Hierarchical Audio Filterbank for Scalable Bitrate Decomposition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional audio compression methods are not scalable to address varying channel capacity, storage, or bit stream data rate requirements, limiting their ability to adjust audio quality dynamically.
Innovation Solution
A method using a hierarchical filterbank to decompose audio signals into multi-resolution time/frequency representations, extracting tonal and residual components, and encoding them with differential and joint channel coding to create a scalable bit stream that can be scaled to arbitrary data rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional audio compression methods are used, then audio quality is fixed at encoding time, but the system cannot adapt to varying channel capacity or storage requirements
Solution Approach 1:
The audio signal is decomposed into multiple frequency sub-bands using a hierarchical filterbank, where each sub-band can be independently encoded and scaled. This segmentation allows the system to selectively retain or discard sub-bands based on available bandwidth, providing adaptability without requiring complete re-encoding.
Solution Approach 2:
The encoding system uses dynamic bit allocation across different frequency sub-bands based on psychoacoustic models. The bit allocation adjusts in real-time according to signal characteristics and available bandwidth, allowing the system to maintain perceived quality while adapting to varying data rate requirements.
2Adaptability or versatility
If error encoding is used to create scalable bit streams, then data rate scalability is achieved, but coding efficiency decreases
Solution Approach 1:
Different encoding precision is applied to different frequency sub-bands based on their perceptual importance. Critical bands receive higher encoding precision while less critical bands use coarser quantization, optimizing coding efficiency while maintaining scalability through selective quality distribution.
Solution Approach 2:
The system changes quantization parameters and bit allocation dynamically across different sub-bands and scalability layers. By adjusting these parameters locally rather than uniformly, the system achieves both scalability and improved coding efficiency compared to uniform error encoding approaches.
3Adaptability or versatility
If multiple compression algorithms are used for layered scaling, then support for multiple data rates is provided, but device complexity increases
Solution Approach 1:
A single hierarchical filterbank-based compression algorithm performs multiple functions: it provides both the base layer encoding and all enhancement layers simultaneously. The same core algorithm structure supports all data rates by selectively combining sub-bands, eliminating the need for multiple separate compression algorithms while maintaining layered scalability.
Data Source
Figure 1
Figure 2a~2b
Figure 3
AI summary
The invention regards a method of hierarchically filtering an input signal to achieve a nearly arbitrary time/frequency decomposition. The method comprises the steps of: (a) buffering samples of the input signal into frames of N samples (2900); (b) multiplying the N samples in each frame by an N-sample window function (2900); (c) applying an N-point transform to produce N/2 transform coefficients (2902); (d) dividing the N/2 residual transform coefficients into P groups of Mi coefficients (2906), such that the sum of the Mi coefficients is N/2 f∈1l2 (e) for each of P groups, applying a (2* Mi)-point inverse transform to the transform coefficients to produce (2* Mi) sub-band samples from each group (2906); (f) in each sub-band i, multiplying the (2* Mi) sub-band samples by a (2* Mi)-point window function (2908); (g) in each sub-band i, overlapping with Mi previous samples and adding corresponding values to produce Mi new samples for each sub-band (2910); and (h) repeating steps (a)-(g) on one or more of the sub-bands of Mi new samples using successively smaller transform sizes N (2912) until the desired time/transform resolution is achieved (2914). The invention further comprises a method of hierarchically reconstructing time samples of an input signal, in which each input frame contains Mi time samples in each of P sub-bands.