Learning Phonetic Features Using Connectionist Networks

Abstract

A method for learning phonetic features from speech data using connectionist networks is described. A temporal flow model is introduced in which sampled speech data flow through a parallel network from input to output units. The network uses hidden units with recurrent links to capture spectral/temporal characteristics of phonetic features. A supervised learning algorithm is presented which performs gradient descent in weight space using a course approximation of the desired output as an evaluation function. A simple connectionist network with recurrent links was trained on a single instance of the work pair “no” and “go,” and successfully learned a discriminatory mechanism. The trained network also correctly discriminated 98% of 25 other tokens of each word by the same speaker. The discriminatory feature was formed without segmentation of the input, and without a direct comparison of the two items. The network formed an internal representation of a single, integrated spectral feature which has a theoretical basis in human acoustic-phonetic perception.