Direct-Diffuse Audio Bandwidth Extension for Artifact Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for enhancing compressed audio signals, such as bandwidth extension and equalization, fail to effectively restore original audio quality due to difficulties in detecting high-frequency cut-off frequencies and addressing 'holes' in the spectrogram that create audible artifacts and loss of sound attacks.
Innovation Solution
An audio signal processing apparatus that decomposes input signals into direct and diffuse components, with a bandwidth extender and combiner to enhance the direct audio signal by extending its frequency range and adding reverberation, while maintaining the natural ambient sound in the diffuse signal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If bandwidth extension techniques are used to regenerate high frequency components, then the perceived audio quality is improved, but the detection of the actual high-frequency cut-off frequency becomes a difficult task and may introduce errors
Solution Approach 1:
The patent introduces an intermediary approach by using the diffuse audio signal as a reference to infer the cut-off frequency characteristics for the direct audio signal. Instead of directly detecting the cut-off frequency in the direct signal (which is difficult), the system uses the easier-to-detect diffuse signal as a mediator to estimate the cut-off frequency, thereby improving detection accuracy without significantly increasing system complexity.
2Manufacturing precision
If conventional bandwidth extension methods are applied to the whole audio signal, then high frequency components are restored, but audible artifacts such as musical noise and birdie sounds are not addressed
Solution Approach 1:
The patent segments the audio signal into two distinct components: direct audio signal and diffuse audio signal. By separating these components, the system can apply different processing strategies to each. The direct signal undergoes bandwidth extension with artifact suppression, while the diffuse signal serves as a reference and is combined back to restore natural ambient sound, thereby reducing audible artifacts while maintaining audio quality.
Solution Approach 2:
The patent extracts the diffuse audio signal component from the overall audio signal and uses it as a reference for processing the direct audio signal. This extraction allows the system to identify and suppress artifacts in the direct signal bandwidth extension while preserving the natural characteristics captured in the diffuse signal, thereby reducing harmful artifacts like musical noise and birdie sounds.
3Measurement precision
If equalization or bass maximization techniques are used to improve perceived quality, then general sound effects are enhanced, but the loss of transients and attacks is not addressed
Solution Approach 1:
The patent applies local quality enhancement by treating the direct and diffuse audio signals differently. The direct audio signal, which contains transient and attack information, undergoes specific bandwidth extension processing with parameters optimized for preserving transient characteristics. The diffuse signal, containing ambient reverberation, is processed separately and combined to restore overall quality. This localized approach ensures transients and attacks are preserved while still achieving perceptual quality enhancement.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates to an audio signal processing apparatus (100) and method for processing an input audio signal (101) into an output audio signal (103). The audio signal processing apparatus (100) comprises a decomposer (105) configured to decompose the input audio signal (101) into a direct audio signal (102a) and a diffuse audio signal (102b), a modifier (107) configured to modify the direct audio signal (102a) in order to obtain a modified direct audio signal (102a'), wherein the modifier (107) comprises a bandwidth extender (107a) configured to extend an upper cutoff frequency of the frequency range of the direct audio signal (102a), and a combiner (109) configured to combine the modified direct audio signal (102a') with the diffuse audio signal (102b) in order to obtain the output audio signal (103).