论文部分内容阅读
提出一种基于动态规划的矢量量化方法DPVQ(DynamicProgramBasedVectorQuantiza-tion。新方法既能有效压缩语音特征,又能充分保持时间序列信息,同时又兼顾码语音不同段的统计特性。
A new dynamic vector quantization (DPVQ) method based on dynamic programming is proposed, which not only can effectively compress the speech features but also keep the time series information fully, while taking into account the statistical characteristics of different segments of the code speech.