Codebook classification rearrangement vector quantization
Xueying Zhang · Journal of Tsinghua University(Science and Technology) · 2013
Vector quantization efficiently reduces the calculational times for speech codecs.Some general vector quantization methods are split-multistage vector quantization(S-MSVQ) and switched split vector quantization.However,balancing their memory requirements,computational complexities and speed to get good decoded speech quality is difficult.This paper presents a codebook classification rearrangement vector quantization(CCRVQ) method that reduces the search area by rearranging and classifying the designed codebooks.Codebook classification rearranging split-multistage vector quantization(CCRS-MSVQ) combines CCRVQ with S-MSVQ.Assuming that the quantization bits and codebook size remain unchanged,tests demonstrate that the CCRS-MSVQ computational complexity is reduced by 90.24% compared to S-MSVQ,which enables the vector quantizer to produce real time transparent speech coding.