Automatic extraction of stop-oriented features from Chinese Speech wave using wavelet transform

The concept of directive features for Chinese Speech is presented in this paper.From which a method and its digital signal processing algorithm for automatically extracting stoiroriented features from Chinese speech waveform by using wavelet transform are presented.This method classifies Chinese voiceless consonants into two stop subsets,BDG:b,d,gand zZHJGPTcCHQK:z,zh,i,g,p) t,c,ch,q,k,and one fricative subset FsSHbX:f,s,sh,x,h.For each speech segment of syllables containing voiceless consonant as its initial,the algorithm calculates detection objectives and outputs one of category symbols out of b.d.g,STOP/BD,f.s.sh.x.h and its segment markers,which provide a new'coars to fine'match-searching structure for Chinese automatic speech recognition system.The validity of the method was tested on a subset of 913 syllables containing voiceless consonant as its initial from a database of real speech consisting of 1267 Chinese all-syllable tokens,with hand-labeled initial and final segment markers as the benchmark of the test,resulting in a classification accuracy of 96.1%,95.1%,and 89.0% for category b.d.g,STOP/BD,and Ls.sh.x.h respectively,and in an average accuracy of 93.6% for all of the 913 synables.