Recorded Audio Noise Reduction via Background Reference Separation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing noise reduction algorithms struggle to effectively separate music signals from noise in audio recordings, often misinterpreting music as noise or vice versa, leading to unsatisfactory noise reduction effects.
Innovation Solution
An audio data processing method that involves acquiring recorded audio, determining a reference audio from a database, and using deep network models to filter and separate voice components from background and noise components, converting noise reduction into noise reduction for processed voice audio to avoid confusion between background and noise components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If conventional noise reduction algorithms are used on recorded audio, then noise suppression is attempted, but music signals are misinterpreted as noise or vice versa, resulting in poor noise reduction effect
Solution Approach 1:
The recorded audio is segmented into three distinct components: background audio (music), voice audio, and noise. This segmentation is achieved by first extracting the background component through matching with reference audio from a database, then isolating the voice component, leaving the noise as the residual. This precise segmentation prevents misclassification of music as noise or vice versa, directly resolving the technical contradiction between noise suppression effectiveness and signal separation accuracy.
2Object-affected harmful factors
If statistical or deep learning noise reduction methods are applied to music recording signals, then noise suppression is performed, but the algorithms cannot distinguish between music and noise in similar frequency spectrums, leading to misinterpretation and unsatisfactory results
Solution Approach 1:
Reference audio from a database serves as an intermediary to bridge the gap between the recorded audio and the noise reduction process. By matching the background component of recorded audio with reference audio, the system can accurately identify and separate music from noise without relying on algorithms that may misinterpret similar frequency spectrums. This intermediary approach enhances both noise reduction capability and adaptability to different music types.
3Device complexity
If direct noise reduction is applied to recorded audio containing both music and voice, then processing is simplified, but the background music and noise components become confused, reducing noise reduction quality
Solution Approach 1:
The audio signal is divided into distinct components (background music, voice, noise) before noise reduction processing. Although this segmentation adds initial processing steps, it prevents the confusion between music and noise during noise reduction, ultimately improving noise reduction quality while maintaining reasonable processing complexity through systematic component separation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An audio data processing method and apparatus, a device and a medium. The method comprises: acquiring recorded audio, wherein the recorded audio comprises a background reference audio component, a voice audio component and an ambient noise component (S101); acquiring, from an audio database, prototype audio that matches the recorded audio (S102); acquiring candidate voice audio from the recorded audio according to the prototype audio, wherein the candidate voice audio comprises the voice audio component and the ambient noise component (S103); determining the difference between the recorded audio and the candidate voice audio to be the background reference audio component contained in the recorded audio (S104); and performing ambient noise reduction processing on the candidate voice audio, so as to obtain noise-reduced voice audio corresponding to the candidate voice audio, and merging the noise-reduced voice audio with the background reference audio component, so as to obtain noise-reduced recorded audio (S105). By means of the method, the noise reduction effect of recorded audio can be improved.