Vector Quantization Bit Allocation for Multi-Rate Audio Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multi-rate lattice vector quantization methods require a high number of bits for encoding, which can be inefficient in terms of bit allocation and encoding complexity, especially when dealing with split multi-rate lattice vector quantization and algebraic vector quantization in audio or voice encoding.
Innovation Solution
An encoding apparatus and method that dynamically adjusts bit allocation by generating quantization parameters and determining whether to encode a codebook indicator or the difference in bits allocated to vector quantization, based on the available bits for encoding sub-vectors, using a control circuitry to manage bit usage and optimize encoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multi-rate lattice vector quantization is applied to audio or voice encoding, then quantization accuracy is improved, but the number of bits for encoding increases
Solution Approach 1:
The patent divides the vector quantization process into multiple rates (first rate and second rate), allowing different precision levels for different segments of the data. This segmentation enables high quantization accuracy where needed while using fewer bits for other portions, thus resolving the contradiction between accuracy and bit consumption.
Solution Approach 2:
The patent applies different quantization strategies to different parts of the data structure. Specifically, codebook indicators are encoded with fewer bits when the codebook is determined by surrounding elements, while only necessary differential information is transmitted. This local quality approach maintains overall accuracy while reducing total bit consumption.
2Manufacturing precision
If split multi-rate lattice vector quantization is used, then encoding precision is improved, but encoding complexity increases
Solution Approach 1:
The patent determines codebooks for sub-vectors in advance based on surrounding codebook indicators, before the actual encoding process. This preliminary determination simplifies the encoding complexity by pre-establishing the quantization structure, while still achieving high encoding precision through the multi-rate approach.
Solution Approach 2:
The patent enables the encoding system to self-determine codebook selections based on surrounding elements and differential information, reducing the need for complex external control mechanisms. The system uses its own structure (surrounding codebook indicators) to guide the encoding process, thereby reducing complexity while maintaining precision.
3Reliability
If algebraic vector quantization is applied, then voice encoding quality is improved, but bit allocation efficiency decreases
Solution Approach 1:
The patent dynamically adjusts bit allocation based on the relationship between codebook indicators and surrounding elements. When a codebook can be determined from surrounding information, fewer bits are allocated; otherwise, more bits are used. This dynamic approach maintains high voice encoding quality while improving bit allocation efficiency.
Solution Approach 2:
The patent changes the encoding parameters (number of bits) based on the specific conditions of each codebook indicator. By varying the bit allocation parameter according to whether the codebook is determined by surrounding elements or requires explicit encoding, the system achieves both high quality and efficient bit usage.
Data Source
AI summary
An encoding apparatus includes: a quantization circuit that generates a quantization parameter including first information related to a codebook of vector quantization and second information related to code vectors included in the codebook; and a control circuit that determines which one of first encoding of the first information for the target sub-vector and second encoding of a second number of bits based on the difference between an allocated number of bits for vector quantization and the number of bits of the quantization parameter is to be executed, in accordance with the number of bits available for encoding sub-vectors including at least a target sub-vector among a plurality of sub-vectors in the vector quantization.