A LINEAR PREDICTIVE HMM FOR VECTOR-VALUED OBSERVATIONS WITH APPLICATIONS TO SPEECH RECOGNITION

被引：62

作者：

KENNY, P ^{[1
]}

LENNIG, M ^{[1
]}

MERMELSTEIN, P ^{[1
]}

机构：

[1] BELL NO RES,MAN MACHINE SYST,MONTREAL,QUEBEC,CANADA

来源：

IEEE TRANSACTIONS ON ACOUSTICS SPEECH AND SIGNAL PROCESSING | 1990年 / 38卷 / 02期

基金：

加拿大自然科学与工程研究理事会;

关键词：

D O I：

10.1109/29.103057

中图分类号：

O42 [声学];

学科分类号：

070206 ; 082403 ;

摘要：

In this paper we describe a new type of Markov model which we have developed to account for the correlations between successive frames of the speech signal. The idea is to treat the sequence of frames as a nonstationary autoregressive process whose parameters are controlled by a hidden Markov chain. We show that this type of model performs better than the standard multivariate Gaussian HMM when it is incorporated into a large-vocabulary isolated-word recognizer. © 1990 IEEE

引用

页码：220 / 225

页数：6