Temporal control improvement of hidden Markov models for automatic speech recognition

A contribution for speaker-independent recognition systems is presented, as well for speech recognition in the presence of large vocabularies as for word spotting. The recognition strategy used is based upon continuous hidden markov models (HMMs) which include two levels (phonemic level and word lev...

Full description

Saved in:
Bibliographic Details
Published in:Microprocessing and microprogramming 1991-08, Vol.32 (1), p.549-556
Main Author: Dours-Senac, Christine
Format: Article
Language:English
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A contribution for speaker-independent recognition systems is presented, as well for speech recognition in the presence of large vocabularies as for word spotting. The recognition strategy used is based upon continuous hidden markov models (HMMs) which include two levels (phonemic level and word level). The aim of the method is double; first to provide a fast algorithm for sub-vocabulary preselection or recognition in the presence of small vocabulary, second to improve the temporal control of the standard HMMs.
ISSN:0165-6074
DOI:10.1016/0165-6074(91)90400-N