14th International Congress of Phonetic Sciences (ICPhS-14)

San Francisco, CA, USA
August 1-7, 1999


The Accurate Estimation of Articulatory Synthesiser Parameters through Reducing the Degree of Saturation in a Neural Network Hidden Layer

K. M. Curtis (1), H. Altun (2)

RISN Group, University of Nottingham (1) Department of Electrical & Electronic Engineering, University of Nottingham, UK
(2) Department of Electrical & Electronic Engineering, Nigde University, Nigde, Turkey

A new method is proposed to correctly estimate the parameters of an articulatory speech synthesiser using a MLP neural network. This is achieved through modifying the statistical characteristic of the acoustic input pattern vectors in order to prevent the activation level of the hidden nodes from approaching saturation. The technique results in considerably faster neural learning and a more accurate estimation of the articulatory synthesiser parameters.

Full Paper

Bibliographic reference.  Curtis, K. M. / Altun, H. (1999): "The accurate estimation of articulatory synthesiser parameters through reducing the degree of saturation in a neural network hidden layer", In ICPhS-14, 2263-2266.