A Temporal Fuzzy-ART Neural Network Architecture as a Model of Phoneme Perception

This paper explores the applicability of Adaptive Resistance Theory- (ART-) type neural networks for finding and encoding linguistic structures, specifically those corresponding to acoustic patterns in natural speech. We build an interpretation of human perceptual response to acoustic pattern in natural speech, translating this to a neural architecture as a model of acquisition, storage, and classification of acoustic speech patterns.LinguisticsMastersUniversity of New Mexico. Dept. of LinguisticsLuger, GeorgeMorford, JillCaudell, Thoma