DocumentCode :
336781
Title :
The HDM: a segmental hidden dynamic model of coarticulation
Author :
Richards, Hywel B. ; Bridle, John S.
Author_Institution :
Dragon Syst. UK, Cheltenham, UK
Volume :
1
fYear :
1999
fDate :
15-19 Mar 1999
Firstpage :
357
Abstract :
This paper introduces a new approach to acoustic-phonetic modelling, the hidden dynamic model (HDM), which explicitly accounts for the coarticulation and transitions between neighbouring phones. Inspired by the fact that speech is really produced by an underlying dynamic system, the HDM consists of a single vector target per phone in a hidden dynamic space in which speech trajectories are produced by a simple dynamic system. The hidden space is mapped to the surface acoustic representation via a non-linear mapping in the form of a multilayer perceptron (MLP). Algorithms are presented for training of all the parameters (target vectors and MLP weights) from segmented and labelled acoustic observations alone, with no special initialisation. The model captures the dynamic structure of speech, and appears to aid a speech recognition task based on the SwitchBoard corpus
Keywords :
multilayer perceptrons; speech recognition; HDM; MLP weights; acoustic-phonetic modelling; coarticulation; dynamic structure; hidden dynamic model; hidden dynamic space; multilayer perceptron; neighbouring phones; nonlinear mapping; segmental hidden dynamic model; speech recognition task; speech trajectories; surface acoustic representation; target vectors; training; underlying dynamic system; Hidden Markov models; Low pass filters; Multidimensional systems; Multilayer perceptrons; Nonlinear acoustics; Nonlinear dynamical systems; Speech recognition; Speech synthesis; Trajectory; Vocabulary;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1999. Proceedings., 1999 IEEE International Conference on
Conference_Location :
Phoenix, AZ
ISSN :
1520-6149
Print_ISBN :
0-7803-5041-3
Type :
conf
DOI :
10.1109/ICASSP.1999.758136
Filename :
758136
Link To Document :
بازگشت