PVQ Shape Search With Adaptive Bit Length for Search Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Structured Pyramid Vector Quantization (PVQ) for speech and audio coding faces challenges in achieving efficient search complexity while maintaining high Signal to Noise Ratio (SNR) in scenarios with memory and search complexity constraints, particularly for high-rate coding where the number of allowed unit pulses and dimensions are high.
Innovation Solution
The method involves determining the need for a longer bit word length based on the maximum pulse amplitude and accumulated energy to perform PVQ shape search efficiently, using a fine search unit to adaptively choose between different bit word lengths for inner loop calculations, and employing cross-multiplication to determine the best position for unit pulse addition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If unconstrained vector quantization is used, then quantization quality is improved, but complexity and memory capacity requirements increase
Solution Approach 1:
The patent segments the vector quantization process into structured components (gain-shape decomposition, pyramid structure with nested loops) that separate the quantization search into manageable parts. This segmentation allows quality preservation while reducing overall search complexity by organizing the search space hierarchically.
Solution Approach 2:
The patent employs dynamic adaptation in the search process through early termination conditions and adaptive precision adjustment. The search can dynamically terminate when quality thresholds are met, and precision can be adapted based on signal characteristics, reducing unnecessary computational operations while maintaining quality.
2Measurement precision
If higher precision bit word length is used in inner loop calculations, then SNR is improved, but computational complexity and memory usage increase
Solution Approach 1:
The patent applies different precision levels locally within the search algorithm. Critical calculations that most impact SNR (such as correlation computations and energy calculations) use higher precision, while less critical operations use lower precision. This local quality differentiation maintains overall SNR performance while reducing total computational complexity.
Solution Approach 2:
The patent dynamically changes precision parameters based on signal characteristics and search progress. The bit word length and accumulator precision are adjusted as parameters based on the current search state and signal properties, optimizing the balance between SNR and computational complexity for different coding scenarios.
3Measurement precision
If more unit pulses and higher dimensions are used in PVQ, then coding quality is improved, but search complexity increases significantly
Solution Approach 1:
The patent implements a nested search structure with outer and inner loops that systematically explore the search space. The nested pyramid structure with correlated search dimensions allows efficient traversal of high-dimensional spaces by reusing computations across nested levels, reducing the exponential search complexity that would otherwise result from higher dimensions and more unit pulses.
Solution Approach 2:
The patent performs preliminary computations before the main search, such as pre-calculating correlation values, energy terms, and normalization factors. These preliminary actions prepare the search space in advance, allowing the main search loop to operate more efficiently with reduced per-iteration complexity, thereby improving overall search productivity for high-rate coding.
Data Source
AI summary
An encoder and a method therein for Pyramid Vector Quantizer, PVQ, shape search, the PVQ taking a target vector x as input and deriving a vector y by iteratively adding unit pulses in an inner dimension search loop. The method comprises, before entering a next inner dimension search loop for unit pulse addition, determining, based on the maximum pulse amplitude, maxampy, of a current vector y, whether more than a current bit word length is needed to represent enloopy, in a lossless manner in the upcoming inner dimension loop. The variable enloopy is related to an accumulated energy of the vector y. The performing of this method enables the encoder to keep the complexity of the search at a reasonable level.