Split Gain-Shape Vector Coding Under Segment Bit Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing gain-shape vector quantization methods face challenges in determining a suitable vector length, leading to inefficiencies in encoding and decoding due to varying vector dynamics and bit allocation constraints, which affect the quality of audio codecs.
Innovation Solution
A method for split gain shape vector encoding and decoding that determines an optimal number of segments based on energy variations and bit budget constraints, allowing for efficient bit allocation and improved perceptual quality by adjusting the number of segments dynamically.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the vector length is increased to capture larger variations within the vector, then the shape quantizer can handle the dynamics of the signal better, but the overhead for gain coding increases, leaving fewer bits for shape coding
Solution Approach 1:
The target vector is divided into multiple segments, each of which is independently quantized. This segmentation allows the system to handle larger vectors by breaking them into manageable parts, reducing the gain coding overhead while maintaining the ability to capture signal dynamics within each segment.
Solution Approach 2:
The number of segments is dynamically adjusted based on the complexity of the signal and available bit budget. This dynamic adaptation allows the system to optimize the trade-off between capturing signal dynamics and maintaining sufficient bits for shape coding, rather than using a fixed vector length.
2Loss of information
If the vector length is decreased to reduce dynamics within the vector, then fewer bits are needed for gain coding, but the lower dimensionality of the shape VQ has less capability to exploit sample correlation
Solution Approach 1:
By segmenting the vector, the system maintains higher effective dimensionality for shape VQ within each segment, preserving the capability to exploit sample correlation locally while keeping the overall bit budget manageable through distributed quantization across multiple segments.
3Device complexity
If a fixed band structure is used to partition the spectrum, then the encoding process is simplified, but there is no adaptation of the band structure to changes in the target vector
Solution Approach 1:
The band structure is made dynamic by allowing the number and size of segments to be adjusted based on the characteristics of the target vector. This enables adaptation to signal changes while maintaining a structured approach to quantization, balancing complexity and versatility.
Data Source
AI summary
The invention relates to an encoder and a decoder and methods therein for supporting split gain shape vector encoding and decoding. The method performed by an encoder, where the encoding of each vector segment is subjected to a constraint related to a maximum number of bits, BMAX, allowed for encoding a vector segment. The method comprises, determining an initial number, Np_<sup2>init</sup2>, of segments for a target vector x; and further determining an average number of bits per segment, BAVG, based on a vector bit budget and Np_<sup2>init</sup2>. The method further comprises determining a final number of segments to be used, for the vector x, in the gain shape vector encoding, based on energies of the Np_<sup2>init </sup2>segments and a difference between BMAX and BAVG. The performing of the method enables an efficient allocation of the bits of the bit budget over the target vector.