ACELP Speech Codebook Optimization Using Toeplitz Autocorrelation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The computationally complex bottleneck in ACELP-type speech coding algorithms, specifically in optimizing the residual codebook, is not viable due to the exhaustive search of a size NP space, leading to non-optimal iterative algorithms that fail to meet hardware complexity limits and perceptual quality requirements.
Innovation Solution
The proposed solution involves modifying the correlation matrix in the ACELP optimization process to an autocorrelation matrix with a Hermitian Toeplitz structure, which reduces computational complexity and memory requirements by considering the impact of the zero impulse response on both current and future frames, while maintaining perceptual quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If exhaustive search algorithm is used for codebook optimization, then coding quality is improved, but computational complexity increases excessively
Solution Approach 1:
The patent segments the codebook optimization process into two distinct stages: a training phase where the autocorrelation matrix is pre-computed and stored, and a coding phase where the pre-computed matrix is used for rapid optimization. This segmentation allows the computationally intensive matrix computation to be performed once during training, while the actual coding process benefits from reduced complexity using the stored matrix, thus resolving the contradiction between achieving optimal coding quality and maintaining manageable computational complexity.
2Measurement precision
If traditional correlation matrix approach is used, then codebook optimization is achieved, but memory requirements increase
Solution Approach 1:
The patent extracts the computationally intensive autocorrelation matrix computation from the real-time coding process and performs it separately during a training phase. The extracted matrix is then stored for reuse during coding, which reduces the memory requirements during active coding operations compared to storing and re-computing the full correlation matrix for each coding operation.
3Device complexity
If non-optimal iterative algorithms are used for codebook optimization, then computational complexity is reduced, but coding quality deteriorates
Solution Approach 1:
The patent performs preliminary computation of the autocorrelation matrix during a training phase before actual coding operations. This preliminary action prepares the optimization data structure in advance, allowing the coding process to use pre-computed values rather than performing iterative optimizations during real-time coding, thus achieving both reduced computational complexity and maintained coding quality.
Data Source
AI summary
An apparatus for encoding a speech signal that determines a codebook vector of a speech coding algorithm includes a matrix determiner for determining an autocorrelation matrix R, and a codebook vector determiner for determining the codebook vector depending on the autocorrelation matrix R. The matrix determiner determines the autocorrelation matrix R by determining vector coefficients of a vector r, wherein the autocorrelation matrix R includes a plurality of rows and a plurality of columns, wherein the vector r indicates one of the columns or one of the rows of the autocorrelation matrix R, wherein R(i, j)=r(|i−j|), wherein R(i, j) indicates the coefficients of the autocorrelation matrix R, wherein i is a first index indicating one of a plurality of rows of the autocorrelation matrix R, and wherein j is a second index indicating one of the plurality of columns of the autocorrelation matrix R.