An embedded audio processing system and method supporting AI language enhancement

Abstract

The application relates to the technical field of audio processing technology, in particular to an embedded audio processing system and method supporting AI language enhancement, which comprises a data acquisition module, an audio analysis module, an AI enhancement module, an audio processing module and an interactive output module. The modules of the system cooperate, the data acquisition module collects audio, the analysis module accurately classifies and calibrates, the enhancement module improves the quality of fuzzy audio, the processing module generates corresponding content, the output module outputs, accurate, efficient and high-quality voice interaction is realized, the speed of vehicle-mounted AI voice answering voice questions is improved, and the accuracy of vehicle-mounted AI voice answering voice questions is improved.

Description

Technical Field

[0001] This invention relates to the field of audio processing technology, and in particular to an embedded audio processing system and method that supports AI language enhancement. Background Technology

[0002] In-vehicle voice control systems need to cope with complex environments such as driving noise and interference from multiple sound sources, often resulting in problems such as inaccurate command recognition and lag in interaction. Traditional audio processing technologies struggle to balance real-time performance and enhancement effects, failing to meet the demands for efficient voice control while driving. Therefore, an in-vehicle voice system that can quickly answer voice questions and boasts high accuracy is needed to address these issues.

[0003] Chinese Patent Publication No. CN117809636A discloses a method and system for enhancing and recognizing target speech in a multi-speaker environment. The method includes designing a frequency-domain-based array flow matrix based on a surface matrix array simulating multiple sensors within the vehicle scene; designing constraints based on the positions of the sensors within the vehicle and using an improved frequency-domain-based constrained minimum variance algorithm to enhance the target direction speech signal; designing a generative adversarial network (GAN) and inputting the processed target direction speech signal into the GAN to generate the enhanced speech signal; and training the GAN using multiple language datasets to improve the model's generalization ability. This invention solves the speech interference problem in multi-speaker environments, providing high-quality enhanced speech, reducing distortion of the target speaker's speech, and significantly improving the accuracy of speech recognition. However, this solution cannot address the problem of slow response speed and low accuracy of in-vehicle AI voice responses due to unclear received speech. Summary of the Invention

[0004] To address this issue, the present invention provides an embedded audio processing system and method that supports AI language enhancement, thereby overcoming the problems in the prior art where in-vehicle AI voice responses to voice questions are slow and have low accuracy due to unclear received voice.

[0005] To achieve the above objectives, in one aspect, the present invention provides an embedded audio processing system supporting AI language enhancement, comprising:

[0006] The data acquisition module is used to acquire audio data;

[0007] The audio analysis module is used to classify audio based on sensor state coefficients in the audio data to obtain classified audio, which includes clear audio and ambiguous audio. It is also used to adjust the audio classification process based on user accuracy feedback coefficients in the audio data, and to calibrate the audio classification adjustment process based on driving optimization coefficients in the audio data.

[0008] The AI ​​enhancement module is used to enhance fuzzy audio based on the audio enhancement student model to obtain enhanced audio. It is also used to correct the enhancement process of fuzzy audio based on the bounce user accurate feedback coefficient in the audio data, and to optimize the enhancement process of fuzzy audio based on the number of sound sources generated in the audio data.

[0009] The audio processing module is used to generate first interactive content based on a first audio interaction model and clear audio, and also to generate second interactive content based on a second audio interaction model and enhanced audio.

[0010] The interactive output module is used to output the first and second interactive content as voice interactive content.

[0011] Furthermore, the audio analysis module compares the sensor state coefficient z with the preset sensor state coefficient z0, judges the audio state based on the comparison result, and classifies the audio based on the judgment result, wherein:

[0012] When z≤z0, the audio analysis module determines that the audio state is clear, classifies the audio, and outputs the clear audio as the classified audio.

[0013] When z > z0, the audio analysis module determines that the audio state is ambiguous, classifies the audio, treats the audio as ambiguous audio, and outputs the ambiguous audio as the classified audio.

[0014] Furthermore, the audio analysis module compares the user's accurate feedback coefficient A with the preset user accurate feedback coefficient A0, judges the accuracy of the audio based on the comparison result, and adjusts the audio classification process based on the judgment result, wherein:

[0015] When A > A0, the audio analysis module determines that the audio is accurate and does not adjust the audio classification process.

[0016] When A ≤ A0, the audio analysis module determines the audio accuracy to be inaccurate, adjusts the audio classification process, and adjusts the preset sensor state coefficient z0 according to the adjustment coefficient tz, setting tz = 0.65 + 0.35 × e -(A0-A) Let e ​​be the base of the natural logarithm. We obtain the adjusted preset sensor state coefficient z0j. We set z0j = tz × z0. We replace the preset sensor state coefficient z0 with the adjusted preset sensor state coefficient z0j, and then compare the sensor state coefficient z with the preset sensor state coefficient z0 again.

[0017] Furthermore, the audio analysis module compares the driving optimization coefficient B with the preset driving optimization coefficient B0, judges the driving status based on the comparison result, and calibrates the audio classification adjustment process based on the judgment result, wherein:

[0018] When B < B0, the audio analysis module determines that the driving state is quiet and does not calibrate the audio classification adjustment process;

[0019] When B≥B0, the audio analysis module determines the driving state to be noisy, calibrates the audio classification adjustment process, cancels the adjustment process, and calibrates the preset sensor state coefficient z0 according to the calibration coefficient jz, setting jz=0.55+0.25×e -(B-B0) e is the base of the natural logarithm. The calibrated preset sensor state coefficient z0r is obtained. z0r is set to jz×z0. The preset sensor state coefficient z0 is replaced with the calibrated preset sensor state coefficient z0r. The sensor state coefficient z is then compared with the preset sensor state coefficient z0 again.

[0020] Furthermore, the AI ​​enhancement module inputs the fuzzy audio into the audio enhancement student model, obtains the enhancement scheme output by the audio enhancement student model, and enhances the fuzzy audio according to the enhancement scheme to obtain the enhanced audio. The audio enhancement scheme includes an algorithm optimization scheme, a power consumption optimization scheme, and a hardware optimization scheme.

[0021] Furthermore, the AI ​​enhancement module compares the rebound user accurate feedback coefficient RA with the preset rebound user accurate feedback coefficient RA0, judges the enhancement effect based on the comparison result, and corrects the enhancement process of the blurred audio based on the judgment result, wherein:

[0022] When RA > RA0, the AI ​​enhancement module determines that the enhancement effect is effective and does not correct the enhancement process of the blurred audio.

[0023] When RA≤RA0, the AI ​​enhancement module determines that the enhancement effect is invalid, corrects the enhancement process of the blurred audio, corrects the algorithm optimization scheme, obtains the corrected algorithm optimization scheme, and enhances the blurred audio according to the corrected algorithm optimization scheme.

[0024] Furthermore, the AI ​​enhancement module compares the number N of sound sources generated in the audio data with the preset number N0 of sound sources generated, judges the noise level of the sound sources based on the comparison result, and optimizes the enhancement process of the blurred audio based on the judgment result, wherein:

[0025] When N < N0, the AI ​​enhancement module determines that the sound source is not noisy and does not optimize the enhancement process of the ambiguous audio.

[0026] When N≥N0, the AI ​​enhancement module determines that the sound source is noisy, optimizes the enhancement process of the fuzzy audio, optimizes the hardware optimization scheme, obtains the optimized algorithm optimization scheme, and enhances the fuzzy audio according to the optimized algorithm optimization scheme.

[0027] Furthermore, the audio processing module inputs clear audio into the first audio interaction model to obtain the first interactive content output by the first audio interaction model, and also inputs enhanced audio into the second audio interaction model to obtain the second interactive content output by the second audio interaction model, and sends the first interactive content and the second interactive content to the interaction output module.

[0028] Furthermore, the interactive output module sends the first interactive content and the second interactive content as voice interactive content to the vehicle-mounted interactive device, which then outputs the voice interactive content.

[0029] On the other hand, the present invention also provides an embedded audio processing method supporting AI language enhancement, the method comprising:

[0030] Step S1: Acquire audio data;

[0031] Step S2: Classify the audio according to the sensor state coefficient in the audio data to obtain classified audio. The classified audio includes clear audio and blurry audio. The audio classification process is also adjusted according to the user accurate feedback coefficient in the audio data, and the audio classification adjustment process is calibrated according to the driving optimization coefficient in the audio data.

[0032] Step S3: Enhance the fuzzy audio according to the audio enhancement student model to obtain the enhanced audio. Also, correct the enhancement process of the fuzzy audio according to the rebound user accurate feedback coefficient in the audio data, and optimize the enhancement process of the fuzzy audio according to the number of sound sources generated in the audio data.

[0033] Step S4: Generate the first interactive content based on the first audio interaction model and the clear audio, and generate the second interactive content based on the second audio interaction model and the enhanced audio.

[0034] Step S5: Output the first and second interactive content as voice interactive content.

[0035] Compared with existing technologies, the beneficial effects of this invention are that the various modules of the system work together to achieve good results. The data acquisition module is responsible for collecting audio data, providing a foundation for subsequent processing; the audio analysis module can classify audio into clear audio and fuzzy audio based on the sensor state coefficients in the audio data, and also adjusts the classification process by combining the accurate feedback coefficients of users, and calibrates this adjustment process through driving optimization coefficients to make the classification more in line with actual needs; the AI ​​enhancement module uses an audio enhancement student model to enhance fuzzy audio, and at the same time corrects the enhancement process based on the accurate feedback coefficients of users, and optimizes the enhancement effect by combining the number of sound sources generated, which greatly improves the quality of fuzzy audio; the audio processing module generates the first interactive content based on the first audio interaction model and clear audio, and generates the second interactive content based on the second audio interaction model and enhanced audio, ensuring the relevance of the interactive content; the interactive output module outputs the two types of interactive content as voice interactive content, ultimately achieving accurate, efficient and high-quality voice interaction, and improving the overall user experience. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the structure of an embedded audio processing system that supports AI language enhancement, as shown in this embodiment.

[0037] Figure 2 This is a flowchart illustrating the embedded audio processing method based on AI language enhancement in this embodiment. Detailed Implementation

[0038] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0039] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0040] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0041] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0042] Please see Figure 1 As shown, this is a schematic diagram of the structure of an embedded audio processing system supporting AI language enhancement in this embodiment. The system includes:

[0043] The data acquisition module is used to acquire audio data;

[0044] The audio analysis module is used to classify audio based on sensor state coefficients in the audio data to obtain classified audio, which includes clear audio and ambiguous audio. It is also used to adjust the audio classification process based on user accuracy feedback coefficients in the audio data, and to calibrate the audio classification adjustment process based on driving optimization coefficients in the audio data.

[0045] The AI ​​enhancement module is used to enhance fuzzy audio based on the audio enhancement student model to obtain enhanced audio. It is also used to correct the enhancement process of fuzzy audio based on the bounce user accurate feedback coefficient in the audio data, and to optimize the enhancement process of fuzzy audio based on the number of sound sources generated in the audio data.

[0046] The audio processing module is used to generate first interactive content based on a first audio interaction model and clear audio, and also to generate second interactive content based on a second audio interaction model and enhanced audio.

[0047] The interactive output module is used to output the first and second interactive content as voice interactive content.

[0048] Specifically, the system's modules work collaboratively to achieve excellent results. The data acquisition module collects audio data, providing a foundation for subsequent processing. The audio analysis module categorizes audio into clear and blurry audio based on sensor state coefficients in the audio data. It also adjusts the classification process based on accurate user feedback coefficients and calibrates this process using driving optimization coefficients to make the classification more aligned with actual needs. The AI ​​enhancement module enhances blurry audio using an audio enhancement student model, while simultaneously correcting the enhancement process based on accurate user feedback coefficients and optimizing the enhancement effect by considering the number of sound sources, significantly improving the quality of blurry audio. The audio processing module generates first interactive content based on the first audio interaction model and clear audio, and second interactive content based on the second audio interaction model and enhanced audio, ensuring the relevance of the interactive content. The interaction output module outputs both types of interactive content as voice interaction content, ultimately achieving accurate, efficient, and high-quality voice interaction and improving the overall user experience.

[0049] Specifically, the audio analysis module compares the sensor state coefficient z with the preset sensor state coefficient z0, judges the audio state based on the comparison result, and classifies the audio based on the judgment result, wherein:

[0050] When z≤z0, the audio analysis module determines that the audio state is clear, classifies the audio, and outputs the clear audio as the classified audio.

[0051] When z > z0, the audio analysis module determines that the audio state is ambiguous, classifies the audio, treats the audio as ambiguous audio, and outputs the ambiguous audio as the classified audio.

[0052] Specifically, the audio analysis module compares the user's accurate feedback coefficient A with a preset user accurate feedback coefficient A0, judges the accuracy of the audio based on the comparison result, and adjusts the audio classification process based on the judgment result, wherein:

[0053] When A > A0, the audio analysis module determines that the audio is accurate and does not adjust the audio classification process.

[0054] When A ≤ A0, the audio analysis module determines the audio accuracy to be inaccurate, adjusts the audio classification process, and adjusts the preset sensor state coefficient z0 according to the adjustment coefficient tz, setting tz = 0.65 + 0.35 × e -(A0-A)Let e ​​be the base of the natural logarithm. We obtain the adjusted preset sensor state coefficient z0j. We set z0j = tz × z0. We replace the preset sensor state coefficient z0 with the adjusted preset sensor state coefficient z0j, and then compare the sensor state coefficient z with the preset sensor state coefficient z0 again.

[0055] Specifically, the audio analysis module compares the driving optimization coefficient B with the preset driving optimization coefficient B0, judges the driving status based on the comparison result, and calibrates the audio classification adjustment process based on the judgment result, wherein:

[0056] When B < B0, the audio analysis module determines that the driving state is quiet and does not calibrate the audio classification adjustment process;

[0057] When B≥B0, the audio analysis module determines the driving state to be noisy, calibrates the audio classification adjustment process, cancels the adjustment process, and calibrates the preset sensor state coefficient z0 according to the calibration coefficient jz, setting jz=0.55+0.25×e -(B-B0) e is the base of the natural logarithm. The calibrated preset sensor state coefficient z0r is obtained. z0r is set to jz×z0. The preset sensor state coefficient z0 is replaced with the calibrated preset sensor state coefficient z0r. The sensor state coefficient z is then compared with the preset sensor state coefficient z0 again.

[0058] Specifically, the AI ​​enhancement module inputs the fuzzy audio into the audio enhancement student model, obtains the enhancement scheme output by the audio enhancement student model, and enhances the fuzzy audio according to the enhancement scheme to obtain the enhanced audio. The audio enhancement scheme includes an algorithm optimization scheme, a power consumption optimization scheme, and a hardware optimization scheme.

[0059] Specifically, the AI ​​enhancement module compares the rebound user accurate feedback coefficient RA with the preset rebound user accurate feedback coefficient RA0, judges the enhancement effect based on the comparison result, and corrects the enhancement process of the blurred audio based on the judgment result, wherein:

[0060] When RA > RA0, the AI ​​enhancement module determines that the enhancement effect is effective and does not correct the enhancement process of the blurred audio.

[0061] When RA≤RA0, the AI ​​enhancement module determines that the enhancement effect is invalid, corrects the enhancement process of the blurred audio, corrects the algorithm optimization scheme, obtains the corrected algorithm optimization scheme, and enhances the blurred audio according to the corrected algorithm optimization scheme.

[0062] Specifically, the AI ​​enhancement module compares the number N of sound sources generated in the audio data with the preset number N0 of sound sources generated, judges the noise level of the sound sources based on the comparison result, and optimizes the enhancement process of the blurred audio based on the judgment result, wherein:

[0063] When N < N0, the AI ​​enhancement module determines that the sound source is not noisy and does not optimize the enhancement process of the ambiguous audio.

[0064] When N≥N0, the AI ​​enhancement module determines that the sound source is noisy, optimizes the enhancement process of the fuzzy audio, optimizes the hardware optimization scheme, obtains the optimized algorithm optimization scheme, and enhances the fuzzy audio according to the optimized algorithm optimization scheme.

[0065] Specifically, the audio processing module inputs clear audio into the first audio interaction model to obtain the first interactive content output by the first audio interaction model, and also inputs enhanced audio into the second audio interaction model to obtain the second interactive content output by the second audio interaction model, and sends the first interactive content and the second interactive content to the interaction output module.

[0066] Specifically, the interactive output module sends the first and second interactive content as voice interactive content to the vehicle-mounted interactive device, which then outputs the voice interactive content.

[0067] Please see Figure 2 As shown, this is a flowchart illustrating the embedded audio processing method based on AI language enhancement in this embodiment. The method includes:

[0068] Step S1: Acquire audio data;

[0069] Step S2: Classify the audio according to the sensor state coefficient in the audio data to obtain classified audio. The classified audio includes clear audio and blurry audio. The audio classification process is also adjusted according to the user accurate feedback coefficient in the audio data, and the audio classification adjustment process is calibrated according to the driving optimization coefficient in the audio data.

[0070] Step S3: Enhance the fuzzy audio according to the audio enhancement student model to obtain the enhanced audio. Also, correct the enhancement process of the fuzzy audio according to the rebound user accurate feedback coefficient in the audio data, and optimize the enhancement process of the fuzzy audio according to the number of sound sources generated in the audio data.

[0071] Step S4: Generate the first interactive content based on the first audio interaction model and the clear audio, and generate the second interactive content based on the second audio interaction model and the enhanced audio.

[0072] Step S5: Output the first and second interactive content as voice interactive content.

[0073] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. An embedded audio processing system supporting AI language enhancement, characterized in that, include: The data acquisition module is used to acquire audio data; The audio analysis module is used to classify audio based on sensor state coefficients in the audio data to obtain classified audio, which includes clear audio and ambiguous audio. It is also used to adjust the audio classification process based on user accuracy feedback coefficients in the audio data, and to calibrate the audio classification adjustment process based on driving optimization coefficients in the audio data. The AI ​​enhancement module is used to enhance fuzzy audio based on the audio enhancement student model to obtain enhanced audio. It is also used to correct the enhancement process of fuzzy audio based on the bounce user accurate feedback coefficient in the audio data, and to optimize the enhancement process of fuzzy audio based on the number of sound sources generated in the audio data. The audio processing module is used to generate first interactive content based on a first audio interaction model and clear audio, and also to generate second interactive content based on a second audio interaction model and enhanced audio. The interactive output module is used to output the first and second interactive content as voice interactive content; The audio analysis module compares the sensor state coefficient z with the preset sensor state coefficient z0, judges the audio state based on the comparison result, and classifies the audio based on the judgment result, wherein: When z≤z0, the audio analysis module determines that the audio state is clear, classifies the audio, and outputs the clear audio as the classified audio. When z > z0, the audio analysis module determines that the audio state is ambiguous, classifies the audio, treats the audio as ambiguous audio, and outputs the ambiguous audio as the classified audio. The audio analysis module compares the driving optimization coefficient B with the preset driving optimization coefficient B0, judges the driving status based on the comparison result, and calibrates the audio classification adjustment process based on the judgment result, wherein: When B < B0, the audio analysis module determines that the driving state is quiet and does not calibrate the audio classification adjustment process; When B≥B0, the audio analysis module determines the driving state to be noisy, calibrates the audio classification adjustment process, cancels the adjustment process, and calibrates the preset sensor state coefficient z0 according to the calibration coefficient jz, setting jz=0.55+0.25×e -(B-B0) e is the base of the natural logarithm. The calibrated preset sensor state coefficient z0r is obtained. z0r is set to jz×z0. The preset sensor state coefficient z0 is replaced with the calibrated preset sensor state coefficient z0r. The sensor state coefficient z is then compared with the preset sensor state coefficient z0 again.

2. The embedded audio processing system supporting AI language enhancement according to claim 1, characterized in that, The audio analysis module compares the user's accurate feedback coefficient A with the preset user accurate feedback coefficient A0, judges the accuracy of the audio based on the comparison result, and adjusts the audio classification process based on the judgment result, wherein: When A > A0, the audio analysis module determines that the audio is accurate and does not adjust the audio classification process. When A ≤ A0, the audio analysis module determines the audio accuracy to be inaccurate, adjusts the audio classification process, and adjusts the preset sensor state coefficient z0 according to the adjustment coefficient tz, setting tz = 0.65 + 0.35 × e -(A0-A) Let e ​​be the base of the natural logarithm. We obtain the adjusted preset sensor state coefficient z0j. We set z0j = tz × z0. We replace the preset sensor state coefficient z0 with the adjusted preset sensor state coefficient z0j, and then compare the sensor state coefficient z with the preset sensor state coefficient z0 again.

3. The embedded audio processing system supporting AI language enhancement according to claim 2, characterized in that, The AI ​​enhancement module inputs the fuzzy audio into the audio enhancement student model, obtains the enhancement scheme output by the audio enhancement student model, and enhances the fuzzy audio according to the enhancement scheme to obtain the enhanced audio. The audio enhancement scheme includes an algorithm optimization scheme, a power consumption optimization scheme, and a hardware optimization scheme.

4. The embedded audio processing system supporting AI language enhancement according to claim 3, characterized in that, The AI ​​enhancement module compares the accurate feedback coefficient RA of the user with the preset accurate feedback coefficient RA0 of the user, judges the enhancement effect based on the comparison result, and corrects the enhancement process of the blurred audio based on the judgment result, wherein: When RA > RA0, the AI ​​enhancement module determines that the enhancement effect is effective and does not correct the enhancement process of the blurred audio. When RA≤RA0, the AI ​​enhancement module determines that the enhancement effect is invalid, corrects the enhancement process of the blurred audio, corrects the algorithm optimization scheme, obtains the corrected algorithm optimization scheme, and enhances the blurred audio according to the corrected algorithm optimization scheme.

5. The embedded audio processing system supporting AI language enhancement according to claim 4, characterized in that, The AI ​​enhancement module compares the number N of sound sources generated in the audio data with the preset number N0 of sound sources generated. Based on the comparison result, it judges the noise level of the sound sources and optimizes the enhancement process for the blurred audio based on the judgment result. When N < N0, the AI ​​enhancement module determines that the sound source is not noisy and does not optimize the enhancement process of the ambiguous audio. When N≥N0, the AI ​​enhancement module determines that the sound source is noisy, optimizes the enhancement process of the fuzzy audio, optimizes the hardware optimization scheme, obtains the optimized algorithm optimization scheme, and enhances the fuzzy audio according to the optimized algorithm optimization scheme.

6. The embedded audio processing system supporting AI language enhancement according to claim 5, characterized in that, The audio processing module inputs clear audio into the first audio interaction model, obtains the first interactive content output by the first audio interaction model, and inputs enhanced audio into the second audio interaction model, obtains the second interactive content output by the second audio interaction model, and sends the first interactive content and the second interactive content to the interaction output module.

7. The embedded audio processing system supporting AI language enhancement according to claim 6, characterized in that, The interactive output module sends the first and second interactive content as voice interactive content to the vehicle-mounted interactive device, which then outputs the voice interactive content.

8. An embedded audio processing method supporting AI language enhancement, applied to an embedded audio processing system supporting AI language enhancement as described in any one of claims 1 to 7, characterized in that, The method includes: Step S1: Acquire audio data; Step S2: Classify the audio according to the sensor state coefficient in the audio data to obtain classified audio. The classified audio includes clear audio and blurry audio. The audio classification process is also adjusted according to the user accurate feedback coefficient in the audio data, and the audio classification adjustment process is calibrated according to the driving optimization coefficient in the audio data. Step S3: Enhance the fuzzy audio according to the audio enhancement student model to obtain the enhanced audio. Also, correct the enhancement process of the fuzzy audio according to the rebound user accurate feedback coefficient in the audio data, and optimize the enhancement process of the fuzzy audio according to the number of sound sources generated in the audio data. Step S4: Generate the first interactive content based on the first audio interaction model and the clear audio, and generate the second interactive content based on the second audio interaction model and the enhanced audio. Step S5: Output the first and second interactive content as voice interactive content.

Images