Balancing method and system for Chirp modulation signal of audio equipment

Abstract

The invention relates to the technical field of signal processing of intelligent voice equipment, and discloses an equalization method and system for a Chirp modulation signal of audio equipment, and the equalization method for the Chirp modulation signal of the audio equipment comprises the steps: analyzing the non-stationary characteristic of the Chirp modulation signal through an information geometric mapping algorithm, and generating non-stationary complexity mapping; processing the Chirp modulation signal by using a multi-scale entropy analysis algorithm to generate a multi-scale complexity spectrum; extracting long-range correlation characteristics of the Chirp modulation signals by adopting a fractal analysis algorithm, and outputting Hurst index characteristic mapping; constructing a complexity-perceived adaptive equalization system; a time-frequency joint optimization algorithm is constructed, and global optimization and adaptive adjustment of equalization parameters are realized; signal complexity characteristics are quantized by using a multi-scale entropy analysis algorithm, and the adaptability of the equalization system to signals with different complexities is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing of intelligent voice equipment, and more specifically, to an equalization method and system for Chirp modulation signals of audio equipment. Background Art

[0002] Chirp modulation signals, as a special type of audio signal, are widely used in near-field communications (NFC) for audio devices such as smart speakers and smartphones. In scenarios like smart homes and wearables, chirp signals enable communication through audio devices without the need for additional hardware. However, in real-world applications, chirp signals often face various acoustic interferences, such as multipath, reverberation, and background noise. These interferences can cause signal distortion and phase distortion, severely impacting communication quality and reliability.

[0003] In the existing technology, the following methods are mainly used for chirp signal equalization: traditional linear equalization methods, which have low computational complexity but their performance drops sharply in nonlinear and non-stationary channel environments; adaptive equalization technology, which can dynamically adapt to environmental changes but does not take into account the non-stationary and complex characteristics of the signal; machine learning-based equalization methods, although they can handle complex environments, have high computational resource requirements and are difficult to run in real time on resource-constrained devices.

[0004] The above method has the following technical problems when processing Chirp modulated signals: it cannot accurately characterize the non-stationary characteristics of Chirp signals, resulting in differences in equalization effects at different times; it lacks the ability to perceive the multi-scale complexity of signals, making it difficult to adapt to the equalization requirements of signals of different complexities; it ignores the long-range correlation characteristics of signals, resulting in unstable equalization performance in continuous Chirp signal transmission; it cannot achieve a balance between real-time and high performance, and the allocation of computing resources is unreasonable.

[0005] Therefore, there is an urgent need for an equalization method that can accurately characterize the non-stationary characteristics, multi-scale complexity and long-range correlation characteristics of Chirp signals to achieve high-performance adaptive equalization under resource-constrained conditions. Summary of the Invention

[0006] The present invention provides an equalization method and system for a Chirp modulated signal of an audio device, which solves the technical problem in related technologies that the non-stationary characteristics, multi-scale complexity and long-range correlation characteristics of the Chirp signal cannot be accurately characterized.

[0007] The present invention provides a method for equalizing a Chirp modulation signal of an audio device, comprising: The information geometry mapping algorithm is used to analyze the non-stationary characteristics of Chirp modulated signals and generate non-stationary complexity mapping; Apply multi-scale entropy analysis algorithm to process Chirp modulated signal and generate multi-scale complexity spectrum; The fractal analysis algorithm is used to extract the long-range correlation characteristics of the Chirp modulation signal and output the Hurst exponent feature map; Based on non-stationary complexity mapping, multi-scale complexity spectrum, and Hurst exponent feature mapping, a complexity-aware adaptive equalization system is constructed to achieve adaptive equalization processing of Chirp signals. Based on the adaptive equalization processing results, a time-frequency joint optimization algorithm is constructed to achieve global optimization and adaptive adjustment of equalization parameters.

[0008] Furthermore, the step of analyzing the non-stationary characteristics of the Chirp modulated signal using the information geometry mapping algorithm includes: Receive the original Chirp modulated signal; Apply short-time Fourier transform to convert the Chirp signal into time-frequency representation; Construct a signal probability distribution model based on time-frequency representation and map the signal onto a Riemannian manifold; Construct the Fisher information matrix and calculate the Fisher information matrix elements corresponding to the probability distribution of adjacent time points on the manifold; Based on the Fisher information matrix, the geodesic distance between two distribution points on the statistical manifold is calculated to quantify the degree of signal non-stationarity; Based on the calculation results of geodesic distance, a non-stationary complexity map of Chirp signal is constructed.

[0009] Furthermore, the step of applying a multi-scale entropy analysis algorithm to process the Chirp modulation signal includes: Segment the original Chirp signal and divide it into multiple data segments; Apply coarse-graining processing to each data segment to generate multiple different scale factors The coarse-grained time series under Calculate the sample entropy of the coarse-grained time series at each scale; Calculate multi-scale entropy indicators and construct entropy spectra that reflect the complexity changes of signals at different time scales; Identify the characteristic scale and key change points of Chirp signals based on entropy spectrum; Combining non-stationary complexity mapping and multi-scale entropy spectrum, a comprehensive complexity feature space is constructed to generate a two-dimensional complexity spectrum.

[0010] Furthermore, the step of extracting the long-range correlation characteristics of the Chirp modulation signal using a fractal analysis algorithm includes: Apply discrete wavelet transform to the original Chirp signal for multi-resolution decomposition to obtain wavelet coefficients of different scales; Calculate the variance of wavelet coefficients at each scale; Analyze the relationship between the variance of wavelet coefficients and scale in a double logarithmic coordinate system and fit a linear model; Calculate the Hausdorff dimension and fractal index of the signal based on the wavelet variance slope; The local Hurst exponent was estimated using the rescaled range analysis method; The local Hurst exponent is calculated based on the sliding window, and the Hurst exponent feature map is constructed.

[0011] Furthermore, the step of constructing a complexity-aware adaptive equalization system includes: Based on non-stationary complexity mapping, multi-scale complexity spectrum and Hurst exponent characteristic mapping, a comprehensive evaluation index of signal complexity is constructed; According to the comprehensive complexity evaluation index, the Chirp signal is adaptively segmented and divided into multiple regions with different complexities; Establish a balanced parameter library, which contains balanced parameter configurations applicable to areas of different complexity; For each signal segment, the optimal equalization parameter configuration is selected according to its complexity characteristics; Construct a complexity adaptive equalizer and apply differentiated equalization strategies to different complexity areas; Apply a smooth transition strategy between adjacent signal areas to avoid abrupt changes in the equalization effect at the area boundaries; Evaluate the quality of the equalized signal and calculate the equalization performance index; Based on the balancing performance indicators, the balancing parameter library is updated online to optimize the balancing performance.

[0012] Furthermore, the calculation formula for constructing the comprehensive evaluation index of signal complexity is: ; in Indicates time Comprehensive evaluation index of signal complexity at ; 、 and They are non-stationary complexity weight, multi-scale complexity weight and long-range correlation weight respectively; is a non-stationary complexity map, which means that the signal The degree of non-stationarity at is the scale factor; is the selected set of scale factors; For scale The entropy value under is the time scale correlation function, characterizing a specific scale In time the importance of For time Hurst index at is the Hurst exponential mapping function, which is used to quantify the contribution of long-range correlation to complexity; For all selected scale factors Summation.

[0013] Furthermore, the step of constructing the time-frequency joint optimization algorithm includes: Construct a joint time-frequency objective function that comprehensively considers the performance of equalization in both the time and frequency domains; Constructing time-domain objective function based on information geometry metric; Construct frequency domain objective function based on spectral distance; Constructing a time-frequency joint objective function based on multi-scale entropy differences; The gradient descent algorithm is used to optimize the time-frequency joint objective function and iteratively update the equalization parameters; Introducing a signal complexity adaptive gradient update strategy and adopting differentiated parameter update methods for different complexity regions; Construct a long-range correlation-aware regularization term to enhance the equalization's ability to handle long-term signal dependencies; Incorporate the long-range correlation regularization term into the objective function to obtain the final optimization goal; The gradient descent algorithm is combined with the signal complexity adaptive gradient update strategy to optimize the final optimization target and obtain the optimal equalization parameter configuration.

[0014] Furthermore, the calculation formula of the time-frequency joint objective function is: ; in represents the time-frequency joint objective function, which is the overall goal of optimizing the equalization parameters. is the time domain objective function, used to evaluate the performance of equalization in the time domain; is the frequency domain objective function, which is used to evaluate the performance of equalization in the frequency domain; is the time-frequency joint objective function, which is used to evaluate the performance of equalization in the time-frequency joint domain; 、 and Represent the weight coefficients of the time domain objective function, frequency domain objective function and time-frequency joint objective function respectively; represents the time variable, represents a frequency variable, Represents a set of equalization parameters.

[0015] Furthermore, the final optimization objective is expressed as: ; in represents the final optimization objective function, represents the time-frequency joint objective function, A regularization term that represents the perception of long-range dependencies; represents the time variable, represents a frequency variable, Represents a set of equalization parameters.

[0016] The present invention provides an equalization system for a Chirp modulation signal of an audio device, which is used to perform the above-mentioned equalization method for a Chirp modulation signal of an audio device, comprising: Information geometry mapping module, used to analyze the non-stationary characteristics of Chirp modulated signals and generate non-stationary complexity mapping; Multi-scale entropy analysis module, used to process Chirp modulated signals and generate multi-scale complexity spectra; Fractal analysis module, used to extract the long-range correlation characteristics of Chirp modulation signals and output Hurst exponent feature map; Complexity-aware adaptive equalization module, used to implement adaptive equalization processing of Chirp signals; The time-frequency joint optimization module is used to achieve global optimization and adaptive adjustment of equalization parameters.

[0017] The beneficial effects of the present invention are: accurately characterizing the non-stationary characteristics of Chirp signals through the information geometry mapping algorithm, solving the limitation problem of traditional equalization technology in non-stationary signal processing; The multi-scale entropy analysis algorithm is used to quantify the signal complexity characteristics, which enhances the adaptability of the equalization system to signals of different complexities. The fractal analysis technology is introduced to extract the long-range correlation characteristics of the signal, which improves the equilibrium stability in the transmission of continuous Chirp signals. The complexity-aware adaptive equalization system built based on the above features can dynamically allocate computing resources according to signal characteristics, reducing computational complexity while ensuring performance; A time-frequency joint optimization algorithm is designed to take into account both time-domain and frequency-domain performance, achieve global optimal configuration of equalization parameters, and improve signal equalization and restoration quality. High-performance equalization is achieved on resource-constrained devices, which improves the transmission reliability of Chirp signals in complex acoustic environments, reduces the bit error rate, and reduces computing resource usage, providing an effective solution for improving the performance of near-field communication systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart of a method for equalizing a Chirp modulation signal of an audio device in the present invention. DETAILED DESCRIPTION

[0019] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that these embodiments are discussed solely to enable those skilled in the art to better understand and implement the subject matter described herein, and that the functions and arrangements of the elements discussed may be varied without departing from the scope of this specification. Various examples may omit, substitute, or add various processes or components as needed. Furthermore, features described in some examples may be combined in other examples.

[0020] At least one embodiment of the present invention discloses a method for equalizing a Chirp modulation signal of an audio device, such as Figure 1 Shown, including: Step 1: Analyze the non-stationary characteristics of the Chirp modulated signal using the information geometry mapping algorithm to generate a non-stationary complexity map; This step is the basis for achieving accurate equalization of Chirp signals, and accurately quantifies their non-stationary characteristics by mapping the signal into a geometric space. It specifically includes the following sub-steps: Step 1.1: Receive the original Chirp modulation signal collected by the audio device and apply short-time Fourier transform to convert the Chirp signal into time-frequency representation; the calculation formula is: ; in is the time-frequency representation of the signal, is the original time domain signal, Represents the time window function, usually Hanning window or Hamming window is selected; is the integral variable; is the imaginary unit, ; Indicates time, Indicates frequency.

[0021] Step 1.2: construct a signal probability distribution model based on the time-frequency representation and map the signal onto the Riemannian manifold; For time The signal at , extract its power spectral density: ; in Indicates time Frequency The power spectral density of represents the square of the modulus of the time-frequency representation; Indicates the a point in time.

[0022] And normalized to a probability distribution: ; in represents the normalized probability distribution; For all frequencies The power spectral density of is integrated as the normalization factor.

[0023] Step 1.3, construct the Fisher information matrix , for adjacent time points on the statistical manifold and The corresponding probability distribution and , calculate its Fisher information matrix elements: ; in and Respectively represent the description time points and Parameters of the signal probability distribution at ; is the frequency The marginal probability distribution of represents the Fisher information matrix Rank Column elements; For all frequencies 's points; Represents a probability distribution Parameters The partial derivative of Represents a probability distribution Parameters The partial derivative of .

[0024] Step 1.4, based on the Fisher information matrix, calculate the geodesic distance between two distribution points on the statistical manifold to quantify the degree of signal non-stationarity: ; in Indicates a time point and The geodesic distance between the probability distributions at Represents connection distribution and The path, and Respectively indicate time points and The probability distribution at is a path parameter, is the derivative of the path, Is a point on the path The Fisher information matrix at , Represents matrix transpose; Indicates that among all possible paths Take the minimum value; represents the geodesic distance function; Indicates path parameters integral from 0 to 1; Represents the square root operation.

[0025] Step 1.5: Based on the geodesic distance calculation results, construct the non-stationary complexity map of the Chirp signal , which reflects the degree of non-stationarity of the signal at each time point: ; in Indicates a time point The non-stationary complexity value at is a normalization factor used to normalize the complexity map to a specific range; is the size of the time window considered, which indicates the number of adjacent time points included in the calculation; is a weighting function used to balance the impact of short-term and long-term non-stationarity; is the index variable within the window; Relative to Offset A time point of a time unit; Indicates all indexes in the window From 1 to The sum of .

[0026] Through the above sub-steps, the non-stationary characteristics of the Chirp signal on the statistical manifold are accurately quantified, and the generated non-stationary complexity map This provides a basis for the adaptive adjustment of the subsequent equilibrium strategy. The output of step 1 is a non-stationary complexity map , this mapping is a time function that quantifies the non-stationarity of the chirp signal at each time point. The numerical range of this mapping is usually between [0,1], and the larger the value, the stronger the non-stationarity of the signal at that time point.

[0027] Step 2, applying a multi-scale entropy analysis algorithm to process the Chirp modulated signal to generate a multi-scale complexity spectrum; This step quantifies the complexity of the Chirp signal at different time scales through a multi-scale entropy analysis method, thereby capturing the changes in the nonlinear characteristics of the signal in different time windows. Step 2 forms a complementary relationship with step 1, and together they construct a comprehensive characterization of the Chirp signal characteristics: step 1 focuses on the analysis of the non-stationary characteristics of the signal in the time-frequency domain, while step 2 focuses on revealing the complexity characteristics of the signal at different time scales. These two characterization methods characterize the characteristics of the Chirp signal from different angles, and their combination can form a more comprehensive description of the signal characteristics. Step 2 receives the non-stationary complexity map of step 1. Finally, it is combined with the results of multi-scale entropy analysis to form a richer complexity representation. The specific steps include the following: Step 2.1, the original Chirp signal Segment processing is performed and divided into pieces of length Multiple data segments: ; in 、 、 Represents the first, second, and data segments; is the number of segments, Indicates the length of each data segment.

[0028] Step 2.2: Apply coarse-graining processing to each data segment to generate multiple different scale factors The coarse-grained time series under For the scale factor The coarse-graining process is to The data points are averaged to form a new time series: ; in represents the index of the coarse-grained sequence, ; represents the scale factor, which determines the degree of coarse-graining; Indicates the first sampling points; Indicates rounding down to calculate the length of the coarse-grained sequence; Expressing arrive All indexes of sum; represents the division operation for finding the average value; The scale factor is The first sequence after coarse-graining elements.

[0029] Step 2.3, for each scale Coarse-grained time series under Calculate the sample entropy (SampEn). First determine the template length and similarity threshold , and then calculate the sample entropy: ; in Indicates that the length of the sequence is found The probability of matching the template, Indicates that the length is found The probability of matching template; is the template length; is the similarity threshold; represents the natural logarithm function; Represents the ratio of two probabilities; represents the sample entropy function.

[0030] Step 2.4, calculate the multi-scale entropy index and construct the entropy spectrum that reflects the complexity change of the signal at different time scales: ; For a range of scale factors: , get the corresponding multi-scale entropy value sequence, and form the entropy spectrum ; in Indicates the maximum scale factor, which is usually determined according to the signal length and analysis requirements; represents the multi-scale entropy function; represents the sample entropy function, represents the coarse-grained sequence.

[0031] Step 2.5, based on entropy spectrum Identify the characteristic scale and key change points of Chirp signals; characteristic scale Defined as a local extreme point or change point in the entropy spectrum: ; in is a pre-set threshold used to determine whether the entropy spectrum slope changes; Represents the entropy spectrum Scale Factor The derivative of , which reflects the rate at which entropy changes with scale; represents a set of characteristic scales; Indicates absolute value operation; Indicates that the conditions are met A collection of values.

[0032] Step 2.6, combined with non-stationary complexity mapping and the multiscale entropy spectrum in step 2 , construct a comprehensive complexity feature space and generate a two-dimensional complexity spectrum : ; in Indicates time and scale The comprehensive complexity value under and are the weighting coefficients of non-stationary complexity and multiscale entropy respectively; is the time scale correlation function, used to describe a specific time point With a specific scale the degree of correlation between them; Indicates a time point The non-stationary complexity value at reflects the non-stationary degree of the signal at that time point; The scale factor is The multi-scale entropy value at , reflects the complexity of the signal at that scale.

[0033] Through the above sub-steps, the complexity analysis of the Chirp signal at multiple time scales is completed. The generated multi-scale complexity spectrum can effectively capture the nonlinear characteristic changes of the signal in different time windows and frequency bands, providing a more comprehensive signal characteristic representation for the subsequent adaptive equalization strategy. The output of step 2 includes the multi-scale entropy spectrum and the two-dimensional complexity spectrum . The multi-scale entropy spectrum is a scale factor function, reflecting the complexity distribution of the signal at different time scales; two-dimensional complexity spectrum It is a two-dimensional function of time and scale, which combines the non-stationary characteristics and multi-scale complexity of the signal.

[0034] Step 3: Use fractal analysis algorithm to extract the long-range correlation characteristics of the Chirp modulation signal and output the Hurst exponent feature map; This step quantifies the long-range correlation and self-similarity of the Chirp signal through fractal analysis, providing a characterization of the long-term dependence of the time series for the equalization algorithm. Step 3 and the previous two steps constitute a complete system for the analysis of Chirp signal characteristics: the non-stationary complexity mapping in step 1 provides the local variation characteristics of the signal in the time-frequency domain, the multi-scale entropy analysis in step 2 provides the complexity distribution of the signal at different time scales, and step 3 supplements the correlation structure characteristics of the signal's long time series. It specifically includes the following sub-steps: Step 3.1: Apply Discrete Wavelet Transform (DWT) to the original Chirp signal for multi-resolution decomposition to obtain wavelet coefficients of different scales; The calculation formula is: ; in is the wavelet basis function, expressed as: ; in Represents the scale index, which determines the degree of expansion and contraction of the wavelet; represents the translation parameter, which determines the position of the wavelet; represents the mother wavelet function; Indicated on scale and location The wavelet coefficients at ; Represents the original Chirp signal; Indicates time Integral from negative infinity to positive infinity; represents the normalization factor of the wavelet function; Represents the scaling factor of the wavelet function; Represents the translation of the wavelet function.

[0035] Step 3.2, calculate each scale The variance of the wavelet coefficients : ; in It's a scale The number of lower wavelet coefficients; Represents the square modulus of the wavelet coefficient, i.e., energy; Representation scale The variance of the wavelet coefficients at this scale reflects the energy distribution of the signal at this scale; represents the division operation for finding the average value; Indicates the number of All translation parameters of sum; Represents the square modulo operation.

[0036] Step 3.3: Analyze the variance of wavelet coefficients in the double logarithmic coordinate system and scale The relationship between , fitting linear model: ; in is the slope of the fitted line, reflecting the scale characteristics of the signal; is a constant term, which represents the intercept of the fitted straight line; represents the logarithmic function with base 2; Representation scale The variance of the lower wavelet coefficients; Represents the scale index.

[0037] Step 3.4, based on the wavelet variance slope Calculate the Hausdorff dimension of a signal and Fractal Index ; Expressed as: ; ; in The fractal dimension of the signal describes its complexity, and its value range is usually between 1 and 2; Represents the power spectral density slope, which is related to the long-range correlation of the signal. The larger the value, the stronger the long-range correlation; represents the wavelet variance slope.

[0038] Step 3.5, use the rescaled range analysis (R / S analysis) method to estimate the local Hurst exponent. First, divide the signal into multiple segments of length For each subsequence of Calculate the average: ; in Indicates the first data points, represents the subsequence length, represents the division operation for finding the average value, Indicates the number of Sum all the indices of represents the average value of the subsequence; Calculate the cumulative deviation series: ; in Before The cumulative deviation of the points from the mean, Indicates the number of Sum all the indices of Indicates the The difference between a data point and the mean; Calculate the range: ; in and Respectively represent the maximum and minimum values, Indicates the range of the cumulative deviation series; Calculate the standard deviation: ; in represents the square root operation, represents the division operation for finding the average value, Indicates the number of Sum all the indices of Indicates the The square of the difference between the data point and the mean, represents the standard deviation of the subsequence; Compute the rescale range statistic: ; in Equivalent to , indicating extremely poor, Represents standard deviation.

[0039] Step 3.6, for different lengths Repeat step 3.5 for the subsequence to obtain a series of data points; Fit a linear model in log-log coordinates: ; in This is the Hurst index, and its value range is usually between 0 and 1; is a constant term, which represents the intercept of the fitted straight line; represents the natural logarithm function; represents the rescaled range statistic; Indicates the subsequence length; Represents the Hurst exponent, which is the slope of the fitted line.

[0040] Step 3.7, calculate the local Hurst exponent based on the sliding window and construct the Hurst exponent feature map ; For time point , with windows around it Apply R / S analysis to obtain the local Hurst index ; in is the window width, which indicates the number of data points used to calculate the local Hurst exponent; Indicates a specific point in time; Indicates the starting position of the window; Indicates the end position of the window; Indicates a time point The local Hurst exponent at .

[0041] Step 3.8, according to Hurst index feature mapping , the signal is divided into the following three areas; Anti-persistence zone: , indicating that the signal has high-frequency fluctuation characteristics in this area; Random walk region: , indicating that the signal is approximately a random process in this region; Continuous area: , indicating that the signal has long-range correlation and trend characteristics in this area.

[0042] Through the above sub-steps, the long-range correlation and self-similarity characteristics of the Chirp signal are accurately quantified. The generated Hurst exponent feature map provides a characterization of the long-term dependence of the signal for the equalization algorithm, which helps to maintain the consistency of the equalization effect in a long time window. The output of step 3 is the Hurst exponent feature map , the mapping is a time function, and its value range is usually between [0,1], which is used to quantify the long-range correlation and self-similarity characteristics of the signal at each time point. The closer the value is to 1, the stronger the persistence and long-range correlation of the signal; the closer it is to 0, the stronger the anti-persistence and high-frequency fluctuation characteristics of the signal. This feature map will be combined with the non-stationary complexity map in step 1. Together with the multi-scale complexity spectrum in step 2, it forms the basis of the complexity-aware adaptive equalization system in step 4, and plays an important role in signal processing over long time windows.

[0043] Step 4: Based on the non-stationary complexity map, multi-scale complexity spectrum, and Hurst exponent feature map, a complexity-aware adaptive equalization system is constructed to achieve adaptive equalization processing of the Chirp signal. This step builds a complexity-aware adaptive equalization system based on the feature map data obtained in the first three steps to achieve differentiated equalization processing of Chirp signal complexity difference areas. Specifically, step 4 converts the non-stationary complexity map output in step 1 into , the multi-scale complexity spectrum output from step 2 and the two-dimensional complexity spectrum , and the Hurst exponential feature map output in step 3 As input, these signal features of different dimensions are integrated to form a comprehensive representation of signal complexity. The adaptive equalization system consists of four functional modules: complexity analysis unit, parameter matching unit, equalization processing unit, and performance evaluation unit. Each module is connected through signal and parameter interfaces to form a closed-loop control structure. Specifically, it includes the following sub-steps: Step 4.1: Based on the feature map obtained in the previous step, the complexity analysis unit constructs a comprehensive evaluation index of signal complexity. , which comprehensively considers the non-stationarity, multi-scale complexity and long-range correlation of the signal: ; in Indicates time Comprehensive evaluation index of signal complexity at ; 、 and They are non-stationary complexity weight, multi-scale complexity weight and long-range correlation weight respectively; is a non-stationary complexity map, which means that the signal The degree of non-stationarity at is the scale factor; is the selected set of scale factors; For scale The entropy value under is the time scale correlation function, characterizing a specific scale In time the importance of For time Hurst index at is the Hurst exponential mapping function, which is used to quantify the contribution of long-range correlation to complexity; For all selected scale factors Summation.

[0044] The complexity analysis unit is implemented as a cascade structure consisting of a feature extractor, a weight adjuster, and a complexity evaluator, and the processing is accelerated through parallel computing.

[0045] Step 4.2: Comprehensive evaluation index based on complexity The complexity analysis unit adaptively segments the Chirp signal into multiple regions of different complexity: ; in 、 、 Represents the first, second, and Segmented areas, is the total number of segments, represents the original Chirp signal, is a segmentation algorithm, is the segmentation threshold, according to The change range is determined adaptively.

[0046] In practical applications, the segmentation algorithm adopts a method based on the combination of intersection detection and dynamic programming to ensure the accuracy and smoothness of the segmentation boundaries.

[0047] Step 4.3: The parameter matching unit establishes the equalization parameter library , including the following equalization parameter configurations for different complexity areas: ; in represents the equalization parameter library, 、 、 Represents the first, second, and Equalization parameter configuration, Indicates the total number of parameter configurations.

[0048] The parameter library adopts a hierarchical storage structure, and classifies and indexes according to the complexity of parameter configuration and applicable scenarios to improve matching efficiency.

[0049] Step 4.4, for each signal segment , the parameter matching unit selects the optimal equalization parameter configuration according to its complexity characteristics: ; in Expressed as The best parameter configuration for segment selection, Indicates signal segment Complexity index and parameter configuration Corresponding standard complexity index The distance between Indicates taking the parameter configuration that minimizes the distance; Indicates parameter configuration Is the parameter library Elements in Indicates the The complexity index of each segment; Representation and parameter configuration The corresponding standard complexity index; Represents a distance function that measures the difference between two complexity metrics.

[0050] In the specific application scenarios of audio equipment, when processing chirp signals in music transmission, the parameter matching unit will give priority to parameter configurations with higher frequency response flatness; when processing chirp signals in voice commands, it will give priority to parameter configurations with higher time resolution.

[0051] In step 4.5, the equalization processing unit constructs a complexity adaptive equalizer and applies differentiated equalization strategies to different complexity regions. The equalization processing unit contains multiple filter structures, parameter adjustment modules, and processing pipelines, and can dynamically reconstruct the processing path according to the complexity characteristics of the input signal: For high complexity areas ( ): Apply a high-order filter structure, typically a 64-order FIR filter or an 8-order IIR filter. Use a smaller adaptive step size, typically 0.01-0.05. Increase the number of iterations, typically 2-3 times the standard configuration. Apply regularization constraints, using and Combination form of norms; For medium complexity areas ( ): Apply an intermediate filter structure, typically a 32-order FIR filter or a 4-order IIR filter. Use a balanced adaptive step size, typically 0.05-0.1. Use a standard number of iterations, typically 10-15. Apply a mild regularization constraint, primarily using norm; For low complexity regions ( ): Apply a low-order filter structure, typically a 16th-order FIR filter or a 2nd-order IIR filter. Use a larger adaptive step size, typically 0.1-0.2, to reduce the number of iterations, usually half the standard configuration. Simplify the calculation process and omit regularization constraints. in represents the threshold value of high complexity area, Threshold indicating low complexity regions.

[0052] In step 4.6, the equalization processing unit applies a smooth transition strategy between adjacent signal areas to avoid abrupt changes in the equalization effect at the area boundaries: ; in Indicates the parameter configuration of the transition region, is a smooth transition function that smoothly transitions from 1 to 0 at the region boundary. and Respectively represent adjacent and Optimal parameter configuration for each signal area; The complement of the smooth transition function; Indicates time Transition parameter configuration at .

[0053] The transition area parameter configuration achieves a smooth transition at the boundaries of adjacent signal areas, avoids sudden changes in the equalization effect, and improves the continuity and stability of the overall equalization processing.

[0054] In actual implementation, the smooth transition function uses a cosine smoothing window, and the length of the transition interval is adaptively adjusted according to the rate of change of the signal, with a typical value of 10-30ms.

[0055] Step 4.7: The performance evaluation unit evaluates the quality of the equalized signal and calculates the equalization performance index. : ; in Indicates time The equilibrium performance index at and are the signal-to-noise ratios before and after equalization, It is a spectral distortion measure, which indicates the degree of difference between the spectrum before and after equalization; represents the signal-to-noise ratio improvement factor; represents the spectral fidelity factor.

[0056] The performance evaluation unit uses a sliding window analysis method to comprehensively evaluate performance indicators at three scales: short-term, medium-term, and long-term, providing accurate feedback for parameter updates.

[0057] Step 4.8, the performance evaluation unit is based on the balance performance index , the collaborative parameter matching unit balances the parameter library Perform online updates to optimize balancing performance: ; in Indicates the updated parameter library, Indicates the current parameter library, is the parameter update function, is the gradient of the performance index with respect to the parameter, indicating the sensitivity of the performance index to the change of the parameter; Indicates time Parameter library at Indicates time Parameter library at Indicates that the current parameter library , performance indicators and gradient is the update function of the input; Indicates the parameters Operator for finding the gradient.

[0058] Parameter updates use a stochastic gradient descent method based on memory cache, retaining historical update data to avoid local oscillations, and continuously optimizing the parameter library during actual use of audio devices to adapt to the usage environments of different users.

[0059] Through the above sub-steps, the complexity-aware adaptive equalization processing of the Chirp signal is realized, and differentiated equalization strategies are applied to different signal complexity areas, which not only ensures the equalization accuracy of high-complexity areas, but also improves the computational efficiency of low-complexity areas. In audio communication scenarios, the system can effectively process various types of Chirp modulation signals, including linear frequency modulation, exponential frequency modulation and multi-segment frequency modulation signals, and improve signal recognition rate and transmission quality. The output of step 4 includes the equalized Chirp signal, the comprehensive evaluation index of signal complexity, and the signal complexity index. , balance parameter configuration and balanced performance indicators The equalized Chirp signal is the main output of this step and is directly used for subsequent information decoding; the comprehensive evaluation index of signal complexity is It is a time function generated by integrating the feature maps of the first three steps, which fully characterizes the complexity of the signal at each time point; equalization parameter configuration It is the optimal parameter setting for different signal segments; equalization performance index It quantifies the changes in the equilibrium effect in the time dimension.

[0060] Step 5: Based on the adaptive equalization processing results, a time-frequency joint optimization algorithm is constructed to achieve global optimization and adaptive adjustment of equalization parameters; This step further improves the equalization parameter configuration obtained in step 4 by constructing a time-frequency joint optimization algorithm, and promotes the local optimal solution to the global optimal solution. and balanced performance indicators Based on the signal features extracted from step 1 to step 3 (non-stationary complexity mapping , multiscale complexity spectrum and Hurst exponential characteristic map ) to guide the optimization process. The time-frequency joint optimization algorithm considers the performance of equalization in both the time and frequency domains, achieving global optimization and adaptive adjustment of equalization parameters, ensuring the consistency and stability of equalization performance throughout the entire signal cycle. This algorithm adopts a multi-objective optimization framework and consists of a time-domain optimization module, a frequency-domain optimization module, a time-frequency joint optimization module, and a parameter update module. These modules work together to achieve global optimization. Specifically, it includes the following sub-steps: Step 5.1, the time-frequency joint optimization module constructs the time-frequency joint objective function; This function comprehensively considers the performance of equalization in the time domain and frequency domain: ; in represents the time-frequency joint objective function, which is the overall goal of equalization parameter optimization; is the time domain objective function, which evaluates the performance of equalization in the time domain; is the frequency domain objective function, which evaluates the performance of equalization in the frequency domain; is the time-frequency joint objective function, which evaluates the performance of equalization in the time-frequency joint domain; 、 and Represent the weight coefficients of the time domain objective function, frequency domain objective function and time-frequency joint objective function respectively; represents the time variable, represents a frequency variable, represents a set of equalization parameters; In practical applications, the weight coefficient is dynamically adjusted according to the specific type of the Chirp signal.

[0061] For example, for a linear frequency modulation Chirp signal, set 、 、 ; For the logarithmic frequency modulated Chirp signal, set 、 、 , to adapt to the characteristics of different types of signals.

[0062] Step 5.2: The time domain optimization module constructs the time domain objective function based on the information geometry metric : ; in represents the time domain objective function; is the equalized signal at time The probability distribution at ; is the ideal probability distribution of the target signal; is the geodesic distance, which is used to measure the difference between two probability distributions; is the time domain weighting function, according to the time point Importance adjustment weights; is the total number of time sampling points; is the index of the time sampling point; Indicates that all time sampling points from 1 to sum; Time domain weighting function An adaptive weighting strategy is used to assign higher weights to key regions of the chirp signal (such as frequency transition points or start and end points) and lower weights to stable regions. In its implementation, the module consists of three components: a distribution estimator, a distance calculator, and a weighted accumulator. It optimizes computational efficiency through a fast approximation algorithm.

[0063] Step 5.3: The frequency domain optimization module constructs the frequency domain objective function based on the spectral distance. : ; in represents the frequency domain objective function; is the power spectral density of the equalized signal; is the ideal power spectral density of the target signal; is the frequency domain weighting function, according to the frequency Importance adjustment weights; Represents the square difference between the power spectral density of the equalized signal and the target signal; Indicates integration of all frequency points; Indicates the square operation; Frequency domain weighting function According to the frequency characteristics of the Chirp signal, a higher weight is assigned to the frequency band where the signal energy is concentrated, and a lower weight is assigned to the low energy area. In practical applications, this function adopts the band-limited weighting form: ; in represents the frequency domain weighting function, is the center frequency, is the bandwidth parameter, is the normalization coefficient, represents the natural exponential function; For different types of audio equipment, and The parameters are adjusted according to the frequency response characteristics of the device.

[0064] For example, in the smart speaker application scenario, for the linear frequency modulation Chirp signal, the typical setting is , For multi-segment frequency hopping Chirp signals, a segment weighting strategy is adopted, setting an independent and Parameters are adjusted to match the frequency hopping characteristics of the signal. This weighting strategy ensures that the equalization algorithm achieves the most accurate optimization effect in the frequency band where the signal energy is most concentrated, while reducing the ineffective investment of computing resources in low-energy areas.

[0065] In step 5.4, the time-frequency joint optimization module constructs the time-frequency joint objective function based on the multi-scale entropy difference: ; in represents the time-frequency joint objective function; is the equalized signal at time and scale Multiscale entropy under ; is the ideal multi-scale entropy of the target signal; is the time-frequency joint weighting function, according to time and scale Importance adjustment weights; represents the scale factor, represents the selected set of scale factors; represents the sum of all selected scale factors.

[0066] This module uses a multiresolution analysis framework to simultaneously consider differences in the signal's entropy characteristics at different time scales, effectively capturing the nonlinear characteristics of chirp signals. In specific audio applications, such as indoor positioning systems based on chirp signals, this function can more accurately match the time-frequency characteristics changes caused by multipath effects in complex reverberant environments.

[0067] In step 5.5, the parameter update module uses the gradient descent algorithm to optimize the time-frequency joint objective function and iteratively update the equalization parameters: ; in Indicates the The equalization parameters of the iteration; For the The equalization parameters of the iteration; is the learning rate, which controls the step size of parameter update; is the objective function with respect to the parameter The gradient of , which indicates the direction of change of the objective function in the parameter space; An index indicating the number of iterations; Indicates the parameters Operator for finding the gradient.

[0068] Gradient calculations employ a hybrid analytical and numerical approach, using analytical gradients for primary parameters and numerical approximations for secondary parameters, balancing accuracy and efficiency. In embedded audio devices, a combination of fixed-point arithmetic and lookup tables is employed to reduce computational complexity and enable real-time processing.

[0069] In step 5.6, the parameter update module introduces a signal complexity adaptive gradient update strategy and adopts differentiated parameter update methods for different complexity regions: ; in Indicates time Adaptive learning rate at ; Is the basic learning rate, which indicates the default learning rate value; is the complexity adjustment function, according to the signal complexity Dynamically adjust the learning rate; Indicates time The comprehensive evaluation index of signal complexity at .

[0070] Complexity adjustment function Using piecewise linear or exponential decay, the learning rate is reduced to increase stability when signal complexity is high, and increased to accelerate convergence when signal complexity is low. In actual audio device operating scenarios, when the device is operating in a noisy environment, the system automatically reduces the learning rate to prevent parameter oscillation; in a quiet environment, the learning rate is increased to accelerate adaptation.

[0071] Step 5.7: Time-frequency joint optimization module constructs long-range correlation-aware regularization term , enhance the equalization's ability to handle long-term signal dependencies: ; in A regularization term that represents the perception of long-range dependencies; and are the equalized signal and the target signal at time The local Hurst exponent at is the regularization coefficient, which controls the weight of the regularization term; Indicates the time difference between the equalized signal and the target signal The squared difference of the local Hurst exponent at ; Indicates the sum of all time sampling points is the total number of time sampling points The index of the time sampling point.

[0072] In different audio application scenarios, adjust the long-range correlation characteristics of the signal For example, when processing a continuous Chirp signal (such as a music watermark), setting a larger When processing short-time Chirp signals (such as control instructions), set a smaller value. This regularization term uses a fast Hurst exponent estimation algorithm in actual implementation to reduce computational overhead.

[0073] In step 5.8, the time-frequency joint optimization module incorporates the long-range correlation regularization term into the objective function to obtain the final optimization goal: ; in Represents the final optimization objective function; represents the time-frequency joint objective function; A regularization term that represents the awareness of long-range dependencies.

[0074] This objective function adopts a hierarchical calculation strategy in the algorithm implementation, first optimizing the basic parameters at the local scale, and then optimizing the long-range correlation parameters at the global scale, reducing the dimension of the parameter space and improving the optimization efficiency.

[0075] Step 5.9: The parameter update module implements global optimization and adaptive adjustment of the equalization parameters based on the above optimization process to obtain the optimal equalization parameter configuration. : ; in represents the optimal equilibrium parameter configuration; Represents the search for the objective function; Parameter configuration to obtain the minimum value; Represents the final optimization objective function.

[0076] Global optimization employs a hybrid strategy combining coarse-grained grid search and fine-grained gradient descent to avoid being trapped in local optima. When deployed on audio devices, the algorithm automatically adjusts the optimization precision based on the device's computing power, performing a full optimization on high-performance devices and a simplified version on low-power devices, achieving a balanced balance between resources and performance.

[0077] Through the above sub-steps, the time-frequency joint optimization of the equalization parameters is achieved, while considering the performance of the signal in the time domain, frequency domain and time-frequency joint domain, and especially enhancing the processing ability of the long-range correlation of the signal to ensure the consistency and stability of the equalization performance throughout the signal cycle. In actual audio communication systems, this algorithm can improve the equalization effect of the Chirp signal in different acoustic environments, and improve the reliability and anti-interference ability of signal transmission. The final output of step 5 is the global optimal equalization parameter configuration and the optimized Chirp signal. Compared with the local optimal parameters in step 4, The comprehensive performance of the time domain, frequency domain, and joint time-frequency domain is considered, with special attention paid to the influence of long-range correlation, and the consistency of equalization performance can be maintained throughout the entire signal cycle. Through the five main steps of the present invention, from signal feature extraction to complexity-aware equalization to global parameter optimization, a complete chirp modulated signal equalization processing flow is constructed. Each step is closely related and progressive, ultimately achieving high-quality equalization and restoration of chirp signals in complex acoustic environments.

[0078] An equalization system for a Chirp modulated signal of an audio device, used to perform the above-mentioned equalization method for a Chirp modulated signal of an audio device, comprising: Information geometry mapping module, used to analyze the non-stationary characteristics of Chirp modulated signals and generate non-stationary complexity mapping; Multi-scale entropy analysis module, used to process Chirp modulated signals and generate multi-scale complexity spectra; Fractal analysis module, used to extract the long-range correlation characteristics of Chirp modulation signals and output Hurst exponent feature map; Complexity-aware adaptive equalization module, used to implement adaptive equalization processing of Chirp signals; The time-frequency joint optimization module is used to achieve global optimization and adaptive adjustment of equalization parameters.

[0079] Here, the present invention provides an implementation example: The application scenario of this embodiment is a near-field device communication system in a smart speaker. In this scenario, the smart speaker needs to communicate quickly and reliably with surrounding smart home devices (such as smart lamps, smart thermostats, and smart TVs). The system uses Chirp modulation signals as the communication carrier, transmitting control commands and status information through an audio channel. The actual application environment has the following characteristics: Complex acoustic environment: The device is usually placed in an indoor environment with multiple reflective surfaces, such as a living room or bedroom, where severe reverberation, echo, and multipath effects may occur.

[0080] Background noise interference: There are various background noises in the home environment, such as the sound of TV, the sound of running home appliances, and human voices, which makes the signal-to-noise ratio unstable.

[0081] Multi-device collaboration: Smart speakers need to communicate with multiple smart devices at the same time, requiring the signal processing system to be able to process multiple Chirp signals simultaneously.

[0082] Low interference requirement: Chirp signals need to be transmitted in a manner that is almost imperceptible to the human ear. The signal energy must be low and have strong anti-interference capabilities.

[0083] To meet the requirements of these application scenarios, the system uses the chirp modulation signal equalization method described in this implementation to achieve high-quality and highly reliable audio channel communication. In actual deployment, this method is implemented as an embedded software module running on the smart speaker's main processor, seamlessly integrating with the device's audio acquisition and processing systems.

[0084] In practical application in the smart speaker communication system, the implementation process of this embodiment is as follows: In the experimental environment, the following hardware equipment and parameters were configured: Smart speaker: uses a four-microphone array, a sampling rate of 48kHz, and 16-bit quantization. Test environment: a typical home living room (about 25 square meters), with a reverberation time RT60 = 350ms. Background noise level: an average of 40dB SPL, with a peak value of up to 65dB SPL. Chirp signal configuration: frequency range 18-20kHz, duration 50-200ms, and signal power controlled at a level that is almost imperceptible to the human ear (about 30dB SPL).

[0085] Three typical Chirp signals were collected as test sets in the experiment: Linear frequency modulation Chirp: frequency changes linearly; Logarithmic frequency modulation Chirp: frequency changes logarithmically; Multi-segment frequency hopping Chirp: contains multiple frequency hopping points; For each signal, 100 sets of samples were collected under different environmental conditions (quiet, low noise, and high noise), constituting a total of 900 sets of test data sets.

[0086] For the linear frequency modulation Chirp signal, the information geometry mapping algorithm is applied to analyze its non-stationary characteristics, and the actual effect of the non-stationary complexity mapping is shown in Table 1: Table 1: Non-stationary complexity mapping results of linear frequency modulation Chirp signals under different environments

[0087] Table 1 shows that the signal's non-stationary complexity varies significantly across different environments. High-noise environments generally exhibit higher non-stationary complexity than quiet environments, reflecting the impact of noise on signal characteristics. Furthermore, the signal exhibits higher non-stationary complexity in the intermediate time period (40-80ms), which is related to the frequency change rate of the chirp signal.

[0088] The collected Chirp signals are subjected to multi-scale entropy analysis and fractal feature extraction. The main characteristic parameters of different types of Chirp signals are shown in Table 2: Table 2: Multi-scale entropy and fractal characteristic parameters of different types of Chirp signals

[0089] The results in Table 2 show that the entropy of multi-segment FH chirps at all scales is higher than that of linear and logarithmic FH chirps, reflecting their higher structural complexity. The Hurst exponents of linear and logarithmic FH chirps are generally greater than 0.5, indicating a clear persistence characteristic, while the Hurst exponent of multi-segment FH chirps is close to 0.5, indicating characteristics closer to a random process.

[0090] Based on the above analysis results, the system builds a complexity-aware adaptive equalization system. The equalization parameters configured for different complexity areas are shown in Table 3: Table 3: Complexity-aware adaptive equalization parameter configuration

[0091] In the actual processing process, the system calculates the complexity index in real time , dynamically switch the above parameter configurations to achieve adaptive equalization processing for areas of different complexity.

[0092] The time-frequency joint optimization algorithm is applied to adjust the equalization parameters, and the objective function weights of different types of Chirp signals are configured as shown in Table 4: Table 4: Time-frequency joint optimization weight configuration for different types of Chirp signals

[0093] The weight configurations in Table 4 reflect the differences in the characteristics of different types of chirp signals: logarithmic frequency modulation chirp focuses more on time domain optimization; multi-band frequency hopping chirp focuses more on frequency domain optimization; at the same time, all signal types maintain a high joint time-frequency weight to ensure the consistency of equalization in the time-frequency domain.

[0094] To verify the equalization accuracy of this implementation, a comparative analysis was conducted between the traditional LMS equalizer, the traditional RLS equalizer, and this implementation. The evaluation metrics were the equalized signal distortion rate (SER) and signal recognition accuracy. The experiments were conducted under different signal-to-noise ratio (SNR) conditions, and the results are shown in Table 5. Table 5: Performance comparison of different equalization methods

[0095] Table 5 shows that this implementation outperforms traditional LMS and RLS equalizers across all SNR conditions. In particular, under low SNR conditions (5dB and 0dB), this implementation maintains low signal distortion and high recognition accuracy, demonstrating strong adaptability to complex noise environments. On average, this implementation reduces signal distortion by approximately 50% compared to traditional methods.

[0096] To verify the computational resource optimization effect of this embodiment, the computational complexity and memory usage of a traditional fixed parameter equalizer and the complexity adaptive equalizer of this embodiment were measured when processing the same signal. The results are shown in Table 6: Table 6: Comparison of computing resource utilization efficiency

[0097] The complexity-adaptive equalization implemented in this implementation reduces computational complexity and memory usage while maintaining superior equalization quality (lower SER). The computational resource efficiency improvement is most pronounced for linear frequency modulation chirps, reaching 41.1%. Even for the most complex multi-frequency hopping chirps, the resource efficiency improvement reaches 31.0%. This demonstrates the computational efficiency optimization achieved by this implementation through complexity analysis and adaptive resource allocation.

[0098] To verify the adaptability of this embodiment in different environments, the stability of the equalization performance was tested in various acoustic environments. The results are shown in Table 7: Table 7: Balanced performance stability under different environments

[0099] The results in Table 7 show that as environmental complexity increases (e.g., reverberation time and background noise), the performance of traditional equalization methods rapidly degrades, while this implementation maintains a good equalization effect. In the most complex and noisy environments, this implementation achieves a 59.9% improvement in stability over traditional methods, demonstrating its excellent environmental adaptability.

[0100] The above practical application examples and technical effect verification have proved that this implementation has advantages in the near-field device communication application scenario of smart speakers, can effectively solve the key technical problems in the equalization processing of Chirp modulation signals, and achieve high-quality and high-stability signal processing effects.

[0101] The above describes an embodiment of the present invention, but this embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Ordinary technicians in this field can also make more forms of equivalent embodiments based on the inspiration of this embodiment, all of which are protected by this embodiment.

Claims

1. A method for equalizing a Chirp modulation signal of an audio device, characterized in that: include: The information geometry mapping algorithm is used to analyze the non-stationary characteristics of Chirp modulated signals and generate non-stationary complexity mapping; Apply multi-scale entropy analysis algorithm to process Chirp modulated signal and generate multi-scale complexity spectrum; The fractal analysis algorithm is used to extract the long-range correlation characteristics of the Chirp modulation signal and output the Hurst exponent feature map; Based on non-stationary complexity mapping, multi-scale complexity spectrum, and Hurst exponent feature mapping, a complexity-aware adaptive equalization system is constructed to achieve adaptive equalization processing of Chirp signals. Based on the adaptive equalization processing results, a time-frequency joint optimization algorithm is constructed to achieve global optimization and adaptive adjustment of equalization parameters.

2. The method for equalizing a Chirp modulation signal of an audio device according to claim 1, wherein: The step of analyzing the non-stationary characteristics of the Chirp modulation signal by using the information geometry mapping algorithm comprises: Receive the original Chirp modulated signal; Apply short-time Fourier transform to convert the Chirp signal into time-frequency representation; Construct a signal probability distribution model based on time-frequency representation and map the signal onto a Riemannian manifold; Construct the Fisher information matrix and calculate the Fisher information matrix elements corresponding to the probability distribution of adjacent time points on the manifold; Based on the Fisher information matrix, the geodesic distance between two distribution points on the statistical manifold is calculated to quantify the degree of signal non-stationarity; Based on the calculation results of geodesic distance, a non-stationary complexity map of Chirp signal is constructed.

3. The method for equalizing a Chirp modulation signal of an audio device according to claim 1, wherein: The step of applying the multi-scale entropy analysis algorithm to process the Chirp modulation signal comprises: Segment the original Chirp signal and divide it into multiple data segments; Apply coarse-graining processing to each data segment to generate multiple different scale factors The coarse-grained time series under ; Calculate the sample entropy of the coarse-grained time series at each scale; Calculate multi-scale entropy indicators and construct entropy spectra that reflect the complexity changes of signals at different time scales; Identify the characteristic scale and key change points of Chirp signals based on entropy spectrum; Combining non-stationary complexity mapping and multi-scale entropy spectrum, a comprehensive complexity feature space is constructed to generate a two-dimensional complexity spectrum.

4. The method for equalizing a Chirp modulation signal of an audio device according to claim 1, wherein: The step of extracting the long-range correlation characteristics of the Chirp modulation signal using the fractal analysis algorithm comprises: Apply discrete wavelet transform to the original Chirp signal for multi-resolution decomposition to obtain wavelet coefficients of different scales; Calculate the variance of wavelet coefficients at each scale; Analyze the relationship between the variance of wavelet coefficients and scale in a double logarithmic coordinate system and fit a linear model; Calculate the Hausdorff dimension and fractal index of the signal based on the wavelet variance slope; The local Hurst exponent was estimated using the rescaled range analysis method; The local Hurst exponent is calculated based on the sliding window, and the Hurst exponent feature map is constructed.

5. The method for equalizing a Chirp modulation signal of an audio device according to claim 1, wherein: The steps of constructing a complexity-aware adaptive equalization system include: Based on non-stationary complexity mapping, multi-scale complexity spectrum and Hurst exponent characteristic mapping, a comprehensive evaluation index of signal complexity is constructed; According to the comprehensive complexity evaluation index, the Chirp signal is adaptively segmented and divided into multiple regions with different complexities; Establish a balanced parameter library, which contains balanced parameter configurations applicable to areas of different complexity; For each signal segment, the optimal equalization parameter configuration is selected according to its complexity characteristics; Construct a complexity adaptive equalizer and apply differentiated equalization strategies to different complexity areas; Apply a smooth transition strategy between adjacent signal areas to avoid abrupt changes in the equalization effect at the area boundaries; Evaluate the quality of the equalized signal and calculate the equalization performance index; Based on the balancing performance indicators, the balancing parameter library is updated online to optimize the balancing performance.

6. The method for equalizing a Chirp modulation signal of an audio device according to claim 5, wherein: The calculation formula for constructing the comprehensive evaluation index of signal complexity is: ； in Indicates time Comprehensive evaluation index of signal complexity at ; 、 and They are non-stationary complexity weight, multi-scale complexity weight and long-range correlation weight respectively; is a non-stationary complexity map, which means the signal is in time The degree of non-stationarity at is the scale factor; is the selected set of scale factors; For scale The entropy value under is the time scale correlation function, characterizing a specific scale In time the importance of For time Hurst index at is the Hurst exponential mapping function, which is used to quantify the contribution of long-range correlation to complexity; For all selected scale factors Summation.

7. The method for equalizing a Chirp modulation signal of an audio device according to claim 1, wherein: The step of constructing the time-frequency joint optimization algorithm includes: Construct a joint time-frequency objective function that comprehensively considers the performance of equalization in both the time and frequency domains; Constructing time-domain objective function based on information geometry metric; Construct frequency domain objective function based on spectral distance; Constructing a time-frequency joint objective function based on multi-scale entropy differences; The gradient descent algorithm is used to optimize the time-frequency joint objective function and iteratively update the equalization parameters; Introducing a signal complexity adaptive gradient update strategy and adopting differentiated parameter update methods for different complexity regions; Construct a long-range correlation-aware regularization term to enhance the equalization's ability to handle long-term signal dependencies; Incorporate the long-range correlation regularization term into the objective function to obtain the final optimization goal; The gradient descent algorithm is combined with the signal complexity adaptive gradient update strategy to optimize the final optimization target and obtain the optimal equalization parameter configuration.

8. The method for equalizing a Chirp modulation signal of an audio device according to claim 7, wherein: The calculation formula of the time-frequency joint objective function is: ； in represents the time-frequency joint objective function, which is the overall goal of optimizing the equalization parameters. is the time domain objective function, used to evaluate the performance of equalization in the time domain; is the frequency domain objective function, which is used to evaluate the performance of equalization in the frequency domain; is the time-frequency joint objective function, which is used to evaluate the performance of equalization in the time-frequency joint domain; 、 and Represent the weight coefficients of the time domain objective function, frequency domain objective function and time-frequency joint objective function respectively; represents the time variable, represents a frequency variable, Represents a set of equalization parameters.

9. The method for equalizing a Chirp modulation signal of an audio device according to claim 7, wherein: The final optimization objective is expressed as: ； in represents the final optimization objective function, represents the time-frequency joint objective function, A regularization term that represents the perception of long-range dependencies; represents the time variable, represents a frequency variable, Represents a set of equalization parameters.

10. An equalization system for a chirp modulation signal of an audio device, characterized in that: The method for equalizing a Chirp modulation signal of an audio device according to any one of claims 1 to 9 comprises: Information geometry mapping module, used to analyze the non-stationary characteristics of Chirp modulated signals and generate non-stationary complexity mapping; Multi-scale entropy analysis module, used to process Chirp modulated signals and generate multi-scale complexity spectra; Fractal analysis module, used to extract the long-range correlation characteristics of Chirp modulation signals and output Hurst exponent feature map; Complexity-aware adaptive equalization module, used to implement adaptive equalization processing of Chirp signals; The time-frequency joint optimization module is used to achieve global optimization and adaptive adjustment of equalization parameters.

Images