SOFM和HMM在矢量量化码本优化中的应用  

Application of SOFM and HMM in Optimization of Codebook of Vector Quantization

作　　者：张雷[1] 孙诗瑛[1] 

机构地区：[1]上海交通大学电子工程系

出　　处：《南京邮电学院学报》1997年第4期57-60,共4页Journal of Nanjing University of Posts and Telecommunications(Natural Science)

摘　　要：在语音压缩编码中，矢量量化占有很大比重和计算量。码本的好坏与方法的选择有很大关系。将基于神经网络的ＳＯＦＭ（自组织特征映射算法）和ＨＭＭ（隐含马尔可夫过程）应用于矢量量化，从而产生更为优良的码本。In the compressive coding of speech signals, the vector quantization has a large amount of computation. The quality of codebook is related to the selection of methods. A method of applying neural network and hidden Markov method in the vector quantization is introduced, several simulated results are also given, thus a better codebook can be obtained by this method.

关 键 词：语声编码 矢量量化编码 SOFM HMM 

分 类 号：TN912.32[电子电信—通信与信息系统]