, and predicting a characteristic section suitable for signal decomposition by analyzing the relation and distribution among the characteristic points. The determination of the characteristic interval is based on the characteristics of the noise signal and the decomposition target, capturing the most representative and influencing parts of the noise signal. And carrying out signal decomposition on the target noise signal based on the predicted multiple signal decomposition characteristic intervals to obtain multiple sub-noise signals. The signal decomposition process proceeds according to the specific characteristics of each characteristic interval, aiming at refining the target noise signal into several sub-noise signals, each representing a specific part or feature in the original noise signal.
Step S104, respectively carrying out inverse sound wave calculation and fusion on a plurality of sub-noise signals through a first initial earphone acoustic model and a second initial earphone acoustic model to obtain a first initial inverse sound wave and a second initial inverse sound wave;
Specifically, a plurality of sub-noise signals are respectively input to the noise recognition networks in the two initial earphone acoustic models. The noise recognition network extracts key noise features in each sub-noise signal using a deep learning algorithm. The noise identification network adopts a structure of a first double-layer long-short-term memory (LSTM) network, a gating circulation unit (GRU) network and a linear normalization layer, and the composite network structure is beneficial to capturing and extracting time sequence characteristics and dynamic changes in noise signals so as to more accurately identify the noise. The noise feature vector is input to an inverse acoustic wave synthesis network comprising an encoder and a decoder. The encoder section processes and compresses the noise signature through an input layer and a two-layer Convolutional Neural Network (CNN) to form a more compact and advanced noise signature representation. The decoder part is composed of a cyclic neural network (RNN) and a two-layer residual error network, and synthesizes the opposite-phase sound wave according to the noise characteristics after encoding. The synthesis process takes into account the fundamental nature of the noise and takes into account the time series characteristics of the noise to ensure that the generated inverted sound waves are effectively cancelled out by the original noise signal. And inputting the synthesized inverted sound wave of each sub-noise signal into an inverted sound wave fusion network, and fusing all the sub-inverted sound waves into a complete inverted sound wave. The fusion network adopts a structure of an attention weighted fusion layer and an output layer, wherein an attention mechanism can automatically adjust the weight of each sub-inverted sound wave according to the importance of each sub-inverted sound wave so as to ensure that more important characteristics are highlighted in the fusion process, thereby improving the quality and the noise reduction effect of the whole inverted sound wave. The whole process is carried out in parallel in two initial earphone acoustic models, and each model respectively processes noise signals of one side earphone, so that the collaborative consistency of the noise reduction performance of the left ear and the right ear is ensured. By processing, each model is able to derive an initial inverted sound wave for its processed signal, which is used in noise reduction calibration of the earphone to achieve a better noise reduction experience for the user.
Step S105, performing inverse filtering frequency equalization processing on the first initial inverted sound wave and the second initial inverted sound wave to obtain a first target inverted sound wave and a second target inverted sound wave;
Specifically, the first initial inverted sound wave is subjected to frequency band intensity characteristic analysis, and the energy distribution condition of the inverted sound wave in different frequency bands is known, so that the first frequency band intensity characteristic is obtained. The characteristics reflect which frequency bands of sound in the inverted sound wave are excessively strengthened or weakened. And similarly, carrying out frequency band intensity characteristic analysis on the second initial inverted sound wave to obtain a second frequency band intensity characteristic. And performing inverse filtering frequency intensity balance adjustment on the first initial inverted sound wave based on the intensity characteristics of the first frequency band. The intensity of sound waves in each frequency band is adjusted through an inverse filtering technology, so that the sound intensity of each frequency band can reach a more ideal and balanced state. The inverse filtering adjusts the sound wave based on an algorithm and an acoustic model to ensure that the adjusted sound wave can more accurately produce a cancellation effect with external noise. And similarly, the second initial inverted sound wave carries out inverse filtering frequency intensity balance adjustment according to the intensity characteristic of the second frequency band to obtain a second target inverted sound wave. Through the adjustment, the noise reduction effect is improved, and the problem of tone quality distortion caused by uneven frequency intensity is avoided. Finally, the obtained first target opposite-phase sound wave and the second target opposite-phase sound wave are more balanced in frequency intensity, and can better adapt to hearing requirements of different users and different external noise environments.
And S106, performing earphone acoustic model parameter compensation through the first target opposite-phase sound wave and the second target opposite-phase sound wave to obtain a first target earphone acoustic model and a second target earphone acoustic model.
Specifically, a first noise reduction evaluation index of the first target inverted sound wave and a second noise reduction evaluation index of the second target inverted sound wave are calculated respectively. The indexes comprehensively consider multiple aspects of the noise reduction effect, such as noise reduction depth, frequency response balance and the like, and provide accurate evaluation basis for parameter optimization. The obtained first and second noise reduction evaluation indexes reflect the noise reduction performance of the respective opposite-phase sound waves. A multi-objective fitness function is defined based on the noise reduction evaluation index, and is used in a genetic optimization algorithm to simultaneously optimize a plurality of noise reduction performance indexes so as to ensure that the final noise reduction effect is efficient and balanced. The genetic optimization algorithm is a search algorithm simulating natural selection and genetic principles, and continuously optimizes the quality of the solution through selection, crossover and mutation operations in an iterative process. Initializing a parameter population for the first initial earpiece acoustic model and the second initial earpiece acoustic model using a genetic optimization algorithm, resulting in a plurality of first compensation parameter sets. This is the starting point for finding the optimal noise reduction parameter configuration, each parameter set representing one possible earphone acoustic model parameter configuration. And evaluating the first compensation parameter set through the multi-target fitness function, calculating the fitness value of each parameter set, and identifying which parameter configurations can improve the noise reduction effect. And generating a plurality of corresponding second compensation parameter sets according to the calculated fitness value, and carrying out optimization solution on the parameter sets to finally obtain a target compensation parameter set. By continuously iterating the optimization, the optimal earphone acoustic model parameter configuration is approximated gradually, so that the optimal noise reduction effect can be ensured. And carrying out earphone acoustic model parameter compensation on the first initial earphone acoustic model and the second initial earphone acoustic model based on the obtained target compensation parameter set to obtain a first target earphone acoustic model and a second target earphone acoustic model.
In the embodiment of the application, the noise characteristics of various complex environments can be accurately captured by collecting the noise signals of the test environment and carrying out the fine noise filtering treatment, so that an accurate data basis is provided for the subsequent noise reduction treatment. The accurate identification and analysis of the environmental noise ensures that the noise reduction system can adapt to changeable environmental conditions and provides stable and effective noise reduction effect. Through initialization and parameter compensation of an acoustic model in the earphone, the noise reduction effect adjustment can be performed in a personalized way according to the use environments and personal preferences of different users by combining an advanced acoustic processing model comprising a noise recognition network, an inverted sound wave synthesis network and an inverted sound wave fusion network. The personalized adjustment not only can improve the use satisfaction degree of users, but also can improve the market competitiveness of earphone products. And carrying out parameter compensation on the acoustic model of the earphone by utilizing a genetic optimization algorithm, so that the noise reduction performance of the earphone can be dynamically optimized and adjusted according to real-time feedback. The dynamic adjustment mechanism not only enhances the adaptability of the earphone to different noise environments, but also enables the earphone to better meet the hearing demand of users along with time change. Through carrying out NKP decomposition on the target noise signal to obtain a plurality of sub-noise signals, and carrying out calculation and fusion of the opposite-phase sound waves on the sub-noise signals, the noise signals can be processed more finely, and the noise reduction accuracy and effect are improved. Meanwhile, the inverse filtering frequency equalization processing further improves the quality of the inverse sound wave, and the maximization of the noise reduction effect is ensured. From the acoustic model initialization to the final model parameter compensation, a closed-loop optimization system is formed. The wireless earphone not only improves the noise reduction effect, but also is beneficial to improving the overall performance of the earphone, thereby realizing the intelligent noise reduction calibration of the wireless earphone and improving the noise reduction effect of the wireless earphone.
In a specific embodiment, the process of executing step S101 may specifically include the following steps:
(1) Performing earphone internal acoustic characteristic simulation on the wireless earphone through ANSYS physical software, and calculating acoustic characteristic parameter sets of the wireless earphone under different frequencies;
(2) Creating a left internal acoustic processing model corresponding to a left earphone in the wireless earphone and a right internal acoustic processing model corresponding to a right earphone in the wireless earphone, wherein the left internal acoustic processing model and the right internal acoustic processing model have the same model framework, and the left internal acoustic processing model or the right internal acoustic processing model comprises a noise identification network, an inverse sound wave synthesis network and an inverse sound wave fusion network;
(3) Generating a left initialization network super-parameter set and a right initialization network super-parameter set through the acoustic characteristic parameter set;
(4) And initializing the network super parameters of the left internal acoustic processing model through the left initializing network super parameter set to obtain a first initial earphone acoustic model, and initializing the network super parameters of the right internal acoustic processing model through the right initializing network super parameter set to obtain a second initial earphone acoustic model.
Specifically, the wireless earphone is subjected to acoustic characteristic simulation through ANSYS physical software, so that phenomena such as sound wave propagation and reflection in the earphone and influence of earphone materials on acoustic performance are simulated. And calculating the acoustic response of the wireless earphone under different frequencies by software to obtain a comprehensive acoustic characteristic parameter set. These parameter sets include the response of the earpiece at various frequency points, also reflecting the effect of the earpiece structure on the sound propagation. Based on the obtained acoustic characteristic parameter set, corresponding internal acoustic processing models are respectively created for a left earphone and a right earphone in the wireless earphone. The two models are consistent in structure, but correspond to left and right channels of the earphone respectively, so that independent noise reduction treatment can be carried out on sounds on two sides. Each internal acoustic processing model comprises a noise identification network, an inverse acoustic wave synthesis network and an inverse acoustic wave fusion network, wherein the three networks are respectively responsible for identifying specific characteristics in environmental noise, synthesizing effective inverse acoustic waves according to the characteristics and effectively fusing the synthesized multiple inverse acoustic waves into a single inverse acoustic wave so as to achieve the optimal noise reduction effect. In order to enable these internal acoustic processing models to perform effective noise reduction processing according to actual earphone acoustic characteristics, left and right initialized network super-parameter sets are generated by the acoustic characteristic parameter sets. The network super parameter set refers to parameters controlling the network training process and the network structure configuration, such as learning rate, network layer number, node number of each layer, etc. The purpose of generating these super parameter sets is to enable the network to better adapt to the acoustic characteristics of the headphones, improving the accuracy and efficiency of network recognition noise and synthesizing inverted sound waves. And initializing the left internal acoustic processing model by using the left initialization network super-parameter set, and initializing the right internal acoustic processing model by using the right initialization network super-parameter set to obtain a first initial earphone acoustic model and a second initial earphone acoustic model which respectively represent the acoustic processing capacity of the left earphone and the right earphone after preliminary optimization.
In a specific embodiment, the process of executing step S102 may specifically include the following steps:
(1) Collecting a test environment noise signal and defining a noise reduction frequency range of the wireless earphone;
(2) Amplifying the test environmental noise signal to obtain an amplified environmental noise signal;
(3) Performing signal standardization processing on the amplified environmental noise signal to obtain a standard environmental noise signal;
(4) High-frequency noise removal is carried out on the standard environmental noise signal according to the noise reduction frequency range, so that a first noise signal is obtained;
(5) Performing low-frequency noise removal on the first noise signal according to the noise reduction frequency range to obtain a second noise signal;
(6) And performing noise enhancement processing on the second noise signal to generate a target noise signal.
Specifically, noise signals in the test environment are collected. During the acquisition process, the high-sensitivity microphone equipment is used for collecting noise data in an actual or simulated use environment, so that the captured noise signal is ensured to reflect the noise characteristics in the real environment as much as possible. The noise reduction frequency range of the wireless earphone is defined, the interference degree of noise with different frequencies on a user is different, and the effective noise reduction frequency range should cover the frequency band most sensitive to human ears, which usually relates to a middle-low frequency range, wherein human voice and most of environmental noise are concentrated. And the signal amplification processing is carried out on the test environment noise signal, so that the identifiability of the noise signal in the subsequent processing is ensured, and the characteristics of the signal are more obvious. And (3) carrying out signal standardization processing on the amplified environmental noise signals, and enabling the amplified environmental noise signals to reach a certain standard level by adjusting the signal intensity. The normalized ambient noise signal is processed through a filter to remove high frequency noise that is not in the noise reduction frequency range. The filter effectively masks out frequency components above a set threshold and allows only low frequency components to pass. This can eliminate unnecessary high frequency interference, focusing on noise bands that have a greater impact on the user. And (3) removing low-frequency noise from the noise signals subjected to filtering, removing frequency components lower than the noise reduction frequency range, and ensuring that the rest noise signals are concentrated in the most critical frequency band. And carrying out noise enhancement processing on the second noise signal, and enhancing key characteristics in the noise signal on the premise of not changing noise characteristics so that the key characteristics are easier to identify and cancel in subsequent noise reduction processing.
In a specific embodiment, the process of executing step S103 may specifically include the following steps:
(1) Performing fast Fourier transform on the target noise signal to obtain a frequency domain noise signal;
(2) Performing feature recognition on the frequency domain noise signals to obtain a noise feature set;
(3) According to the noise characteristic set, carrying out decomposition characteristic point prediction on the target noise signal to obtain a plurality of signal decomposition characteristic points;
(4) Predicting signal decomposition sections of the signal decomposition feature points to obtain a plurality of signal decomposition feature sections;
(5) And carrying out signal decomposition on the target noise signal based on the plurality of signal decomposition characteristic intervals to obtain a plurality of sub-noise signals.
Specifically, the target noise signal is subjected to fast fourier transformation, the noise signal in the time domain is converted into a frequency domain representation, and the noise signal is easier to analyze and process in the frequency domain. The fast fourier transform allows the system to identify the presence and intensity of individual frequency components in the noise signal. Then, feature recognition is performed on the frequency domain noise signal. The frequency domain noise signals are analyzed to identify representative noise characteristics, such as peaks of specific frequencies, energy distribution patterns, etc., and the noise characteristic sets contain key information of the noise signals. And performing decomposition feature point prediction based on the identified noise feature set. The feature points playing a key role in the noise signal are predicted according to the feature set of the noise signal, and the feature points are the most representative parts in the noise signal and can be regarded as key nodes of noise processing. Then, signal decomposition section prediction is performed on the plurality of predicted signal decomposition feature points. And predicting a series of signal decomposition characteristic intervals by taking the identified characteristic points as references, wherein the intervals are divided according to the positions of the characteristic points and the characteristics of noise signals, and noise components in each interval have certain commonality or continuity. The purpose of signal decomposition interval prediction is to divide the noise signal more accurately so that subsequent signal decomposition and processing can be more targeted. And carrying out signal decomposition on the target noise signal based on the predicted multiple signal decomposition characteristic intervals to finally obtain multiple sub-noise signals. Each sub-noise signal represents a portion of the original noise signal but has more definite characteristics and frequency components.
In a specific embodiment, the process of executing step S104 may specifically include the following steps:
(1) Respectively inputting a plurality of sub-noise signals into a noise recognition network in a first initial earphone acoustic model, and extracting a first noise characteristic vector of each sub-noise signal through the noise recognition network, wherein the noise recognition network comprises a first double-layer LSTM network, a first GRU network and a first linear normalization layer;
(2) Respectively inputting a first noise characteristic vector of each sub-noise signal into an inverted sound wave synthesis network in a first initial earphone acoustic model, and respectively synthesizing a first sub-inverted sound wave of each sub-noise signal through the inverted sound wave synthesis network, wherein the inverted sound wave synthesis network comprises a first encoder and a first decoder, the first encoder comprises a first input layer and a first two-layer CNN network, and the first decoder comprises a first RNN network and a first two-layer residual network in sequence;
(3) Respectively inputting the first sub-opposite-phase sound wave of each sub-noise signal into an opposite-phase sound wave fusion network in a first initial earphone acoustic model, and carrying out opposite-phase sound wave fusion on the first sub-opposite-phase sound wave of each sub-noise signal through the opposite-phase sound wave fusion network to obtain a first initial opposite-phase sound wave, wherein the opposite-phase sound wave fusion network comprises a first attention weighting fusion layer and a first output layer;
(4) Respectively inputting a plurality of sub-noise signals into a noise recognition network in a second initial earphone acoustic model, and extracting a second noise characteristic vector of each sub-noise signal through the noise recognition network, wherein the noise recognition network comprises a second double-layer LSTM network, a second GRU network and a second linear normalization layer;
(5) Respectively inputting a second noise characteristic vector of each sub-noise signal into an inverted acoustic wave synthesis network in a second initial earphone acoustic model, and respectively synthesizing a second sub-inverted acoustic wave of each sub-noise signal through the inverted acoustic wave synthesis network, wherein the inverted acoustic wave synthesis network comprises a second encoder and a second decoder, the second encoder comprises a second input layer and a second two-layer CNN network, and the second decoder comprises a second RNN network and a second two-layer residual error network in sequence;
(6) And respectively inputting the second sub-opposite-phase sound waves of each sub-noise signal into an opposite-phase sound wave fusion network in a second initial earphone acoustic model, and carrying out opposite-phase sound wave fusion on the second sub-opposite-phase sound waves of each sub-noise signal through the opposite-phase sound wave fusion network to obtain second initial opposite-phase sound waves, wherein the opposite-phase sound wave fusion network comprises a second attention weighting fusion layer and a second output layer.
Specifically, the sub-noise signals are input to a noise recognition network in the earphone acoustic model, and each sub-noise signal is recognized and characteristic extracted. The noise recognition network adopts a deep learning technology, comprises a double-layer long and short-term memory (LSTM) network and a gating and circulating unit (GRU) network, and a linear normalization layer, and can work together to extract key feature vectors from complex noise signals. The first dual-layer LSTM network works in conjunction with the first GRU network to capture time-series dependencies in the noise signal, thereby effectively extracting temporally continuous noise features. Through linear normalization layer processing, the numerical stability of the feature vector is ensured, and standardized input is provided for subsequent inverse acoustic wave synthesis. The eigenvectors of each sub-noise signal are input into an inverted acoustic wave synthesizing network designed to synthesize an effective inverted acoustic wave from the noise eigenvectors. The synthesis network is composed of an encoder and a decoder, wherein the encoder comprises an input layer and a two-layer Convolutional Neural Network (CNN), the CNN layer extracting and compressing noise features. The decoder operates through a cyclic neural network (RNN) and a two-layer residual network, recovering and generating an inverted acoustic wave having a target phase based on the compressed noise characteristics. This relies on neural network models to model the complexity of the noise while utilizing residual networks to improve the learning ability of the network to ensure that the generated anti-phase sound waves can effectively cancel out the original noise signal. And inputting each generated sub-inverted sound wave into an inverted sound wave fusion network. The network adopts an attention weighted fusion layer which can automatically adjust the weights of each sub-inverted sound wave in the final output according to the contribution degree of each sub-inverted sound wave so as to realize the optimal noise cancellation effect. By means of weighted fusion, it can be ensured that sub-inverted sound waves which have a greater contribution to noise reduction are given higher priority, so that high-quality initial inverted sound waves are generated at the output layer. The steps are respectively carried out on the left channel and the right channel of the earphone so as to ensure the consistency and the optimization of the double-ear noise reduction effect. The noise recognition, the inverse sound wave synthesis and the fusion network of the left and right sound channels are consistent in structure, but the processed signals and the generated inverse sound waves are different, and the independent noise reduction requirements of the two sound channels of the earphone are reflected.
In a specific embodiment, the process of executing step S105 may specifically include the following steps:
(1) Performing frequency band intensity characteristic analysis on the first initial inverted sound wave to obtain a first frequency band intensity characteristic;
(2) Performing frequency band intensity characteristic analysis on the second initial inverted sound wave to obtain a second frequency band intensity characteristic;
(3) Performing inverse filtering frequency intensity balance adjustment on the first initial inverted sound wave according to the intensity characteristics of the first frequency band to obtain a first target inverted sound wave;
(4) And performing inverse filtering frequency intensity balance adjustment on the second initial inverted sound wave according to the intensity characteristic of the second frequency band to obtain a second target inverted sound wave.
Specifically, the first initial inverted sound wave is subjected to frequency segment intensity characteristic analysis, the energy distribution situation of the inverted sound wave in different frequency segments is quantified, and the sound intensities of the frequency segments are known to be beyond or below the expected level. And meanwhile, carrying out frequency band intensity characteristic analysis on the second initial inverted sound wave to obtain a second frequency band intensity characteristic. The comprehensive consideration of the output of the binaural earphone is ensured, so that the noise reduction processing can be uniformly and effectively applied to the left side and the right side of the binaural earphone, and the consistency and the balance of binaural hearing experience are ensured. And then, carrying out inverse filtering frequency intensity balance adjustment on the initial inverted sound wave according to the intensity characteristics of the frequency band. According to the intensity characteristics of the first frequency band, the first initial inverted sound wave is adjusted through an inverse filtering technology, and the sound intensity of each frequency band is balanced, so that the sound intensity is closer to an ideal noise reduction performance curve. The inverse filtering process includes reducing the frequency content of the excessively strong and enhancing certain frequency bins of the excessively weak to ensure the equalization and effectiveness of the inverted acoustic wave over the entire frequency range. Similarly, the second initial inverted sound wave undergoes an inverse filtering frequency intensity equalization adjustment process according to the intensity characteristics of the second frequency band. Ensuring that the right ear inverted sound wave also achieves the optimized frequency response characteristics. Through the above steps, the first and second initial inverted sound waves are adjusted to the first and second target inverted sound waves, respectively. After the inverse filtering frequency intensity is regulated, the target inverse sound waves can more accurately offset external noise, and the problem of tone quality distortion possibly introduced is minimized, so that a more natural and clear hearing experience is provided.
In a specific embodiment, the process of executing step S106 may specifically include the following steps:
(1) Respectively calculating a first noise reduction evaluation index of the first target inverted sound wave and a second noise reduction evaluation index of the second target inverted sound wave;
(2) Defining a multi-objective fitness function of a genetic optimization algorithm based on the first noise reduction evaluation index and the second noise reduction evaluation index;
(3) Initializing noise reduction calibration compensation parameter populations of the first initial earphone acoustic model and the second initial earphone acoustic model through a genetic optimization algorithm to obtain a plurality of first compensation parameter sets;
(4) Carrying out fitness calculation on a plurality of first compensation parameter sets through a multi-target fitness function to obtain a fitness value of each first compensation parameter set;
(5) Generating a plurality of corresponding second compensation parameter sets according to the fitness value, and carrying out optimization solution on the plurality of second compensation parameter sets to obtain a target compensation parameter set;
(6) And performing earphone acoustic model parameter compensation on the first initial earphone acoustic model and the second initial earphone acoustic model based on the target compensation parameter set to obtain a first target earphone acoustic model and a second target earphone acoustic model.
Specifically, a first noise reduction evaluation index of the first target inverted sound wave and a second noise reduction evaluation index of the second target inverted sound wave are calculated respectively, and these indexes can quantitatively reflect the quality of the noise reduction effect, including noise reduction depth, uniformity of frequency response, suppression efficiency of environmental noise, and the like. A multi-objective fitness function of the genetic optimization algorithm is defined based on the evaluation index. The genetic optimization algorithm is a search algorithm simulating natural selection and genetic mechanism in the biological evolution process, and finds the optimal solution of the problem through iterative evolution of the population. The multi-objective fitness function aims at simultaneously optimizing a plurality of noise reduction performance indexes to achieve a comprehensive optimal noise reduction effect, and ensures that not only a single performance index is concerned in the optimization process, but all aspects of the noise reduction effect are comprehensively considered to obtain the optimal overall noise reduction performance. And initializing noise reduction calibration compensation parameter populations for the first initial earphone acoustic model and the second initial earphone acoustic model through a genetic optimization algorithm. A plurality of candidate compensation parameter sets are generated, each parameter set representing one possible earphone acoustic model configuration. And calculating the fitness of the candidate compensation parameter set through the multi-target fitness function. The fitness value of each parameter set reflects the performance of its corresponding configuration in terms of noise reduction performance, with a high fitness value meaning a better noise reduction effect. And by comparing the fitness values of different parameter sets, identifying which parameter configurations are closer to the ideal noise reduction effect. And generating a plurality of optimized compensation parameter sets according to the fitness value by the algorithm. And carrying out optimization solution on the optimized parameter set, and finding out a final target compensation parameter set through iterative optimization. The target parameter set represents the earphone acoustic model configuration capable of achieving the optimal noise reduction effect under the current algorithm and model conditions. And carrying out final parameter compensation on the first initial earphone acoustic model and the second initial earphone acoustic model based on the target compensation parameter set to obtain a first target earphone acoustic model and a second target earphone acoustic model. The two models respectively provide optimal noise reduction configurations for left and right channels of the earphone, and ensure that the wireless earphone can provide excellent noise reduction effects in various environments.
The method for calibrating the noise reduction of the wireless earphone in the embodiment of the present application is described above, and the system for calibrating the noise reduction of the wireless earphone in the embodiment of the present application is described below, referring to fig. 2, an embodiment of the system for calibrating the noise reduction of the wireless earphone in the embodiment of the present application includes:
An initialization module 201, configured to initialize an acoustic model inside the wireless headset, so as to obtain a first initial headset acoustic model and a second initial headset acoustic model;
The filtering module 202 is configured to collect a test environmental noise signal and perform noise filtering processing on the test environmental noise signal to generate a target noise signal;
A decomposition module 203, configured to perform NKP decomposition on the target noise signal to obtain a plurality of sub-noise signals;
The calculating module 204 is configured to calculate and fuse the inverted sound waves of the plurality of sub-noise signals through the first initial earphone acoustic model and the second initial earphone acoustic model, so as to obtain a first initial inverted sound wave and a second initial inverted sound wave;
The equalization module 205 is configured to perform inverse filtering frequency equalization on the first initial inverted sound wave and the second initial inverted sound wave to obtain a first target inverted sound wave and a second target inverted sound wave;
And the compensation module 206 is configured to perform earphone acoustic model parameter compensation through the first target inverted acoustic wave and the second target inverted acoustic wave, so as to obtain a first target earphone acoustic model and a second target earphone acoustic model.
Through the cooperation of the components, the noise characteristics of various complex environments can be accurately captured by collecting the noise signals of the test environment and carrying out refined noise filtering treatment, so that an accurate data base is provided for subsequent noise reduction treatment. The accurate identification and analysis of the environmental noise ensures that the noise reduction system can adapt to changeable environmental conditions and provides stable and effective noise reduction effect. Through initialization and parameter compensation of an acoustic model in the earphone, the noise reduction effect adjustment can be performed in a personalized way according to the use environments and personal preferences of different users by combining an advanced acoustic processing model comprising a noise recognition network, an inverted sound wave synthesis network and an inverted sound wave fusion network. The personalized adjustment not only can improve the use satisfaction degree of users, but also can improve the market competitiveness of earphone products. And carrying out parameter compensation on the acoustic model of the earphone by utilizing a genetic optimization algorithm, so that the noise reduction performance of the earphone can be dynamically optimized and adjusted according to real-time feedback. The dynamic adjustment mechanism not only enhances the adaptability of the earphone to different noise environments, but also enables the earphone to better meet the hearing demand of users along with time change. Through carrying out NKP decomposition on the target noise signal to obtain a plurality of sub-noise signals, and carrying out calculation and fusion of the opposite-phase sound waves on the sub-noise signals, the noise signals can be processed more finely, and the noise reduction accuracy and effect are improved. Meanwhile, the inverse filtering frequency equalization processing further improves the quality of the inverse sound wave, and the maximization of the noise reduction effect is ensured. From the acoustic model initialization to the final model parameter compensation, a closed-loop optimization system is formed. The wireless earphone not only improves the noise reduction effect, but also is beneficial to improving the overall performance of the earphone, thereby realizing the intelligent noise reduction calibration of the wireless earphone and improving the noise reduction effect of the wireless earphone.
The application also provides a noise reduction calibration device of the wireless earphone, which comprises a memory and a processor, wherein the memory stores computer readable instructions, and the computer readable instructions, when executed by the processor, cause the processor to execute the steps of the noise reduction calibration method of the wireless earphone in the above embodiments.
The present application also provides a computer readable storage medium, which may be a non-volatile computer readable storage medium, and may also be a volatile computer readable storage medium, where instructions are stored in the computer readable storage medium, when the instructions are executed on a computer, cause the computer to perform the steps of the noise reduction calibration method of the wireless headset.
It will be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, systems and units may refer to the corresponding processes in the foregoing method embodiments, which are not repeated herein.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present application. The storage medium includes a U disk, a removable hard disk, a read-only memory (ROM), a random access memory (random acceS memory, RAM), a magnetic disk, an optical disk, or other various media capable of storing program codes.
While the application has been described in detail with reference to the foregoing embodiments, it will be understood by those skilled in the art that the foregoing embodiments may be modified or equivalents may be substituted for some of the features thereof, and that the modifications or substitutions do not depart from the spirit and scope of the embodiments of the application.