On increasing structural complexity of finite state speech models

Author

Vaseghi, S.V. ; Conner, P.

Author_Institution

Sch. of Inf. Syst., East Anglia Univ., Norwich, UK

Volume

1

fYear

1992

fDate

23-26 Mar 1992

Firstpage

537

Abstract

Some methods of modeling speech spectral features and duration within the framework of finite-state models are discussed. On observation modeling, the use of cepstral-time matrices, instead of cepstral vectors, as the observation unit is investigated. On duration modeling, a new HMM is introduced in which state transition and duration probabilities are combined to form duration-dependent transition probabilities. The duration dependent transitions are derived from the cumulative density function (CDF) of state duration

Keywords

hidden Markov models; probability; speech analysis and processing; speech recognition; HMM; cepstral vectors; cepstral-time matrices; cumulative density function; duration modeling; duration-dependent transition probabilities; finite state speech models; observation modeling; observation unit; spectral features; speech duration; speech recognition; state duration probability; state transition probability; structural complexity; Automatic speech recognition; Bit rate; Cepstral analysis; Density functional theory; Distribution functions; Hidden Markov models; Information systems; Probability; Quantization; Speech recognition;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 1992. ICASSP-92., 1992 IEEE International Conference on

Conference_Location

San Francisco, CA

ISSN

1520-6149

Print_ISBN

0-7803-0532-9

Type

conf

DOI

10.1109/ICASSP.1992.225852

Filename

225852