AI-based auditory range adaptive high-fidelity audio coding and decoding algorithm and compression optimization method

Abstract

The invention discloses an AI-based high-fidelity audio coding and decoding algorithm adaptive to an auditory range and a compression optimization method, and the algorithm comprises an audio signal preprocessing module which is used for carrying out the preprocessing of an input audio signal, including but not limited to denoising, frequency filtering, gain adjustment, time domain / frequency domain transformation and the like, and carrying out the compression optimization of the audio signal; and the preprocessing module is used for performing feature extraction through a convolutional neural network (CNN) or an auto-encoder network so as to improve the signal-to-noise ratio of an audio signal and remove environmental noise, and generating optimized audio features. Through an AI-based auditory range adaptation algorithm, dynamic optimization can be performed according to individual auditory characteristics, so that the audio coding and decoding process is more in line with the perception characteristics of human ears, better balance is obtained between the compression ratio and the tone quality, higher audio experience is provided, and through the adaptive quantization and code rate control technology, the audio coding and decoding efficiency is improved. According to the method, on the premise of keeping the high-fidelity tone quality, the deep learning algorithm is utilized to optimize the allocation of computing resources, and the consumption of the computing resources is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of audio coding and decoding, and specifically to a high-fidelity audio coding and decoding algorithm and compression optimization method based on AI that adapt to the auditory range. Background Art

[0002] The high-fidelity audio coding and decoding algorithm based on AI that adapts to the auditory range aims to improve the efficiency and quality of audio coding and decoding through artificial intelligence technology. Traditional audio coding and decoding methods usually rely on fixed parameters and general compression models, and it is difficult to fully consider the perceptual characteristics of the auditory system. The AI algorithm, on the other hand, can automatically adjust the audio compression strategy according to the auditory sensitivity of different people and environmental noise conditions, achieving more efficient compression and more accurate sound quality reconstruction. This algorithm uses technologies such as deep learning and convolutional neural network (CNN) to identify and optimize the important frequency components in the audio signal, reducing the parts that are insensitive to the human ear, thereby significantly reducing the requirements for data transmission and storage while retaining high-fidelity sound quality. In addition, the compression optimization method further improves the compression ratio and sound quality balance in the audio coding and decoding process by intelligently adjusting the encoder parameters, dynamically adapting to adjust the bit rate and quantization strategy.

[0003] Although the AI-based high-fidelity audio codec algorithm and compression optimization method adapted to the auditory range have significant advantages in audio quality and compression efficiency, there are still some drawbacks in the existing technologies: High computational complexity: AI-driven audio codec algorithms usually require a large amount of computational resources, especially in the training and inference stages of deep learning models. This may lead to performance bottlenecks for real-time audio processing, low-latency applications, or devices with limited computational resources (such as mobile devices, embedded systems, etc.); Dependence on model training and data requirements: The training of AI models usually requires a large amount of labeled data to ensure the accuracy and generalization ability of the algorithms. If the training data is insufficient or unrepresentative, it may lead to overfitting of the model, thus affecting audio quality and adaptability. In addition, data privacy and copyright issues may also limit the acquisition of data; Personalization issues in the auditory range: Although AI can be optimized according to the perceptual characteristics of the human ear, the auditory sensitivity of each person is different, especially under different hearing conditions (such as hearing loss). This makes the existing adaptive adjustments may not achieve the best results for each user. To achieve personalized optimization, additional individualized training or real-time adjustment mechanisms may be required, which also increases the system complexity; Generalizability and cross-device adaptability: Different audio devices (such as headphones, speakers, mobile phones, etc.) have different frequency response characteristics, and AI algorithms need to be adjusted for different hardware to ensure consistent sound quality performance. However, there may be certain limitations in the cross-device adaptation of existing algorithms, and it is difficult to achieve complete generality; Balance between compression ratio and sound quality: Although AI methods can improve compression efficiency to a certain extent, excessive compression may lead to loss of sound quality, especially in high-dynamic-range audio or complex audio scenarios. How to balance the compression ratio and sound quality remains a challenge, especially under extreme conditions, where it may not be possible to meet the requirements of low latency, high compression ratio, and high-fidelity sound quality simultaneously; Interpretability issues of the algorithm: Deep learning models are often regarded as "black boxes" and lack intuitive interpretability. In practical applications, especially in the optimization process of audio codecs, the lack of transparency may lead to difficulties in understanding and tuning the algorithm behavior. This poses challenges to debugging, optimization, and the maintainability of the system; Real-time and latency issues: Although AI can bring higher efficiency and quality in audio compression and decoding, the latency issue in real-time applications still exists. Especially for application scenarios that require low-latency processing, such as voice communication or online games, the complex calculations of AI algorithms may lead to unacceptable latency, affecting the user experience.

[0004] Therefore, we propose an AI-based high-fidelity audio codec algorithm and compression optimization method adapted to the auditory range. Summary of the Invention

[0005] To achieve the above object, the present invention provides the following technical solutions: A high-fidelity audio encoding and decoding algorithm and compression optimization method adapted to the auditory range based on AI, including: An audio signal preprocessing module for preprocessing the input audio signal, including but not limited to denoising, frequency filtering, gain adjustment, time-domain / frequency-domain transformation, etc., to improve the signal-to-noise ratio of the audio signal and remove environmental noise. The preprocessing module extracts features through a convolutional neural network (CNN) or an autoencoder (AE) network to generate optimized audio features;

[0006] An auditory range analysis module, based on the user's personalized auditory data, uses deep learning algorithms to analyze the audio signal, automatically generates the user's auditory sensitivity model, and dynamically adjusts the encoding strategy according to this model to optimize the frequency response and compression parameters of the audio signal. The model can adjust the compression range according to the user's degree of hearing loss (such as high-frequency hearing loss, low-frequency hearing sensitivity);

[0007] An audio encoding module, based on the auditory model and compression parameters output by the auditory range analysis module, encodes the audio using adaptive quantization and bitrate control algorithms, dynamically adjusts the quantization scale and bitrate during encoding to optimize the balance between compression ratio and sound quality. The encoding module performs encoding optimization based on transform coding (such as MDCT, DCT) and neural network-based predictive coding (such as autoregressive model, convolutional neural network);

[0008] An audio decoding module, based on the compressed data output by the audio encoding module, decodes the compressed audio using a deep neural network (DNN) or a convolutional neural network (CNN) decoding network to restore a high-fidelity sound quality as close as possible to the original audio. The decoding module restores the high-frequency or low-frequency loss part through adaptive denoising and compensation techniques;

[0009] An optimization feedback module for dynamically adjusting the parameters in the encoding and decoding process according to the user's real-time feedback and environmental changes to ensure that the sound quality remains optimal in different environments. The feedback module adjusts the relevant parameters in the auditory range analysis module and the encoding module in real time by monitoring information such as environmental noise level and sound quality evaluation.

[0010] Preferably, the audio signal preprocessing module further includes an adaptive noise suppression algorithm for dynamically adjusting the noise suppression intensity of the audio signal according to the environmental noise intensity. The noise suppression algorithm performs noise identification based on a deep convolutional network (CNN) and performs noise reduction processing on the audio signal using time-frequency analysis.

[0011] Preferably, the auditory range analysis module trains on the user's hearing test data to establish a personalized auditory model, optimizing the frequency response and compression strategy of the audio. The personalized auditory model is based on self-supervised learning or transfer learning techniques and can dynamically adjust the frequency bandwidth of audio compression according to the user's hearing curve.

[0012] Preferably, the audio encoding module adopts a dynamic encoding strategy based on deep learning and dynamically selects an adaptive encoding mode according to the frequency response adjustment strategy output by the auditory range analysis module. The dynamic encoding strategy includes adaptive quantization, bitrate allocation, dynamic audio feature extraction, etc.

[0013] Preferably, the audio decoding module includes an audio restoration technique based on a neural network. By performing deep learning on the high-order features of the compressed audio, it restores a high-fidelity sound quality close to the original audio. The audio restoration technique uses generative adversarial network (GAN) or variational autoencoder (VAE) techniques for audio reconstruction to repair the high-frequency or detailed information lost during the compression process.

[0014] Preferably, the optimization feedback module automatically adjusts the encoding and decoding parameters based on real-time user feedback, environmental noise assessment, and sound quality evaluation to improve sound quality and system efficiency. The optimization feedback module performs feedback and parameter optimization based on a reinforcement learning (RL) algorithm. By collecting the user's sound quality scores, environmental noise, and device characteristics in real time, it dynamically adjusts the encoding and decoding parameters.

[0015] Preferably, the audio signal preprocessing module further includes a time-frequency analysis module based on deep learning. It uses time-frequency domain transformation (such as short-time Fourier transform (STFT) or wavelet transform) to convert the audio signal into a time-frequency representation and uses a convolutional neural network to extract and optimize the time-frequency features to improve the fidelity of the audio signal.

[0016] Preferably, the audio decoding module further includes an adaptive restoration module. By analyzing the difference between the decoded audio and the original audio, it automatically adjusts the parameters during the decoding process to improve the audio quality. The adaptive restoration module models and repairs the audio error based on a deep convolutional network (CNN).

[0017] Preferably, the audio encoding and decoding method is applicable to consumer electronic products such as smart headphones, voice assistants, and smart speakers. The algorithm can optimize audio encoding and decoding according to the frequency response characteristics of different devices to provide a consistent high-fidelity sound quality performance.

[0018] Preferably, the system further includes an audio compression ratio evaluation module for evaluating the balance between the audio compression ratio and sound quality in real time during the encoding process to ensure the best audio quality effect can still be maintained under different bandwidth conditions.

[0019] Compared with the prior art, the present invention provides a high-fidelity audio coding and decoding algorithm and compression optimization method based on AI that adapts to the auditory range, and has the following beneficial effects:

[0020] 1. The high-fidelity audio coding and decoding algorithm and compression optimization method based on AI that adapts to the auditory range can perform dynamic optimization according to individual auditory characteristics through the AI-based auditory range adaptation algorithm, making the audio coding and decoding process more in line with the perceptual characteristics of the human ear. Thus, a better balance is achieved between the compression ratio and the sound quality, providing a higher audio experience. Through adaptive quantization and bitrate control technologies, this method can significantly improve the compression efficiency while maintaining high-fidelity sound quality. At the same time, deep learning algorithms are used to optimize the allocation of computing resources and reduce the consumption of computing resources.

[0021] 2. The high-fidelity audio coding and decoding algorithm and compression optimization method based on AI that adapts to the auditory range can automatically adjust the coding strategy according to the auditory characteristics of different users and changes in environmental noise, thereby providing personalized and flexible audio coding and decoding services in various practical applications. Especially in a noisy environment, it can automatically enhance the clarity of the audio signal. The optimized feedback module can collect and respond to user feedback and changes in environmental noise in real time, ensuring that the system can perform audio processing under low-latency conditions and is suitable for scenarios with high real-time requirements such as voice communication and online meetings.

[0022] 3. The high-fidelity audio coding and decoding algorithm and compression optimization method based on AI that adapts to the auditory range can, through the intelligent adjustment of a deep neural network, improve the audio compression ratio while ensuring low latency, enabling the system to not only provide high-quality audio even in the case of limited bandwidth but also be flexibly adapted among various devices to provide a consistent sound quality performance. Through the optimized decoding strategy, it can adapt to the characteristics of different audio playback devices and ensure the consistency of sound quality. Whether on headphones, speakers, or mobile phones, it can provide the best audio experience. Detailed implementation manners

[0023] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0024] Embodiment

[0025] Embodiment of the high-fidelity audio coding and decoding algorithm and compression optimization method based on AI that adapts to the auditory range

[0026] AI-based High-fidelity Audio Coding and Decoding Algorithm and Compression Optimization Method Adapted to Auditory Range, including:

[0027] An audio signal preprocessing module for preprocessing the input audio signal, including but not limited to denoising, frequency filtering, gain adjustment, time-domain / frequency-domain transformation, etc., to improve the signal-to-noise ratio of the audio signal and remove environmental noise. The preprocessing module extracts features through a convolutional neural network (CNN) or an autoencoder (AE) network to generate optimized audio features;

[0028] An auditory range analysis module that analyzes the audio signal using deep learning algorithms based on the user's personalized auditory data, automatically generates a user's auditory sensitivity model, and dynamically adjusts the coding strategy according to this model to optimize the frequency response and compression parameters of the audio signal. The model can adjust the compression range according to the user's degree of hearing loss (such as high-frequency hearing loss, low-frequency hearing sensitivity);

[0029] An audio coding module that encodes the audio using adaptive quantization and bitrate control algorithms based on the auditory model and compression parameters output by the auditory range analysis module, dynamically adjusts the quantization scale and bitrate during encoding to optimize the balance between compression ratio and sound quality. The coding module optimizes encoding based on transform coding (such as MDCT, DCT) and neural network-based predictive coding (such as autoregressive model, convolutional neural network);

[0030] An audio decoding module that decodes the compressed audio using a deep neural network (DNN) or a convolutional neural network (CNN) decoding network based on the compressed data output by the audio coding module to restore a high-fidelity sound quality as close as possible to the original audio. The decoding module restores the high-frequency or low-frequency loss part through adaptive denoising and compensation techniques;

[0031] An optimization feedback module for dynamically adjusting the parameters in the encoding and decoding processes according to the user's real-time feedback and environmental changes to ensure that the sound quality remains optimal in different environments. The feedback module adjusts the relevant parameters in the auditory range analysis module and the coding module in real time by monitoring information such as environmental noise level and sound quality evaluation.

[0032] Specifically, the audio signal preprocessing module further includes an adaptive noise suppression algorithm for dynamically adjusting the noise suppression intensity of the audio signal according to the environmental noise intensity. The noise suppression algorithm identifies noise based on a deep convolutional network (CNN) and performs noise reduction processing on the audio signal using time-frequency analysis.

[0033] Specifically, the auditory range analysis module trains on the user's hearing test data to establish a personalized auditory model, optimizing the frequency response and compression strategy of the audio. The personalized auditory model is based on self-supervised learning or transfer learning techniques and can dynamically adjust the frequency bandwidth of audio compression according to the user's hearing curve.

[0034] Specifically, the audio encoding module adopts a dynamic encoding strategy based on deep learning and dynamically selects an adaptive encoding mode according to the frequency response adjustment strategy output by the auditory range analysis module. The dynamic encoding strategy includes adaptive quantization, bitrate allocation, dynamic audio feature extraction, etc.

[0035] Specifically, the audio decoding module includes a neural network-based audio restoration technique. By performing deep learning on the high-order features of the compressed audio, it restores a high-fidelity sound quality close to the original audio. The audio restoration technique uses generative adversarial network (GAN) or variational autoencoder (VAE) techniques for audio reconstruction to repair the high-frequency or detailed information lost during the compression process.

[0036] Specifically, the optimization feedback module automatically adjusts the encoding and decoding parameters based on real-time user feedback, environmental noise assessment, and sound quality evaluation to improve sound quality and system efficiency. The optimization feedback module performs feedback and parameter optimization based on a reinforcement learning (RL) algorithm. By collecting the user's sound quality scores, environmental noise, and device characteristics in real time, it dynamically adjusts the encoding and decoding parameters.

[0037] Specifically, the audio signal preprocessing module also includes a deep learning-based time-frequency analysis module. It uses time-frequency domain transformation (such as short-time Fourier transform (STFT) or wavelet transform) to convert the audio signal into a time-frequency representation and uses a convolutional neural network to extract and optimize the time-frequency features to improve the fidelity of the audio signal.

[0038] Specifically, the audio decoding module further includes an adaptive restoration module. By analyzing the difference between the decoded audio and the original audio, it automatically adjusts the parameters during the decoding process to improve the audio quality. The adaptive restoration module models and repairs the audio error based on a deep convolutional network (CNN).

[0039] Specifically, the audio encoding and decoding method is applicable to consumer electronic products such as smart headphones, voice assistants, and smart speakers. The algorithm can optimize audio encoding and decoding according to the frequency response characteristics of different devices to provide a consistent high-fidelity sound quality performance.

[0040] Specifically, the system also includes an audio compression ratio evaluation module for evaluating the balance between the audio compression ratio and sound quality in real time during the encoding process to ensure the best audio quality effect under different bandwidth conditions.

[0041] Through the above technical solutions, in the present invention, through the AI-based auditory range adaptation algorithm, dynamic optimization can be performed according to individual auditory characteristics, making the audio encoding and decoding process more in line with the perceptual characteristics of the human ear, thereby achieving a better balance between the compression ratio and sound quality, providing a higher audio experience. Through the adaptive quantization and bitrate control technologies, this method can significantly improve the compression efficiency while maintaining high-fidelity sound quality. At the same time, the deep learning algorithm is used to optimize the computing resource allocation, reducing the consumption of computing resources. The system can automatically adjust the encoding strategy according to the auditory characteristics of different users and the changes in environmental noise, thus providing personalized and flexible audio encoding and decoding services in various practical applications. Especially in a noisy environment, it can automatically enhance the clarity of the audio signal. The optimization feedback module can collect and respond to the user's feedback and environmental noise changes in real time, ensuring that the system can perform audio processing under low-latency conditions, and is suitable for scenarios with high real-time requirements such as voice communication and online meetings. Through the intelligent adjustment of the deep neural network, while ensuring low latency, the audio compression ratio can be increased, enabling the system to not only provide high-quality audio even under limited bandwidth conditions, but also be flexibly adapted among various devices, providing a consistent sound quality performance. Through the optimized decoding strategy, it can adapt to the characteristics of different audio playback devices, ensuring the consistency of sound quality. Whether it is on headphones, speakers or mobile phones, the best audio experience can be provided.

[0042] Audio signal preprocessing module: This module uses time-frequency analysis technology based on convolutional neural network (CNN) to process the input audio signal in real time, removing background noise and enhancing the target signal. The noise suppression algorithm can be automatically adjusted based on the environmental noise, making the processed audio signal purer.

[0043] Auditory range analysis module: This module is based on deep learning models (such as long short-term memory network LSTM or convolutional neural network CNN), and through training on a large amount of hearing test data of users, an individualized auditory model is extracted. This model will adjust the compression strategy and frequency response range of the audio signal according to factors such as each user's hearing sensitivity, age, and hearing loss. For example, for users with lower frequency sensitivity, the encoding pressure on the high-frequency part can be reduced, thereby improving the compression efficiency.

[0044] Audio encoding module: This module adopts a deep learning-based encoding strategy, combined with adaptive quantization and adaptive bitrate control technologies, to automatically adjust the audio compression parameters. In low-complexity scenarios, it can provide a high compression ratio; while in high-complexity scenarios, it ensures that the audio quality is not lost. During encoding, the network will dynamically select the most suitable encoding configuration according to the output of the auditory range analysis module.

[0045] Audio decoding module: It adopts an audio decoding algorithm based on neural network. By performing deep learning on the features of compressed audio, it can restore high-fidelity quality close to the original audio. During decoding, when the network performs inverse processing on the compressed data, it automatically compensates for the loss of low-frequency or high-frequency parts, thereby improving the sound quality.

[0046] Optimization feedback module: This module optimizes the key parameters in the algorithm in real time by collecting real-time feedback from users (such as sound quality scores, device types, environmental noise, etc.) and monitoring the current audio performance. For example, the frequency response characteristics of the audio playback device and the changes in the current noise environment can all affect the adjustment of the optimization strategy to provide continuous optimal audio effects. Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An AI-based high-fidelity audio codec algorithm and compression optimization method adapted to the auditory range, characterized in that Including: An audio signal preprocessing module for preprocessing the input audio signal, including but not limited to denoising, frequency filtering, gain adjustment, time-domain / frequency-domain transformation, etc., to improve the signal-to-noise ratio of the audio signal and remove environmental noise. The preprocessing module extracts features through a convolutional neural network (CNN) or an autoencoder (AE) network to generate optimized audio features; An auditory range analysis module that analyzes the audio signal using deep learning algorithms based on the user's personalized auditory data, automatically generates a user's auditory sensitivity model, and dynamically adjusts the encoding strategy according to this model to optimize the frequency response and compression parameters of the audio signal. The model can adjust the compression range according to the user's degree of hearing loss (such as high-frequency hearing loss, low-frequency hearing sensitivity); An audio encoding module that encodes the audio using adaptive quantization and bitrate control algorithms based on the auditory model and compression parameters output by the auditory range analysis module, and dynamically adjusts the quantization scale and bitrate during encoding to optimize the balance between the compression ratio and the sound quality. The encoding module performs encoding optimization based on transform coding (such as MDCT, DCT) and neural network-based predictive coding (such as autoregressive model, convolutional neural network); An audio decoding module that decodes the compressed audio using a deep neural network (DNN) or a convolutional neural network (CNN) decoding network based on the compressed data output by the audio encoding module to restore a high-fidelity sound quality as close as possible to the original audio. The decoding module restores the high-frequency or low-frequency loss part through adaptive denoising and compensation techniques; An optimization feedback module for dynamically adjusting the parameters in the encoding and decoding processes according to the user's real-time feedback and environmental changes to ensure that the sound quality remains optimal in different environments. The feedback module adjusts the relevant parameters in the auditory range analysis module and the encoding module in real time by monitoring information such as environmental noise level and sound quality evaluation.

2. The high-fidelity audio codec algorithm and compression optimization method based on AI for adapting to the auditory range according to claim 1, characterized in that: The audio signal preprocessing module further includes an adaptive noise suppression algorithm for dynamically adjusting the noise suppression intensity of the audio signal according to the environmental noise intensity. The noise suppression algorithm performs noise identification based on a deep convolutional network (CNN) and uses time-frequency analysis to denoise the audio signal.

3. The high-fidelity audio codec algorithm and compression optimization method based on AI for adapting to the auditory range according to claim 1, characterized in that: The auditory range analysis module trains on the user's hearing test data to establish a personalized auditory model, optimizing the frequency response and compression strategy of the audio. The personalized auditory model is based on self-supervised learning or transfer learning techniques and can dynamically adjust the frequency bandwidth of audio compression according to the user's hearing curve.

4. The high-fidelity audio codec algorithm and compression optimization method based on AI for adapting to the auditory range according to claim 1, characterized in that: The audio encoding module adopts a deep learning-based dynamic encoding strategy and dynamically selects an adaptive encoding mode according to the frequency response adjustment strategy output by the auditory range analysis module. The dynamic encoding strategy includes adaptive quantization, bitrate allocation, dynamic audio feature extraction, etc.

5. The high-fidelity audio codec algorithm and compression optimization method based on AI for adapting to the auditory range according to claim 1, characterized in that: The audio decoding module includes a neural network-based audio restoration technology. By performing deep learning on the high-order features of the compressed audio, it restores a high-fidelity sound quality close to the original audio. The audio restoration technology uses generative adversarial network (GAN) or variational autoencoder (VAE) technology for audio reconstruction to repair the high-frequency or detailed information lost during the compression process.

6. The high-fidelity audio codec algorithm and compression optimization method based on AI for adapting to the auditory range according to claim 1, characterized in that: The optimization feedback module automatically adjusts the encoding and decoding parameters based on real-time user feedback, environmental noise assessment, and sound quality evaluation to improve sound quality and system efficiency. The optimization feedback module performs feedback and parameter optimization based on the reinforcement learning (RL) algorithm. By collecting real-time user sound quality scores, environmental noise, and device characteristics, it dynamically adjusts the encoding and decoding parameters.

7. The high-fidelity audio codec algorithm and compression optimization method based on AI for adapting to the auditory range according to claim 1, characterized in that: The audio signal preprocessing module further includes a deep learning-based time-frequency analysis module. It uses time-frequency domain transformation (such as short-time Fourier transform (STFT) or wavelet transform) to convert the audio signal into a time-frequency representation and uses a convolutional neural network to extract and optimize the time-frequency features to improve the fidelity of the audio signal.

8. The high-fidelity audio codec algorithm and compression optimization method based on AI for adapting to the auditory range according to claim 1, characterized in that: The audio decoding module further includes an adaptive restoration module. By analyzing the difference between the decoded audio and the original audio, it automatically adjusts the parameters during the decoding process to improve the audio quality. The adaptive restoration module models and repairs the audio error based on a deep convolutional network (CNN).

9. The high-fidelity audio codec algorithm and compression optimization method based on AI for adapting the auditory range according to claim 1, characterized in that: The audio encoding and decoding method is applicable to consumer electronic products such as smart headphones, voice assistants, and smart speakers. The algorithm can optimize audio encoding and decoding according to the frequency response characteristics of different devices to provide a consistent high-fidelity sound quality performance.

10. The high-fidelity audio codec algorithm and compression optimization method based on AI for adapting to the auditory range according to claim 1, characterized in that: The system further includes an audio compression ratio evaluation module, which is used to evaluate the balance between the audio compression ratio and the sound quality in real time during the encoding process to ensure that the best sound quality effect can still be maintained under different bandwidth conditions.