Perceptual Audio Coding with Adaptive Non-Uniform Tiling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current audio coding technologies face challenges in efficiently managing quantization noise while maintaining low audible artifacts, particularly due to the non-uniform time/frequency resolution of the human auditory system, which is not effectively addressed by uniform filterbanks.
Innovation Solution
The implementation of a non-uniform filterbank with a cascaded lapped critically sampled transform and time domain aliasing reduction (TDAR) stage, allowing for adaptive control of subband merging and joint channel processing to optimize coding efficiency and reduce aliasing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a uniform filterbank (MDCT) is used, then the filterbank structure is simple and uniform, but it cannot adapt to the non-uniform time/frequency resolution requirements of the human auditory system
Solution Approach 1:
The filterbank is segmented into multiple subbands with different time-frequency resolutions. Each subband can be independently configured to match the characteristics of different frequency regions of the human auditory system, allowing adaptation without requiring a completely complex redesign of the entire filterbank structure.
Solution Approach 2:
Different subbands are assigned different time-frequency resolution characteristics according to the local requirements of the human auditory system. Low-frequency subbands use longer windows for better frequency resolution, while high-frequency subbands use shorter windows for better time resolution, implementing local optimization throughout the frequency spectrum.
2Productivity
If more quantization noise is introduced to reduce bitrate, then coding efficiency improves, but audible artifacts increase
Solution Approach 1:
The filterbank applies different quantization strategies to different subbands based on the local masking thresholds of the human auditory system. In frequency regions where the auditory system is less sensitive, higher quantization noise can be introduced without creating audible artifacts, thereby improving overall coding efficiency while maintaining perceptual quality.
Solution Approach 2:
The system dynamically adjusts quantization parameters for each subband according to the local audio content and masking thresholds. By changing quantization step sizes and precision levels adaptively across different frequency bands, the system maximizes bitrate reduction while keeping quantization noise below perceptible thresholds.
3Manufacturing precision
If a non-uniform filterbank is implemented to match human auditory characteristics, then perceptual quality improves, but the complexity of controlling multiple parameters increases
Solution Approach 1:
The complex parameter control space is segmented into independent subband configurations. Each subband has its own set of parameters (window length, overlap, quantization precision) that can be optimized independently, reducing the overall complexity of parameter control while achieving high perceptual quality through localized optimization.
Solution Approach 2:
The filterbank parameters are made dynamic and adaptive rather than fixed. The system automatically adjusts time-frequency resolution and quantization parameters for each subband based on the local characteristics of the audio signal and the human auditory model, reducing manual parameter control complexity while maintaining high perceptual quality.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiments provide an audio processor for processing an audio signal to obtain a subband representation of the audio signal. The audio processor is configured to perform a cascaded lapped critically sampled transform on at least two partially overlapping blocks of samples of the audio signal, to obtain a set of subband samples on the basis of a first block of samples of the audio signal, and to obtain a corresponding set of subband samples on the basis of a second block of samples of the audio signal. Further, the audio processor is configured to perform a weighted combination of two corresponding sets of subband samples, one obtained on the basis of the first block of samples of the audio signal and one obtained on the basis on the second block of samples of the audio signal, to obtain an aliasing reduced subband representation of the audio signal; wherein performing a cascaded lapped critically sampled transform comprises segmenting a set of bins obtained on the basis of the first block of samples using at least two window functions, and to obtain at least two segmented sets of bins based on the segmented set of bins corresponding to the first block of samples; wherein performing a cascaded lapped critically sampled transform comprises segmenting a set of bins obtained on the basis of the second block of samples using the at least two window functions, and to obtain at least two sets of bins based on the segmented set of bins corresponding to the second block of samples; and wherein the sets of bins are processed using a second lapped critically sampled transform of the cascaded lapped critically sampled transform, wherein the second lapped critically sampled transform comprises performing lapped critically sampled transforms having the same framelength for at least one set of bins.