Title :
A non-linear filtering approach to stochastic training of the articulatory-acoustic mapping using the EM algorithm
Author_Institution :
Dept. of Electr. & Comput. Eng., Waterloo Univ., Ont., Canada
Abstract :
Current techniques for training representations of the articulatory-acoustic mapping from data rely on artificial simulations to provide codebooks of articulatory and acoustic measurements, which are then modelled by simple functional approximations. This paper outlines a stochastic framework for adapting an artificial model to real speech from acoustic measurements alone, using the EM algorithm. It is shown that parameter and state estimation problems for articulatory-acoustic inversion can be solved by adopting a statistical approach based on non-linear filtering
Keywords :
acoustic variables measurement; filtering theory; function approximation; maximum likelihood estimation; parameter estimation; speech recognition; speech synthesis; state estimation; statistical analysis; stochastic processes; EM algorithm; acoustic measurements; articulatory measurements; articulatory-acoustic inversion; articulatory-acoustic mapping; artificial simulations; codebooks; expectation maximization; functional approximation; nonlinear filtering approach; parameter estimation; real speech; state estimation; stochastic training; Acoustic measurements; Acoustical engineering; Computational modeling; Extraterrestrial measurements; Filtering algorithms; Speech recognition; Speech synthesis; State estimation; Stochastic processes; Training data;
Conference_Titel :
Spoken Language, 1996. ICSLP 96. Proceedings., Fourth International Conference on
Conference_Location :
Philadelphia, PA
Print_ISBN :
0-7803-3555-4
DOI :
10.1109/ICSLP.1996.607167
