Scaled Codebook Search for Low-Complexity Audio Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech and audio codecs require high memory and computational complexity for encoding input vectors, particularly in nearest neighbor searches within multiple scale lattice codebooks, which is undesirable for mobile terminal communications.
Innovation Solution
A method is introduced to determine potential basis code vectors for encoding input vectors by associating each set of basis code vectors with scale representatives, allowing code vectors to be scaled by respective scale representatives, thereby reducing the search complexity and memory requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a multiple scale lattice codebook with several lattice truncations is used for speech and audio coding, then coding efficiency is improved, but memory requirements and computational complexity increase
Solution Approach 1:
The codebook search process is segmented into two stages: first identifying potential basis code vectors from each set of basis code vectors, then determining the final code vector from a subset of scaled code vectors derived from these potential basis vectors. This segmentation reduces the overall search complexity while maintaining coding efficiency.
Solution Approach 2:
The patent performs preliminary identification of potential basis code vectors before the final code vector determination. By pre-identifying which basis code vectors are most relevant and then only scaling and evaluating those specific vectors, the method avoids the computational burden of evaluating all code vectors in the multiple scale lattice codebook.
2Manufacturing precision
If a multiple scale lattice codebook with several lattice truncations is used for speech and audio coding, then coding efficiency is improved, but memory requirements increase
Solution Approach 1:
The codebook is effectively segmented into multiple sets of basis code vectors, each associated with scale representatives. Only the necessary scaled versions of potential basis code vectors are stored and evaluated, rather than storing all possible code vectors, thereby reducing memory requirements.
Solution Approach 2:
Each set of basis code vectors serves multiple purposes: it can be scaled by different scale representatives to generate multiple code vectors. This multi-functionality allows the same basis code vectors to contribute to multiple scale levels, reducing the total number of code vectors that need to be stored in memory.
3Measurement precision
If distortion is evaluated for all leader classes in each lattice truncation, then encoding accuracy is maintained, but computational complexity increases
Solution Approach 1:
The patent extracts only the essential components for accurate encoding: potential basis code vectors from each set. By taking out only these critical vectors and their scaled versions for evaluation, rather than evaluating all leader classes, the method maintains encoding accuracy while significantly reducing computational complexity.
Solution Approach 2:
Instead of performing exhaustive distortion evaluation on all leader classes, the method performs partial evaluation on a selected subset of scaled code vectors derived from potential basis code vectors. This partial action is sufficient to achieve good encoding accuracy without the excessive computational cost of complete evaluation.
Data Source
AI summary
It is inter alia disclosed to determine, for each set of basis code vectors of a plurality of sets of basis code vectors, a potential basis code vector for encoding an input vector, wherein each set of basis code vectors is associated with at least one scale representative of a plurality of scale representatives, and to determine a code vector for encoding the input vector from a subset of code vectors, said subset of code vectors comprising, for each determined potential basis code vector and each scale representative associated with the set of basis code vectors of the respective potential basis code vector, a code vector based on the respective potential basis code vector scaled by the respective scale representative.