EQ filter parameter optimization method and device, storage medium and electronic equipment

Abstract

The invention provides an EQ filter parameter optimization method and device, a storage medium and electronic equipment, and is applied to the technical field of audio signal processing. The method comprises the following steps: acquiring original acoustic frequency response data and target acoustic frequency response data corresponding to a hearing scene; according to the hearing scene, target EQ filter configuration is determined from a preset filter configuration strategy, and the target EQ filter configuration comprises a filter type, a filter number and initial filter parameters; based on the initial filter parameters, performing optimization fitting on the original acoustic frequency response data to obtain fitted acoustic frequency response data; constructing a target function based on the difference between the fitting acoustic frequency response data and the target acoustic frequency response data; and the optimization objective function is taken as an objective, and the filter parameter optimization result corresponding to the hearing scene is iteratively determined, so that the efficiency, the accuracy and the intelligent level are improved.

Description

Technical Field

[0001] This application relates to the field of audio signal processing technology, and in particular to an EQ filter parameter optimization method, apparatus, storage medium and electronic device. Background Technology

[0002] With the rapid development of wireless audio technology, true wireless stereo (TWS) earbuds have become mainstream personal audio devices. To enhance the listening experience, modern TWS earbuds generally have a built-in digital equalizer (EQ), which adjusts the gain of different frequency bands to correct the inherent acoustic characteristics of the earbuds, making their output more consistent with the target curve or the frequency response of the user's subjective preferences.

[0003] Currently, the design and optimization of EQ filter parameters includes the following steps: First, acoustic engineers use dedicated simulation software on an acoustic simulation platform to perform finite element modeling and simulation analysis of the headphone's acoustic structure, or lumped parameter simulation analysis based on electroacoustic analogy, to generate the headphone's original acoustic frequency response data. Then, tuning engineers obtain the headphone's target acoustic frequency response data and the original acoustic frequency response data. Based on their personal experience, they determine the required EQ filter type and quantity in the parametric equalizer tuning interface, and manually and repeatedly adjust the EQ filter parameters. The original acoustic frequency response data is then fitted using the EQ filter parameters to obtain fitted acoustic frequency response data. By observing the deviation between the fitted acoustic frequency response data and the target acoustic frequency response data, a trial-and-error iterative process is conducted until the two are subjectively considered to have reached an acceptable level of matching.

[0004] However, the above methods suffer from low efficiency, poor accuracy, and low level of intelligence. Summary of the Invention

[0005] This application provides an EQ filter parameter optimization method, apparatus, storage medium, and electronic device to improve efficiency, accuracy, and intelligence.

[0006] Firstly, this application provides a method for optimizing EQ filter parameters, including:

[0007] The system acquires raw acoustic frequency response data and target acoustic frequency response data corresponding to the listening scenario. The target acoustic frequency response data is obtained in response to the user's scenario selection operation on the graphical interface. The target acoustic frequency response data is different for different listening scenarios.

[0008] Based on the listening scenario, the target EQ filter configuration is determined from the preset filter configuration strategy. The target EQ filter configuration includes the filter type, the number of filters, and the initial filter parameters.

[0009] Based on the initial filter parameters, the original acoustic frequency response data is optimized and fitted to obtain the fitted acoustic frequency response data.

[0010] An objective function is constructed based on the difference between the fitted acoustic frequency response data and the target acoustic frequency response data;

[0011] With the objective function as the goal, the optimization results of the filter parameters corresponding to the listening scenario are determined iteratively.

[0012] In one possible implementation, the EQ filter parameter optimization method includes: dynamically updating the target acoustic frequency response data in response to a user's editing operation on the acoustic frequency response data in a graphical interface.

[0013] In one possible implementation, the editing operation includes semantic interactive editing, dynamically updating the target acoustic frequency response data, including:

[0014] Obtain the modified description text for the acoustic frequency response data input by the user;

[0015] Based on natural language processing technology, target semantic keywords are identified from the modified description text;

[0016] Based on the preset mapping relationship between semantic keywords and frequency response adjustment operations, the target frequency response adjustment operation corresponding to the target semantic keyword is determined;

[0017] Based on the target frequency response adjustment operation, the target acoustic frequency response data is dynamically updated.

[0018] In one possible implementation, acquiring raw acoustic frequency response data includes:

[0019] The simulation frequency response data and the measured frequency response data are obtained. The simulation frequency response data is generated based on the acoustic structure model through finite element modeling and simulation analysis or lumped parameter modeling and simulation analysis based on electroacoustic analogy technology. The measured frequency response data is obtained by measuring the physical prototype.

[0020] The simulated frequency response data and the measured frequency response data are fused together to generate the original acoustic frequency response data.

[0021] In one possible implementation, the original acoustic frequency response data is optimized and fitted based on initial filter parameters to obtain fitted acoustic frequency response data, including:

[0022] Based on the initial filter parameters, determine the transfer function in the frequency domain for each filter corresponding to the listening scenario;

[0023] The transfer functions of each filter are cascaded to obtain the overall equalized transfer function;

[0024] The total equalization transfer function is multiplied in the frequency domain with the original acoustic frequency response data to obtain the fitted acoustic frequency response data.

[0025] In one possible implementation, with the objective function as the goal, the optimized filter parameters corresponding to the auditory scene are iteratively determined, including:

[0026] A genetic algorithm is used to perform a global optimization search within the search space of filter parameters to obtain the preliminary optimized parameters of the filter corresponding to the listening scenario.

[0027] Starting with the initial optimized parameters of the filter, a local optimization based on stochastic gradient descent is used to iteratively update the filter parameters. The iteration stops when the iteration termination condition is met, and the filter parameters at the time of iteration stop are taken as the optimized filter parameters corresponding to the auditory scene.

[0028] In one possible implementation, with the objective function as the goal, the optimized filter parameters corresponding to the auditory scene are iteratively determined, including:

[0029] The stochastic gradient descent method is used to perform a global optimization search in the search space of filter parameters to iteratively update the filter parameters. The iteration stops when the iteration termination condition is met, and the filter parameters at the time of iteration stop are taken as the optimized filter parameters corresponding to the listening scenario.

[0030] Secondly, this application provides an EQ filter parameter optimization device, comprising:

[0031] The acquisition module is used to acquire the raw acoustic frequency response data and the target acoustic frequency response data corresponding to the listening scene. The target acoustic frequency response data is obtained in response to the user's scene selection operation on the graphical interface. The target acoustic frequency response data corresponding to different listening scenes are different.

[0032] The determination module is used to determine the target EQ filter configuration from the preset filter configuration strategy based on the listening scenario. The target EQ filter configuration includes the filter type, the number of filters, and the initial filter parameters.

[0033] The fitting module is used to optimize and fit the original acoustic frequency response data based on the initial filter parameters to obtain the fitted acoustic frequency response data.

[0034] The module is used to construct an objective function based on the difference between the fitted acoustic frequency response data and the target acoustic frequency response data;

[0035] The iteration module is used to iteratively determine the optimized filter parameters for the corresponding auditory scene with the objective function as the goal.

[0036] In one possible implementation, the EQ filter parameter optimization device further includes an update module, which is used to dynamically update the target acoustic frequency response data in response to a user's editing operation on the acoustic frequency response data in a graphical interface.

[0037] In one possible implementation, the editing operation includes semantic interactive editing, and the update module is specifically used to: obtain the modified description text for acoustic frequency response data input by the user; identify target semantic keywords from the modified description text based on natural language processing technology; determine the target frequency response adjustment operation corresponding to the target semantic keywords based on the preset mapping relationship between semantic keywords and frequency response adjustment operations; and dynamically update the target acoustic frequency response data based on the target frequency response adjustment operation.

[0038] In one possible implementation, the acquisition module is specifically used to: acquire simulated frequency response data and measured frequency response data, wherein the simulated frequency response data is generated based on an acoustic structure model through finite element modeling simulation analysis or integrated total parameter modeling simulation analysis based on electroacoustic analogy technology, and the measured frequency response data is obtained by measuring a physical prototype; and perform data fusion processing on the simulated frequency response data and the measured frequency response data to generate original acoustic frequency response data.

[0039] In one possible implementation, the fitting module is specifically used to: determine the transfer function in the frequency domain of each filter corresponding to the listening scene based on the initial filter parameters; cascade the transfer functions of each filter to obtain the overall equalization transfer function; and multiply the overall equalization transfer function with the original acoustic frequency response data in the frequency domain to obtain the fitted acoustic frequency response data.

[0040] In one possible implementation, the iterative module is specifically used to: employ a genetic algorithm to perform a global optimization search within the search space of filter parameters to obtain preliminary optimized filter parameters corresponding to the auditory scene; starting from the preliminary optimized filter parameters, use a stochastic gradient descent method to perform local optimization to iteratively update the filter parameters; stop iterating when the iteration termination condition is met; and use the filter parameters at the time of iteration termination as the optimized filter parameters corresponding to the auditory scene.

[0041] In one possible implementation, the iteration module is further configured to: perform a global optimization search within the search space of filter parameters using stochastic gradient descent to iteratively update the filter parameters; stop the iteration when the iteration termination condition is met, and use the filter parameters at the time of iteration termination as the optimized filter parameters corresponding to the auditory scene.

[0042] Thirdly, this application provides an electronic device, including: a memory and a processor;

[0043] The memory stores instructions that the computer executes;

[0044] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0045] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible embodiments of the first aspect.

[0046] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0047] The EQ filter parameter optimization method, apparatus, storage medium, and electronic device provided in this application acquire raw acoustic frequency response data and target acoustic frequency response data corresponding to the listening scenario. The target acoustic frequency response data is obtained in response to the user's scene selection operation on the graphical interface, and the target acoustic frequency response data is different for different listening scenarios. Based on the listening scenario, a target EQ filter configuration is determined from a preset filter configuration strategy. The target EQ filter configuration includes filter type, number of filters, and initial filter parameters. Based on the initial filter parameters, the raw acoustic frequency response data is optimized and fitted to obtain fitted acoustic frequency response data. Based on the difference between the fitted acoustic frequency response data and the target acoustic frequency response data, an objective function is constructed. The optimized filter parameters corresponding to the listening scenario are iteratively determined with the objective function as the target. In this process, when the user selects a listening scenario, the target acoustic frequency response data corresponding to the listening scenario is automatically acquired, achieving rapid matching of scenario-based data without requiring the user to manually search for or redesign the target acoustic frequency response data, thus achieving flexibility in scene adaptation. Based on the mapping relationship between listening scenarios and filter configurations, the system automatically selects the appropriate EQ filter configuration, achieving intelligent initial configuration, lowering the user's operational threshold, and improving user experience. Then, it automatically executes an iterative optimization process, independent of the user's parameter tuning experience, improving the accuracy and consistency of filter parameter optimization. Ultimately, this achieves an end-to-end automated process for EQ filter parameter optimization, reducing accuracy limitations caused by human factors, lowering labor costs, and significantly improving the efficiency, consistency, and intelligence of filter design and optimization. Furthermore, differentiated processing is applied to different listening scenarios, improving scenario adaptation accuracy, ensuring the sound quality stability of subsequent audio products, and ultimately enhancing user satisfaction. Attached Figure Description

[0048] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0049] Figure 1 A schematic diagram illustrating a scenario for the EQ filter parameter optimization method provided in this application embodiment;

[0050] Figure 2 A flowchart illustrating the EQ filter parameter optimization method provided in this application embodiment. Figure 1 ;

[0051] Figure 3 A flowchart illustrating the EQ filter parameter optimization method provided in this application embodiment. Figure 2 ;

[0052] Figure 4 This is a schematic diagram of a dual-mode architecture for optimizing EQ filter parameters provided in an embodiment of this application;

[0053] Figure 5 Schematic diagram of the EQ filter parameter optimization device provided in the embodiments of this application Figure 1 ;

[0054] Figure 6 Schematic diagram of the EQ filter parameter optimization device provided in the embodiments of this application Figure 2 ;

[0055] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0056] The accompanying drawings have illustrated specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to specific embodiments. Detailed Implementation

[0057] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0058] The terms “first,” “second,” etc., used in the specification and claims of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, products, or apparatus.

[0059] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with relevant laws, regulations and standards, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0060] Currently, the design and optimization of EQ filter parameters rely on manual operation for data transfer between each stage, resulting in low efficiency and a high risk of errors. Regarding the acquisition of target acoustic frequency response data, some solutions use uniform target acoustic frequency response data regardless of the listening scenario; while others require manual searching or redesigning of target acoustic frequency response data corresponding to the listening scenario, failing to automatically adapt to different scenarios. In parameter optimization, manual intervention is required to first determine the necessary EQ filter type and quantity in the equalizer tuning interface, demanding a high level of skill. Then, EQ filter parameters are adjusted based on manual experience. After each round of adjustment, the adjusted EQ filter parameters are used to fit the original acoustic frequency response data, subjectively judging whether the deviation between the fitted acoustic frequency response data and the target acoustic frequency response data meets expectations. If not, the EQ filter parameters are adjusted again based on tuning experience until the deviation meets subjective expectations. This tuning process lacks systematic algorithmic support, and the optimization results are easily affected by the subjectivity and randomness of manual operation, leading to poor consistency and low accuracy. In summary, existing solutions suffer from low efficiency, poor accuracy, and low levels of intelligence.

[0061] To address the aforementioned technical issues, this application provides an EQ filter parameter optimization scheme. It offers a graphical interface for user interaction, automatically acquiring target acoustic frequency response data corresponding to the selected listening scenario. This enables rapid matching of scenario-based data, eliminating the need for users to manually search for or redesign target acoustic frequency response data, thus achieving flexible scenario adaptation. Based on the mapping relationship between listening scenarios and filter configurations, the system automatically selects the EQ filter configuration, achieving intelligent initial configuration, lowering the user's operational threshold, and improving the user experience. Then, it automatically executes an iterative optimization process, independent of the user's parameter tuning experience, improving the accuracy and consistency of filter parameter optimization. Ultimately, this achieves an end-to-end automated process for EQ filter parameter optimization, reducing accuracy limitations caused by human factors, lowering labor costs, and significantly improving efficiency, consistency, and intelligence.

[0062] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.

[0063] First, the application scenarios of the EQ filter parameter optimization method provided in this application embodiment will be described. This EQ filter parameter optimization method is applicable to scenarios such as audio device research and development (e.g., TWS earphones, speaker products, etc.), production line debugging, and personalized sound customization.

[0064] Figure 1 This is a schematic diagram illustrating a scenario for the EQ filter parameter optimization method provided in this application embodiment, such as... Figure 1 As shown, this application scenario includes a client 11 and a server 12. There can be at least one client 11. The server 12 is equipped with a highly integrated platform required for the entire EQ filter design process. This integrated platform can cover functions such as generating simulated frequency response data, generating measured frequency response data, optimizing EQ filter parameters, and burning EQ filter parameters to audio firmware. It can effectively solve the problem in related technologies where functions such as frequency response simulation, test data acquisition, EQ filter design, and parameter debugging are scattered across different software platforms, resulting in fragmented toolchain functions, data flow relying on manual operation, and low efficiency.

[0065] In practical applications, for example, when server 12 detects a command submitted by client 11 to generate simulated frequency response data for an audio device, it calls relevant simulation tools and acoustic structure models to simulate and obtain the simulated frequency response data for the audio device. For instance, when server 12 detects a command submitted by client 11 to optimize EQ filter parameters, it executes the EQ filter parameter optimization method provided in this application embodiment to determine the filter parameter optimization result corresponding to the listening scenario.

[0066] It should be noted that the client 11 can be a computer or laptop, etc. The server 12 can also be replaced by a server cluster or other computing devices with a certain computing power. In addition, the server 12 can be an edge computing device or a cloud server. There is no specific limit to the number of servers 12. When there are multiple servers 12, the multiple servers will work together to complete the entire EQ filter design process.

[0067] Figure 2 A flowchart illustrating the EQ filter parameter optimization method provided in this application embodiment. Figure 1 ,like Figure 2 As shown, the EQ filter parameter optimization method includes:

[0068] S201. Obtain the original acoustic frequency response data and the target acoustic frequency response data corresponding to the listening scenario.

[0069] The target acoustic frequency response data is obtained in response to the user's scene selection operation on the graphical interface. The target acoustic frequency response data is different for different listening scenarios. That is, each listening scenario corresponds to a unique sound spectrum energy distribution.

[0070] It should be understood that the target acoustic frequency response data is the ideal frequency response curve that the audio device (such as TWS earphones or speaker products) is expected to achieve after EQ filter processing. Depending on different listening scenarios, such as bass enhancement, treble enhancement, soothing, brightness, vividness, and high fidelity, it is predefined and stored, for example, on server 12. The target acoustic frequency response data serves as the fitting benchmark in the EQ filter parameter optimization process; that is, the fitted acoustic frequency response data must infinitely approximate the target acoustic frequency response data.

[0071] For example, on the graphical interface of client 11, a scene selection control with icons such as "General," "Subwoofer," and "Treble Boost" is displayed to the user. The user's command to click the "Subwoofer" scene icon is received, and the target acoustic frequency response data uniquely associated with the "Subwoofer" tag is retrieved from the storage space of server 12. The raw acoustic frequency response data of the audio device currently without EQ filter processing is imported from the storage space of server 12 (such as a test system or simulation software).

[0072] For example, the raw acoustic frequency response data includes simulated frequency response data or measured frequency response data. Simulated frequency response data is generated through finite element analysis based on the acoustic structural model of the audio device, while measured frequency response data is obtained by measuring a physical prototype (audio device prototype).

[0073] S202. Based on the listening scenario, determine the target EQ filter configuration from the preset filter configuration strategy. The target EQ filter configuration includes the filter type, the number of filters, and the initial filter parameters.

[0074] The filter configuration strategy consists of multiple pre-defined rules or mapping relationships, which are used to automatically match the set of filters that the equalizer should use and the initial filtering parameters of each filter according to the listening scenario selected by the user.

[0075] For example, filter types include low-profile filters, high-profile filters, and peak filters, and filter parameters include center frequency ( ,unit ), gain ( ,unit ), quality factor ( Value (dimensionless).

[0076] For example, based on the "bass" scene tag, the preset configuration mapping table is queried. If the "bass" listening scene is matched, it is recommended to use one low-profile filter and two peak filters. The initial filtering parameters of the low-profile filter are as follows: , The initial filtering parameters for peak filter 1 are: , The initial filtering parameters for peak filter 2 are: , .

[0077] S203. Based on the initial filter parameters, optimize and fit the original acoustic frequency response data to obtain the fitted acoustic frequency response data.

[0078] This step is based on the initial filter parameters (actually a set of multiple filter parameters), and applies the standard mathematical model in filter theory to perform pure mathematical calculations on the original acoustic frequency response data to generate predictive equalization processing results, i.e., fitting the acoustic frequency response data.

[0079] It should be understood that after each iteration in the filter parameter iterative optimization, the original acoustic frequency response data will be optimized and fitted based on the filter parameters generated in the current iteration to obtain the fitted acoustic frequency response data corresponding to the current iteration.

[0080] For example, in some embodiments, the original acoustic frequency response data is optimized and fitted based on the initial filter parameters to obtain fitted acoustic frequency response data, including: determining the transfer function in the frequency domain of each filter corresponding to the listening scene according to the initial filter parameters; cascading the transfer functions of each filter to obtain the overall equalization transfer function; and multiplying the overall equalization transfer function with the original acoustic frequency response data in the frequency domain to obtain the fitted acoustic frequency response data.

[0081] A data equalizer is implemented by cascading multiple filters in the signal stream. Each filter independently affects a local part of the spectrum, and the total effect of the multiple filters is equal to the frequency response of each filter (the frequency response of a filter can be determined by a finite set of parameters). (Complete description) The product in the complex frequency domain, that is, by adjusting the parameters of each filter, an arbitrary frequency response curve can be synthesized.

[0082] For example, based on the initial filter parameters, the complex transfer function of each filter at each frequency point is calculated separately. Then, the complex transfer functions of all filters are multiplied to obtain the overall equalized transfer function after cascading. Finally, the amplitude response of the total equalization transfer function is... Compared with the original acoustic frequency response data Multiplication in the frequency domain, i.e., linear superposition on the dB scale, generates a new frequency response curve, which is equivalent to fitting the acoustic frequency response data. It is expressed as follows:

[0083] =

[0084] S204. Based on the difference between the fitted acoustic frequency response data and the target acoustic frequency response data, construct the objective function.

[0085] The objective function is used to quantify the difference between the fitted acoustic frequency response data and the target acoustic frequency response data. The goal of the optimization process is to find a set of filter parameters that minimizes the value of the objective function.

[0086] For example, at the same frequency points, the fitted acoustic frequency response data are calculated. Acoustic frequency response data of the target The difference, expressed as Then, different weights are assigned to the differences between different frequency bands. For example, the 1kHz-4kHz frequency band, which is sensitive to human hearing, has a higher weight. Furthermore, for example, a weighted mean square error can be used to construct the objective function. .

[0087] It should be noted that the above-described form of objective function construction is only an example, and this embodiment does not limit the form of objective function construction.

[0088] S205. With the objective function as the goal, iteratively determine the optimized filter parameters corresponding to the listening scenario.

[0089] This step can be implemented using various automatic optimization algorithms, such as population-based metaheuristic algorithms, gradient-based local search algorithms, and Bayesian optimization algorithms based on probabilistic models. This embodiment does not impose any limitations on this.

[0090] Optionally, after determining the filter parameter optimization results, the optimized filter parameters can be automatically generated into EQ filter code or configuration file with preset rules (or format). An interface can be provided to directly write the EQ filter code or configuration file into the firmware of audio devices (such as headphones) or export it to the design tool, effectively improving convenience and intelligence.

[0091] It should be understood that the EQ filter parameter optimization method of this application embodiment can optimize the EQ filter parameters of multiple listening scenarios in parallel, depending on the number of listening scenarios selected by the user on the graphical interface. Simultaneously generating filter parameter optimization results for multiple listening scenarios greatly improves efficiency.

[0092] For example, consider optimizing the EQ filter parameters for a specific TWS earphone in a high-frequency enhancement scenario. When the user clicks the high-frequency enhancement button on the graphical interface, the system immediately loads the target acoustic frequency response data corresponding to that listening scenario, such as a curve with a gently rising frequency above 3kHz. The system retrieves the original acoustic frequency response data of the TWS earphone from the database, which may have attenuation at extremely high frequencies. Based on the high-frequency enhancement strategy, the system determines to use one overhead filter (responsible for overall high-frequency enhancement) and one peak filter (used for fine-tuning a specific high-frequency band). The initial filter parameters are preset as follows: overhead filter (… =5kHz, =+4dB, =0.7); Peak filter ( =12kHz, =+3dB, =2.5). First, the initial set of filter parameters is used to fit the original acoustic frequency response data. The resulting fitted acoustic frequency response data (curve) may still have a dip around 10kHz. The optimization algorithm then iteratively fits this data, gradually adjusting the parameters of the two filters until the fitted acoustic frequency response data (curve) continuously approximates the target acoustic frequency response data (curve). Finally, a set of optimized parameters is obtained, making the fitted curve highly consistent with the target curve corresponding to treble enhancement in the 3kHz-20kHz frequency band. Furthermore, this set of optimized parameters is automatically written into the headphone firmware. When the user selects the treble enhancement mode, the high-frequency detail performance of the headphones will be significantly and as expected improved, greatly enhancing the user experience.

[0093] This application's embodiments automatically acquire target acoustic frequency response data corresponding to the user's selected listening scenario, achieving rapid scenario-based data matching without requiring manual searching or redesign of target acoustic frequency response data, thus enabling flexible scenario adaptation. Based on the mapping relationship between listening scenarios and filter configurations, the system automatically selects EQ filter configurations, achieving intelligent initial configuration, reducing the user's operational threshold, and improving user experience. Then, it automatically executes an iterative optimization process, independent of the user's parameter tuning experience, improving the accuracy and consistency of filter parameter optimization. Ultimately, it achieves an end-to-end automated process for EQ filter parameter optimization, reducing accuracy limitations caused by human factors, lowering labor costs, and significantly improving the efficiency, consistency, and intelligence level of filter design and optimization. Furthermore, differentiated processing for different listening scenarios improves scenario adaptation accuracy, ensures the sound quality stability of subsequent audio products, and ultimately enhances user satisfaction.

[0094] In actual automated EQ filter parameter optimization processes, the fixed and rigid target acoustic frequency response data may hinder the fulfillment of diverse audio product requirements, personalized user needs (such as engineers being dissatisfied with the preset target acoustic frequency response curve, or users wanting to fine-tune and emphasize bass in a specific listening scenario), and precision requirements (for example, in acoustic development, engineers often need to precisely correct specific issues in a particular frequency band (such as the 5kHz resonance peak)). Therefore, to address these issues, after obtaining the preset target acoustic frequency response data, some embodiments of the EQ filter parameter optimization method further include: dynamically updating the target acoustic frequency response data in response to user editing operations on the graphical interface.

[0095] Editing acoustic frequency response data includes directly editing the data and indirectly adjusting it by editing filter parameters. This provides users with two complementary approaches to defining optimization goals: one starting from auditory objectives (curve shape) and the other from technical implementation (filter parameters), catering to the operating habits of users with different knowledge backgrounds.

[0096] Direct editing of acoustic frequency response data:

[0097] For example, in the central area of ​​the graphical interface, three curves are displayed simultaneously in an overlapping or side-by-side manner: the original acoustic frequency response curve, the target acoustic frequency response curve, and the fitted acoustic frequency response curve. Among them, the original acoustic frequency response curve is set as an interactive object, and several draggable control points are preset on the curve. For example, the control points are distributed at key frequency positions, and the part of the curve between two control points can be automatically generated by a smoothing algorithm (such as spline interpolation).

[0098] In one implementation, the user clicks a control point with a mouse or touchscreen and drags it vertically upwards or downwards to adjust the sound pressure level at that frequency location. The system then invokes a built-in curve interpolation algorithm to instantly recalculate and redraw the entire target acoustic frequency response curve based on the new positions of all control points. Accordingly, after the target acoustic frequency response curve changes, the system can automatically trigger iterative updates to the fitted acoustic frequency response curve, using the new target acoustic frequency response curve as a constraint.

[0099] In another implementation, the user selects a frequency band (e.g., 200Hz to 500Hz) on the target acoustic frequency response curve using a mouse or touchscreen, and then selects the amount of sound pressure level to be adjusted in that frequency band, for example, +4dB. Correspondingly, the system automatically increases the sound pressure level of that frequency band by a uniform 4dB, and performs smooth transition processing on the points at the edge of the area to avoid steep jumps in the curve.

[0100] The acoustic frequency response data can be indirectly adjusted by editing the filter parameters.

[0101] For example, the graphical interface includes a filter parameter control panel. This control panel displays a set of control controls, such as numeric input boxes or sliders, for each filter configured for the target acoustic frequency response curve, to adjust the gain. Center frequency and quality factor After the user adjusts the filter parameters using the above controls, the system automatically corrects the target acoustic frequency response curve based on the adjusted filter parameters.

[0102] In this embodiment of the application, under the premise of automation framework, the optimization target is dynamically, visually and in real time redefined through interactive editing operations, so as to achieve an effective unity between automation efficiency and precise human control, and achieve human-machine collaborative intelligent optimization, thereby better meeting the differentiated needs of audio products, the personalized needs of users and the needs of precision.

[0103] In practical applications, for example, if an engineer listens to the acoustic effect of an audio device after optimizing the filter parameters but is still unsatisfied, there's no need to manually determine the specific EQ filter parameters to be adjusted. Instead, a voice or text description can be directly input to redefine the target acoustic frequency response data. In some embodiments, the editing operation includes semantic interactive editing, dynamically updating the target acoustic frequency response data. This includes: acquiring user-inputted modified description text for the acoustic frequency response data; identifying target semantic keywords from the modified description text using natural language processing technology; determining the target frequency response adjustment operation corresponding to the target semantic keywords based on a preset mapping relationship between semantic keywords and frequency response adjustment operations; and dynamically updating the target acoustic frequency response data based on the target frequency response adjustment operation.

[0104] For example, the graphical interface provides a voice input button or a text input box. For instance, a user clicks the voice input button and enters "a bit more bass, but not too booming." The system then calls the speech recognition engine to convert the speech stream into text in real time, i.e., modifying the descriptive text. Alternatively, the user types in the text input box "boost the 3kHz area by 2dB to make the voice stand out more." Then, an application-adaptive hierarchical semantic understanding model performs word segmentation, part-of-speech tagging, audio domain entity recognition, and operation intent extraction on the modified descriptive text, ultimately outputting a structured semantic understanding result (i.e., the target semantic keywords). These target semantic keywords can be further mapped to standard keywords; for example, "booming" can be mapped to "too much low frequency causes discomfort." Using standardized target semantic keywords as an index, a search is performed within the mapping relationship between semantic keywords and frequency response adjustment operations to determine the target frequency response adjustment operation corresponding to the target semantic keywords. This mapping relationship is a predefined rule base or knowledge base, establishing a correspondence between natural language vocabulary and specific acoustic processing operations (e.g., mapping "bass" to "20-250Hz band," and mapping "powerful" to "overall enhancement of this frequency band and appropriate increase of Q value to enhance impact"). This mapping relationship can be constructed by acoustic experts. Finally, the target frequency response adjustment operation is converted into mathematical operations on the target acoustic frequency response data. The new target acoustic frequency response data after the operation is smoothed and constrained to prevent unnatural sharp changes, violations of physical constraints, or violations of auditory constraints in the new target acoustic frequency response data.

[0105] Optionally, the system can also record the user's semantic commands and the final confirmed modification of the target acoustic frequency response data. By analyzing massive amounts of data, it can automatically discover new mapping patterns and continuously update the mapping relationship between semantic keywords and frequency response adjustment operations.

[0106] This application embodiment, by deeply integrating natural language processing and acoustic engineering knowledge, understands the user's intention to adjust the sound effects and automatically converts it into precise modification of the target acoustic frequency response data, greatly improving efficiency and lowering the usage threshold, and further enhancing intelligence.

[0107] In some embodiments, acquiring raw acoustic frequency response data includes:

[0108] Step 1.1: Obtain simulated frequency response data and measured frequency response data.

[0109] Simulated frequency response data is generated based on acoustic structural models through finite element modeling simulation analysis or lumped parameter modeling simulation analysis based on electroacoustic analogy techniques. For example, based on the three-dimensional acoustic structural model of headphones, speaker unit parameters, and material properties, numerical calculation methods such as finite element analysis or boundary element analysis are used to simulate and calculate the sound pressure level of the headphones at the eardrum reference point of a standard coupler (such as IEC 711) as a function of frequency (i.e., simulated frequency response data). Measured frequency response data is obtained by measuring a physical prototype, for example, by mounting a physical headphone prototype to an IEC 711 coupler or an artificial ear according to standard wearing methods. An audio analyzer is used to generate a logarithmic sweep sine wave signal to drive the headphones to produce sound, while simultaneously acquiring the response signal from the microphone within the coupler, thereby calculating the measured frequency response data.

[0110] For example, preprocessing is performed on both simulated and measured frequency response data. Preprocessing includes format standardization, reference alignment, and filtering. For instance, the simulated and measured frequency response data are converted to a suitable data structure, ensuring that their frequency vectors use the same logarithmic distribution (e.g., 1 / 24 octave) and the same number of frequency points (e.g., 401 points). Missing frequency points are interpolated to ensure the data lies on the same frequency grid. Since there may be system offsets between the simulated and measured absolute sound pressure level references, amplitude alignment is required for both data. Therefore, after format standardization, reference alignment is performed on the simulated and measured frequency response data. Filtering can also be performed, for example, by removing outliers through moving averages or wavelet transforms to improve data quality.

[0111] Step 1.2: Perform data fusion processing on the simulated frequency response data and the measured frequency response data to generate the original acoustic frequency response data.

[0112] For example, data fusion processing includes weighted averaging, frequency band selection fusion, or confidence-based fusion strategies. Taking one type of adaptive weighted fusion algorithm based on frequency-varying confidence as an example, the confidence weights for different frequency bands are preset. For each frequency point, the preset confidence weights are used to calculate the sound pressure level after fusing the simulated and measured frequency response data, resulting in the fused acoustic frequency response data. Optionally, the fused acoustic frequency response data can be lightly smoothed and filtered to eliminate minor discontinuities that may be caused by sudden weight changes. Optionally, outlier correction can also be performed on the fused acoustic frequency response data. For example, if the deviation between a certain frequency point and the measured or simulated data exceeds a physical reasonableness threshold (e.g., 10 dB), the measured data is used as the primary basis for correction. Finally, the original acoustic frequency response data is obtained.

[0113] In this embodiment, considering that simulation data is susceptible to model simplification and measured data is susceptible to random errors and one-off manufacturing deviations, intelligent fusion overcomes the inherent defects of a single data source, obtaining an estimate of the actual product performance that is closer to the ideal design state. This significantly improves the accuracy and reliability of the data, laying a reliable data foundation for high-precision EQ filter parameter optimization.

[0114] In some embodiments, the optimization of the filter parameters corresponding to the auditory scene is iteratively determined with the objective function as the goal, including:

[0115] S2051. A genetic algorithm is used to perform a global optimization search within the search space of filter parameters to obtain the preliminary optimized parameters of the filter corresponding to the listening scenario.

[0116] For example, the initial optimization parameters of the filter can be generated through the following process:

[0117] Population initialization: Based on the initial filter parameters, an initial population of P individuals is randomly generated within a preset mutation range, with each individual representing a set of filter parameter combinations.

[0118] Fitness calculation: For each individual, calculate the difference between its corresponding fitted acoustic frequency response data and the target acoustic frequency response data, and use the negative value of the objective function as the fitness.

[0119] Selection operation: Based on fitness, roulette wheel selection or tournament selection is used to select the best individuals to enter the next generation.

[0120] Crossover operation: Perform parameter crossover on the selected individuals with preset crossover probabilities to generate new individuals.

[0121] Mutation operation: Randomly perturb the parameters of the new individual with a preset mutation probability.

[0122] If the genetic algorithm fails to significantly improve the optimal fitness after repeating the evolution for a preset number of generations or after several consecutive generations, the genetic algorithm phase is terminated, and the filter parameters corresponding to the individual with the highest fitness in the historical iterations are output as the initial optimization parameters of the filter.

[0123] S2052. Starting with the initial optimized parameters of the filter, local optimization is performed using the stochastic gradient descent method to iteratively update the filter parameters. The iteration stops when the iteration termination condition is met, and the filter parameters at the time of iteration stop are taken as the optimized filter parameters corresponding to the listening scenario.

[0124] For example, the initial optimized filter parameters in step S2051 are used as the initial parameters for the local optimization stage of stochastic gradient descent. Based on these initial parameters, steps S203 and S204 are executed. Then, the gradient of the objective function with respect to each filter parameter is calculated, and the filter parameters are updated along the negative gradient direction. An adaptive learning rate adjustment strategy or a momentum acceleration strategy is adopted to improve the convergence speed. The objective function value is recalculated after each parameter update. The iteration termination condition is continuously checked. When any iteration termination condition is met, the iteration is stopped immediately, and the filter parameters at this time (the t-th iteration) are output as the final filter parameter optimization result. The iteration termination condition includes any of the following: the number of iterations reaches a preset maximum value; the relative decrease of the objective function value in K consecutive iterations converges; the norm of the gradient vector converges; and the change of the filter parameters converges.

[0125] In this embodiment, the global exploration capability of the genetic algorithm is utilized to effectively avoid the risk of traditional gradient methods falling into local optima due to poor starting points. This ensures that the starting point of the stochastic gradient descent method is located in the vicinity of the global optimum. Furthermore, the stochastic gradient descent method is used to achieve efficient and accurate local convergence near the high-quality solution. Through this hybrid strategy, the convergence speed is improved while avoiding local optima, achieving a perfect balance between global optima and convergence accuracy.

[0126] In some embodiments, the optimization of the objective function is used as the objective to iteratively determine the filter parameter optimization result corresponding to the listening scenario, including: using stochastic gradient descent to perform a global optimization search in the search space of the filter parameters to iteratively update the filter parameters; stopping the iteration when the iteration termination condition is met, and taking the filter parameters at the time of iteration termination as the filter parameter optimization result corresponding to the listening scenario.

[0127] For example, stochastic gradient descent efficiently fits target acoustic frequency response data through numerical gradient calculation and the Adam optimizer. Specifically, it includes the following steps:

[0128] Initialization phase: The initial filter parameters are used as the initial parameters for stochastic gradient descent, and the relevant parameters of the Adam optimizer are initialized.

[0129] Iterative optimization phase: In each iteration, the following sub-steps are executed:

[0130] 1) Create a filter bank based on the current filter parameters;

[0131] 2) Calculate the frequency response data of the current filter bank across the entire frequency band;

[0132] 3) Determine the loss value of the objective function by comparing the current frequency response data with the target acoustic frequency response data;

[0133] 4) The gradient of each filter parameter is calculated in parallel using the central difference method;

[0134] 5) Update filter parameters using the Adam optimizer;

[0135] 6) Constrain the updated filter parameters to ensure they are within a reasonable range;

[0136] 7) Check if the iteration termination condition is met. If the iteration termination condition is met, terminate the iteration.

[0137] Result output stage: The filter parameters at the time of iteration stop are output as the filter parameter optimization results corresponding to the listening scene.

[0138] In some embodiments, after obtaining the optimized filter parameters corresponding to the auditory scene, the method further includes: recalculating and updating the fitted acoustic frequency response data in real time in response to the user's manual adjustment of the current filter parameters; and displaying a description of the auditory perception change corresponding to the current parameter adjustment on a graphical interface.

[0139] Based on the above embodiments, combined with Figure 3 The EQ filter parameter optimization method of the embodiments of this application will be described in detail. Figure 3 A flowchart illustrating the EQ filter parameter optimization method provided in this application embodiment. Figure 2 ,like Figure 3 As shown, the EQ filter parameter optimization method includes:

[0140] S301. Obtain the original acoustic frequency response data and the target acoustic frequency response data corresponding to the listening scenario.

[0141] For example, simulated frequency response data and measured frequency response data are acquired. These data are then preprocessed, including format standardization, reference alignment, and filtering. Finally, the preprocessed simulated and measured frequency response data are fused to generate the original acoustic frequency response data.

[0142] S302. Based on the listening scenario, determine the target EQ filter configuration from the preset filter configuration strategy. The target EQ filter configuration includes the filter type, the number of filters, and the initial filter parameters.

[0143] S303. Based on the initial filter parameters, optimize and fit the original acoustic frequency response data to obtain the fitted acoustic frequency response data.

[0144] S304. Based on the difference between the fitted acoustic frequency response data and the target acoustic frequency response data, construct the objective function.

[0145] S305. A genetic algorithm is used to perform a global optimization search within the search space of filter parameters to obtain the preliminary optimized parameters of the filter corresponding to the listening scenario.

[0146] S306. Starting with the initial optimized parameters of the filter, local optimization is performed using the stochastic gradient descent method to iteratively update the filter parameters. The iteration stops when the iteration termination condition is met, and the filter parameters at the time of iteration stop are taken as the optimized filter parameters corresponding to the listening scenario.

[0147] S307, Interactive Debugging and Real-time Preview.

[0148] For example, in response to a user's manual adjustment of the current filter parameters, the fitted acoustic frequency response data is recalculated and updated in real time.

[0149] S308. Export and apply the filter parameter optimization results.

[0150] For example, the filter parameter optimization results are automatically generated into EQ filter code or configuration files with preset rules (or forms), and an interface is provided to directly write the EQ filter code or configuration files into the firmware of audio devices (such as headphones) or export them to design tools.

[0151] Next, this application further proposes an EQ filter parameter optimization based on a dual-mode architecture.

[0152] Figure 4 This is a schematic diagram of a dual-mode architecture for optimizing EQ filter parameters provided in an embodiment of this application. (Combined with...) Figure 4 This paper describes the dual-mode computing architecture that combines cloud and edge computing.

[0153] Understandably, edge mode is suitable for latency-sensitive tasks such as real-time interactive debugging, rapid preview, single-sense scenario optimization, and lightweight optimization algorithms (such as simplified genetic algorithms and simplified gradient descent algorithms). Edge mode enables low-latency response, allowing engineers to adjust parameters and view frequency response changes in real time during debugging, thus improving debugging efficiency. Cloud mode is suitable for computationally intensive tasks such as concurrent optimization of multi-sense scenarios, complex algorithm calculations (such as genetic algorithms and gradient descent algorithms), large-scale data fitting, and model training. Leveraging the powerful computing capabilities of cloud servers, it supports parallel processing of multiple tasks, significantly improving optimization efficiency and processing power, and is especially suitable for batch debugging in production lines or parallel development of multiple projects. In addition, the cloud can communicate with the edge through RESTful APIs or message queues, receive tasks, and return optimization results. The cloud should support Docker containerized deployment to ensure environmental consistency.

[0154] For example, the EQ filter parameter optimization method is executed in a cloud-edge collaborative architecture. For instance, a user issues an EQ filter parameter optimization task through a client, the edge receives the EQ filter parameter optimization task, and the intelligent task scheduler in the edge evaluates the complexity, real-time requirements, network status, edge resource utilization, and cloud resource utilization of the EQ filter parameter optimization task. Then, according to the elastic scheduling algorithm or preset scheduling rules, the EQ filter parameter optimization task is allocated to the edge or the cloud for execution. Alternatively, the EQ filter parameter optimization task can be broken down into multiple sub-tasks, with some sub-tasks executed at the edge and some sub-tasks executed in the cloud.

[0155] Optionally, it supports breakpoint resumption and status synchronization to ensure that tasks are not interrupted and data is not lost when the network is unstable or when switching modes. It provides a unified data interface and protocol to ensure data format consistency and communication reliability between the edge and the cloud.

[0156] In this embodiment, intelligent scheduling enables collaborative computing between the cloud and the edge. This dual-mode architecture achieves dynamic allocation and efficient utilization of resources, improving system concurrency and flexibility.

[0157] Figure 5 Schematic diagram of the EQ filter parameter optimization device provided in the embodiments of this application Figure 1 ,like Figure 5 As shown, the EQ filter parameter optimization device 50 provided in this embodiment includes: an acquisition module 51, a determination module 52, a fitting module 53, a construction module 54, and an iteration module 55. Wherein:

[0158] The acquisition module 51 is used to acquire the original acoustic frequency response data and the target acoustic frequency response data corresponding to the listening scene. The target acoustic frequency response data is acquired in response to the user's scene selection operation on the graphical interface. The target acoustic frequency response data corresponding to different listening scenes are different.

[0159] The determination module 52 is used to determine the target EQ filter configuration from the preset filter configuration strategy according to the listening scenario. The target EQ filter configuration includes the filter type, the number of filters, and the initial filter parameters.

[0160] The fitting module 53 is used to optimize and fit the original acoustic frequency response data based on the initial filter parameters to obtain the fitted acoustic frequency response data.

[0161] Module 54 is used to construct an objective function based on the difference between the fitted acoustic frequency response data and the target acoustic frequency response data;

[0162] Iteration module 55 is used to iteratively determine the optimized filter parameters corresponding to the listening scenario with the objective function as the goal.

[0163] like Figure 6 As shown, in one possible implementation, the EQ filter parameter optimization device 50 further includes an update module 56, which is used to dynamically update the target acoustic frequency response data in response to a user's editing operation on the acoustic frequency response data in the graphical interface.

[0164] In one possible implementation, the editing operation includes semantic interactive editing, and the update module 56 is specifically used to: obtain the modified description text for the acoustic frequency response data input by the user; identify the target semantic keywords from the modified description text based on natural language processing technology; determine the target frequency response adjustment operation corresponding to the target semantic keywords based on the preset mapping relationship between the semantic keywords and the frequency response adjustment operation; and dynamically update the target acoustic frequency response data based on the target frequency response adjustment operation.

[0165] In one possible implementation, the acquisition module 51 is specifically used to: acquire simulated frequency response data and measured frequency response data, wherein the simulated frequency response data is generated based on the acoustic structure model through finite element analysis, and the measured frequency response data is obtained by measuring the physical prototype; and perform data fusion processing on the simulated frequency response data and the measured frequency response data to generate the original acoustic frequency response data.

[0166] In one possible implementation, the fitting module 53 is specifically used to: determine the transfer function in the frequency domain of each filter corresponding to the listening scene based on the initial filter parameters; cascade the transfer functions of each filter to obtain the overall equalization transfer function; and multiply the overall equalization transfer function with the original acoustic frequency response data in the frequency domain to obtain the fitted acoustic frequency response data.

[0167] In one possible implementation, the iteration module 55 is specifically used to: employ a genetic algorithm to perform a global optimization search within the search space of filter parameters to obtain preliminary optimized filter parameters corresponding to the listening scenario; starting from the preliminary optimized filter parameters, employing a stochastic gradient descent method to perform local optimization to iteratively update the filter parameters; stopping the iteration when the iteration termination condition is met; and using the filter parameters at the time of iteration termination as the optimized filter parameters corresponding to the listening scenario.

[0168] In one possible implementation, the iteration module 55 is further configured to: perform a global optimization search in the search space of filter parameters using stochastic gradient descent to iteratively update the filter parameters; stop the iteration when the iteration termination condition is met, and use the filter parameters at the time of iteration termination as the filter parameter optimization result corresponding to the auditory scene.

[0169] The EQ filter parameter optimization device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0170] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 7 As shown, the electronic device 70 provided in this embodiment includes at least one processor 701 and a memory 702. Optionally, the electronic device 70 further includes a communication component 703. The processor 701, memory 702, and communication component 703 are connected via a bus 704.

[0171] In a specific implementation, at least one processor 701 executes computer execution instructions stored in memory 702, causing at least one processor 701 to perform the above-described method.

[0172] The specific implementation process of processor 701 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0173] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0174] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0175] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0176] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0177] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0178] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0179] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0180] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0181] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0182] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0183] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0184] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0185] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for optimizing EQ filter parameters, characterized in that, include: The raw acoustic frequency response data and the target acoustic frequency response data corresponding to the listening scene are obtained. The target acoustic frequency response data is obtained in response to the user's scene selection operation on the graphical interface. The target acoustic frequency response data corresponding to different listening scenes are different. Based on the listening scenario, a target EQ filter configuration is determined from a preset filter configuration strategy. The target EQ filter configuration includes filter type, number of filters, and initial filter parameters. Based on the initial filter parameters, the original acoustic frequency response data is optimized and fitted to obtain fitted acoustic frequency response data. Based on the difference between the fitted acoustic frequency response data and the target acoustic frequency response data, an objective function is constructed; With the objective function as the goal, the optimization results of the filter parameters corresponding to the auditory scene are determined iteratively.

2. The EQ filter parameter optimization method according to claim 1, characterized in that, Also includes: The target acoustic frequency response data is dynamically updated in response to the user's editing operation on the graphical interface.

3. The EQ filter parameter optimization method according to claim 2, characterized in that, The editing operation includes semantic interactive editing, and the dynamic updating of the target acoustic frequency response data includes: Obtain the modified description text for the acoustic frequency response data input by the user; Based on natural language processing technology, target semantic keywords are identified from the modified description text; Based on the preset mapping relationship between semantic keywords and frequency response adjustment operations, the target frequency response adjustment operation corresponding to the target semantic keyword is determined; Based on the target frequency response adjustment operation, the target acoustic frequency response data is dynamically updated.

4. The EQ filter parameter optimization method according to any one of claims 1 to 3, characterized in that, The acquisition of raw acoustic frequency response data includes: The simulation frequency response data and the measured frequency response data are obtained. The simulation frequency response data is generated based on the acoustic structure model through finite element modeling simulation analysis or lumped parameter modeling simulation analysis based on electroacoustic analogy technology. The measured frequency response data is obtained by measuring the physical prototype. The simulated frequency response data and the measured frequency response data are fused together to generate the original acoustic frequency response data.

5. The EQ filter parameter optimization method according to any one of claims 1 to 3, characterized in that, The optimization and fitting of the original acoustic frequency response data based on the initial filter parameters to obtain fitted acoustic frequency response data includes: Based on the initial filter parameters, determine the transfer function in the frequency domain for each filter corresponding to the auditory scene; The transfer functions of the filters are concatenated to obtain the overall equalized transfer function; The total equalization transfer function is multiplied in the frequency domain with the original acoustic frequency response data to obtain the fitted acoustic frequency response data.

6. The EQ filter parameter optimization method according to any one of claims 1 to 3, characterized in that, The step of iteratively determining the optimized filter parameters corresponding to the auditory scene with the objective function as the goal includes: A genetic algorithm is used to perform a global optimization search within the search space of filter parameters to obtain the preliminary optimized parameters of the filter corresponding to the hearing scenario. Starting with the initial optimized parameters of the filter, a local optimization based on stochastic gradient descent is used to iteratively update the filter parameters. The iteration stops when the iteration termination condition is met, and the filter parameters at the time of iteration stop are taken as the optimized filter parameters corresponding to the auditory scene.

7. The EQ filter parameter optimization method according to any one of claims 1 to 3, characterized in that, The step of iteratively determining the optimized filter parameters corresponding to the auditory scene with the objective function as the goal includes: The stochastic gradient descent method is used to perform a global optimization search within the search space of filter parameters in order to iteratively update the filter parameters; The iteration stops when the iteration termination condition is met, and the filter parameters at the time of iteration stop are taken as the filter parameter optimization result corresponding to the auditory scene.

8. An EQ filter parameter optimization device, characterized in that, include: The acquisition module is used to acquire raw acoustic frequency response data and target acoustic frequency response data corresponding to the listening scene. The target acoustic frequency response data is acquired in response to the user's scene selection operation on the graphical interface. The target acoustic frequency response data corresponding to different listening scenes are different. The determination module is used to determine the target EQ filter configuration from a preset filter configuration strategy based on the listening scenario. The target EQ filter configuration includes filter type, number of filters, and initial filter parameters. The fitting module is used to optimize and fit the original acoustic frequency response data based on the initial filter parameters to obtain fitted acoustic frequency response data. The construction module is used to construct an objective function based on the difference between the fitted acoustic frequency response data and the target acoustic frequency response data; The iterative module is used to iteratively determine the optimized filter parameters corresponding to the auditory scene with the objective function as the goal.

9. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed, are used to implement the method as described in any one of claims 1 to 7.

11. A computer program product, characterized in that, It includes a computer program that, when executed, implements the method of any one of claims 1 to 7.

Images