Gain-Shape Vector Coding with Dynamic Segment Bit Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing gain-shape vector quantization methods face challenges in determining a suitable vector length, leading to increased overhead in gain coding and reduced capability in exploiting sample correlation, particularly due to fixed band structures that do not adapt to changes in the target vector, resulting in suboptimal encoding and decoding efficiency.
Innovation Solution
A method for determining an optimal number of segments in gain-shape vector encoding by analyzing the target vector's energies and bit budget constraints, allowing for efficient bit allocation and improved decoding performance, which involves increasing the number of segments based on energy variations and bit availability, thereby enhancing the encoding and decoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the vector length is increased to capture larger variations, then the shape quantizer needs to handle more dynamics, but the overhead for gain coding increases and fewer bits are left for shape coding
Solution Approach 1:
The target vector is divided into multiple segments, and the quantization is performed separately on each segment. This segmentation allows the system to handle variations within each segment more effectively while reducing the overall complexity and bit requirements compared to quantizing the entire long vector at once.
Solution Approach 2:
The number of segments is determined dynamically based on the energy variations in the target vector and the available bit budget. This adaptive approach allows the system to optimize the trade-off between quantization precision and bit allocation, adjusting the segmentation level according to the specific characteristics of each vector.
2Loss of information
If the vector length is decreased to reduce dynamics, then the shape VQ has less capability to exploit sample correlation, but the gain coding overhead is reduced
Solution Approach 1:
By segmenting the vector, the system preserves local sample correlations within each segment while reducing the overall vector length. This allows the shape VQ to effectively exploit correlations within segments without requiring excessive bits for coding the entire vector.
Solution Approach 2:
Each segment is quantized independently with its own shape vector and gain, allowing local optimization of the quantization process. This local quality approach ensures that sample correlations are exploited effectively within each segment while maintaining overall efficiency.
3Device complexity
If a fixed band structure is used for partitioning, then the encoding process is simplified, but the structure cannot adapt to changes in the target vector
Solution Approach 1:
The band structure is made dynamic by allowing the number of segments to vary based on the energy characteristics of the target vector. This enables the encoding structure to adapt to different signal conditions while maintaining a relatively simple overall framework.
Solution Approach 2:
The parameters of the band structure, specifically the number of segments, are changed adaptively based on the target vector's energy distribution. This allows the system to optimize performance for different signal types without requiring a completely different encoding structure.
Data Source
Figure 1~2
Figure 3a~4
Figure 5~6
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_init, 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_init. 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_initsegments and a difference between B and BAVG. The performing of the method enables an efficient allocation of the bits of the bit budget over the target vector.