Adaptive Codebook Search for Low-Complexity Signal Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing coding technologies face challenges in reducing computation complexity while maintaining performance, particularly in searching pulse positions in algebraic codebooks, leading to high complexity and unstable performance due to either excessive search times or local maximum values.
Innovation Solution
The method selects different codebook search algorithms based on the characteristics of input signals, using low-complexity algorithms for signals with periodic characteristics and high-complexity algorithms for signals with white noise characteristics, thereby optimizing computation resources and ensuring coding quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If depth-first tree search procedure is used to search pulse positions in algebraic codebook, then search result quality is ensured, but computation complexity and search times increase significantly
Solution Approach 1:
The patent applies dynamics by making the search algorithm adaptive to different signal types. The encoder dynamically selects between depth-first tree search and global pulse replacement algorithms based on whether the input signal is voiced or unvoiced, allowing the system to optimize between search quality and computation complexity in real-time based on signal characteristics
Solution Approach 2:
The patent changes the algorithmic parameter (search strategy) based on signal type detection. For voiced signals, global pulse replacement is used which has lower complexity, while for unvoiced signals, depth-first tree search is employed to ensure higher search quality, thus adapting the computational approach to match the signal characteristics
2Productivity
If global pulse replacement procedure is used to reduce computation complexity, then search times are reduced, but performance becomes unstable due to local maximum values
Solution Approach 1:
The patent changes the search algorithm parameter based on signal type. Global pulse replacement is applied to voiced signals where it performs well with stable results, while depth-first tree search is applied to unvoiced signals to avoid local maximum issues, thus maintaining both efficiency and stability across different signal conditions
Solution Approach 2:
The system dynamically adjusts the search strategy based on real-time signal analysis. By detecting whether the signal is voiced or unvoiced, the encoder dynamically selects the appropriate algorithm, ensuring that the search process adapts to signal characteristics to maintain performance stability while optimizing search efficiency
3Measurement precision
If full search is performed to traverse all possible position combinations, then optimal codebook vector is guaranteed, but computation complexity becomes prohibitively high
Solution Approach 1:
The patent segments the search space based on signal type characteristics. By dividing signals into voiced and unvoiced categories, the search problem is segmented into two distinct sub-problems that can be solved with different algorithms, avoiding the need for exhaustive full search while still finding optimal or near-optimal solutions
Solution Approach 2:
The patent changes the search approach parameter based on signal type. For voiced signals, global pulse replacement provides efficient optimization without full search, while for unvoiced signals, depth-first tree search provides a balanced approach between optimality and complexity, eliminating the need for prohibitively expensive full search in all cases
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
A coding method is adapted to select different codebook search algorithms according to varied types of input signals. An encoder using the coding method is also provided. As appropriate search algorithms may be selected according to all possible structural features of the input signals, certain types of signals for which satisfactory results may be obtained through simple computations may match with search algorithms suitable for these signal types and having low computation complexities, so as to achieve better performance with fewer system resources. Meanwhile, other types of signals that need complicated computations may be processed by more sophisticated search algorithms, thereby ensuring the coding quality.