Summary: COMBINING NEURAL NETWORKS AND HIDDEN MARKOV MODELS
M
FOR CONTINUOUS SPEECH RECOGNITION
ichael Cohen*, David Rumelhart**, Nelson Morgan***,
*
Horacio Franco*, Victor Abrash*, and Yochai Konig***
Speech Research Program, SRI International, Menlo Park, CA 94025
*
** Stanford University, Dept. of Psychology, Stanford, CA 94305
** Intl. Computer Science Inst., 1947 Center Street, Suite 600, Berkeley, CA 94704
W
ABSTRACT
e present a speaker­independent, continuous­speech recog­
(
nition system based on a hybrid multilayer perceptron
MLP)/hidden Markov model (HMM). The system com­
e
bines the advantages of both approaches by using MLPs to
stimate the state­dependent observation probabilities of an
e

Source: Abrash, Victor - Speech Technology & Research Laboratory, SRI International

Collections: Computer Technologies and Information Sciences