Broadband Decorrelation Filtering for Multichannel Audio Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio codecs like xHE-AAC and AMR-WB+ face issues with unnatural coloration and artifacts in speech signals due to reverberant processing, inconsistent output at different sampling rates, and inefficiencies in handling weakly correlated channels and out-of-phase signals, leading to degraded audio quality and bitrate inefficiencies.
Innovation Solution
A mixed approach using a decorrelation filter to generate a filling signal in the time domain, combined with a multichannel processor applying narrow band processing in the spectral domain, to reconstruct missing residual components in encoded multichannel signals, particularly for speech signals, ensuring high audio quality while maintaining low bitrates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of substance
If a fully parametric coding method is used to achieve bitrate efficiency, then bitrate is reduced, but audio quality degrades due to spectral smearing and artifacts in speech signals
Solution Approach 1:
The patent segments the audio signal processing into distinct stages: downmix signal processing, filling signal generation through decorrelation filtering, and spectral domain reconstruction. This segmentation allows parametric coding to operate on the downmix while preserving speech characteristics through the filling signal, resolving the contradiction between bitrate efficiency and audio quality
Solution Approach 2:
The filling signal acts as an intermediary between the downmix signal and the final stereo output. Generated through decorrelation filtering, this intermediate signal provides the missing spatial information without the spectral smearing of traditional parametric methods, enabling both low bitrate and high quality
2Adaptability or versatility
If subband filters are used to generate the synthetic second signal, then stereo parameters can be estimated, but the spectral shape of the input signal is heavily smeared over time
Solution Approach 1:
The patent replaces the mechanical subband filtering system with a decorrelation filter based on allpass filters. This substitution maintains the ability to generate stereo parameters while avoiding the temporal smearing inherent in subband approaches, preserving spectral shape through the allpass filter's phase manipulation without magnitude filtering
3Loss of substance
If a low order FIR filter is used to predict the side signal from the mid signal, then bitrate is reduced, but the method fails for weakly correlated channels and out-of-phase signals
Solution Approach 1:
The patent changes the fundamental parameter of the prediction filter from a low order FIR filter to a decorrelation filter based on allpass filters. This parameter change enables the system to handle weakly correlated and out-of-phase signals by generating a filling signal that preserves spatial relationships, while still maintaining bitrate efficiency through parametric coding of the downmix
Data Source
AI summary
An apparatus for decoding an encoded multichannel signal includes: a base channel decoder for decoding an encoded base channel to acquire a decoded base channel; a decorrelation filter for filtering at least a portion of the decoded base channel to acquire a filling signal; and a multichannel processor for performing a multichannel processing using a spectral representation of the decoded base channel and a spectral representation of the filling signal, wherein the decorrelation filter is a broad band filter and the multichannel processor is configured to apply a narrow band processing to the spectral representation of the decoded base channel and the spectral representation of the filling signal.