Comparison among time-delay neural networks, LVQ2 discrete parameter HMM and continuous parameter HMM

Author

Nakagawa, Seiichi ; Hirata, Yoshimitsu

Author_Institution

Toyohashi Univ. of Technol., Japan

fYear

1990

fDate

3-6 Apr 1990

Firstpage

509

Abstract

A continuous-parameter hidden Markov model (HMM) is proposed and compared with a discrete-parameter HMM, a time delay neural network (TDNN), and LVQ2 by using the same training and testing database. It is found that the proposed model´s performance is comparable to that of TDNN or LVQ2. Higher performance (96~97%) is obtained for all Japanese phonemes in isolated words. The HMM approach is superior to others for the recognition of time-sequential patterns like continuous speech

Keywords

Markov processes; speech recognition; HMM; Japanese phonemes; LVQ2; continuous-parameter hidden Markov model; speech recognition; time delay neural network; time-sequential patterns; Automatic speech recognition; Cepstrum; Context modeling; Databases; Delay effects; Hidden Markov models; Laboratories; Machine learning; Neural networks; Pattern recognition; Probability distribution; Speech recognition; Testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 1990. ICASSP-90., 1990 International Conference on

Conference_Location

Albuquerque, NM

ISSN

1520-6149

Type

conf

DOI

10.1109/ICASSP.1990.115761

Filename

115761

Link To Document

https://search.isc.ac/dl/search/defaultta.aspx?DTC=49&DC=2897226