摘要
提出了一种模糊聚类分析算法ＳＦＣＭ，并将其用于语音特征的矢量量化，最终形成码本尺寸为２５６的码本．用ＳＦＣＭ算法得到的码本分布合理，没有空类．采用此码本的语音识别实验表明了这种量化方法对语音识别的有效性．
A fuzzy clustering analysis algorithm SFCM is presented and is applied to the vector quantization of speech feature, to build up the code vector of 256 quantization degrees. The distribution of the code vector obtained by SFCM is reasonable and there are not almost empty classes. The experiment results demonstrate the efficiency of this quantization method for speech recognition.
出处
《东北大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
1998年第6期595-598,共4页
Journal of Northeastern University(Natural Science)
基金
国家自然科学基金
国家教委博士点基金
关键词
模糊聚类分析
语音识别
矢量量化
SFCM
fuzzy clustering analysis, speech recognition, vector quantization.
