Scalable Audio Encoding Subband Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing scalable coding/decoding methods for speech and audio signals at low bit rates fail to adequately distinguish and prioritize perceptually important parameters, leading to suboptimal sound quality, especially when decoding at partial bit rates.
Innovation Solution
A coding apparatus and method that divides spectrum data into subbands, performs neighborhood searches, calculates lattice vectors, and multi-rate indexing, selecting a specific subband group based on energy and bit allocation to prioritize coding bits for perceptually important components, and a corresponding decoding apparatus that decodes only the selected subband group.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If scalable coding techniques are used to enable decoding at partial bit rates, then transmission efficiency is improved, but decoded signal quality deteriorates because perceptually important parameters are not prioritized
Solution Approach 1:
The patent applies local quality by differentiating the treatment of different subbands based on their perceptual importance. The decoding apparatus selectively decodes only those subbands that are determined to be perceptually important, while skipping less important ones. This ensures that limited decoding resources are concentrated on the most critical frequency components, thereby maintaining decoded signal quality even when operating at partial bit rates.
Solution Approach 2:
The patent changes the parameter of subband selection based on perceptual importance evaluation. The decoding apparatus dynamically determines which subbands to decode by evaluating their perceptual importance, and adjusts the decoding process accordingly. This parameter change allows the system to adapt to different bit rate conditions while prioritizing the reconstruction of perceptually critical signal components.
2Manufacturing precision
If all coded parameters are decoded to maintain signal quality, then decoded signal quality is improved, but decoding complexity and processing time increase
Solution Approach 1:
The patent applies the taking out principle by extracting and selectively decoding only the perceptually important subbands from the full set of coded parameters. Instead of decoding all subbands uniformly, the decoding apparatus identifies and extracts the critical subbands based on perceptual importance evaluation, decoding only those while skipping the rest. This extraction approach maintains signal quality for the most important components while significantly reducing decoding complexity.
Solution Approach 2:
The patent applies partial action by decoding only a subset of the total coded parameters rather than all of them. The decoding apparatus performs partial decoding on the perceptually important subbands, which is sufficient to maintain acceptable signal quality under bit rate constraints, while avoiding the excessive processing that would result from decoding all subbands at full quality.
3Adaptability or versatility
If coded parameters from multiple layers are combined, then coding flexibility is improved, but difficulty in identifying perceptually important parameters increases
Solution Approach 1:
The patent applies preliminary action by performing perceptual importance evaluation on subbands before the actual decoding process. The decoding apparatus first evaluates the perceptual importance of each subband based on the received coded parameters, and then uses this pre-evaluation to guide the selective decoding process. This preliminary evaluation simplifies the subsequent decoding by providing a clear prioritization framework, making it easier to identify which parameters to decode first regardless of their layer origin.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Disclosed is an encoding device capable of improving decoded signal quality. A local search unit (302) conducts a local search on a plurality of sub-bands generated by dividing spectrum data, and calculates lattice vectors for the spectra in the plurality of sub-bands. A multi-rate indexing unit (303) uses the lattice vectors to perform multi-rate indexing on each of the sub-bands, and generates indexing information showing the results thereof. A band selection unit (304) determines certain sub-bands from amongst the plurality of sub-bands in a plurality of encoding layers as perceptually important sub-band groups, where these are: within a selection range of sub-bands wherein the total number of encoding bits allocated to each of the plurality of sub-bands in the indexing information is equal to or less than an already set value, and within a sub-band selection range with the highest total energy of each of the plurality of sub-bands.