Frequency Envelope Vector Quantization With Adaptive Codebooks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing frequency envelope quantization methods face challenges with insufficient quantization precision and excessive complexity due to limited bits in the codebook and a large number of frequency envelopes, particularly when using vector quantization.
Innovation Solution
The proposed method divides frequency envelopes into vectors with fewer dimensions, allowing for vector quantization using a codebook with a smaller number of bits, reducing complexity and ensuring quantization quality by adjusting the codebook based on previous quantization results and using a secondary codebook with reduced bit requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If scalar quantization is used for each frequency envelope, then quantization precision is maintained, but bit consumption becomes excessive
Solution Approach 1:
The patent combines multiple frequency envelopes into a single vector for joint quantization. Instead of quantizing each frequency envelope separately using scalar quantization, the method merges N frequency envelopes into one N-dimensional vector, which is then quantized using vector quantization with a codebook of size 2^B. This merging approach maintains quantization precision while significantly reducing bit consumption from N*B bits to just B bits.
2Measurement precision
If vector quantization is used with a large codebook, then quantization quality improves, but computational complexity increases
Solution Approach 1:
The patent segments the large codebook search process into two stages: first, it divides the N-dimensional frequency envelope vector into multiple smaller sub-vectors; second, it performs separate codebook searches for each sub-vector using a smaller codebook of size 2^(B/2). This segmentation reduces the computational complexity from searching through 2^B codebook entries to searching through 2^(B/2) entries multiple times, while still achieving good quantization quality.
Data Source
Figure 1
Figure 2~3
AI summary
Embodiments of the present invention proposes a frequency envelope vector quantization method and apparatus, where the method includes: dividing N frequency envelopes in one frame into N1 vectors, where each vector in the N1 vectors includes M frequency envelopes; quantizing a first vector in the N1 vectors by using a first codebook, to obtain a code word corresponding to the quantized first vector, where the first codebook is divided into 2B1 portions; determining, according to the code word corresponding to the quantized first vector, that the quantized first vector is associated with the ith portion in the 2B1 portions of the first codebook; determining a second codebook according to the codebook of the ith portion; and quantizing a second vector in the N1 vectors based on the second codebook. In the embodiments of the present invention, frequency envelopes are divided into a plurality of vectors with fewer dimensions, so that vector quantization can be performed on frequency envelope vectors by using a codebook with a smaller quantity of bits. Therefore, complexity of vector quantization can be reduced, and an effect of vector quantization can also be ensured.