A study of two-formant models for vowel identification

Abstract

An experiment has been performed where various two-formant vowel models reported in the literature were assessed as to their ability to predict the formant frequencies obtained in a vowel identification task. An alternative model is proposed in which the auditory processing of vowel sounds is assumed to take place in two stages: the peripheral processing stage and the central processing stage. In the peripheral processing stage, the speech spectrum is transformed to its auditory equivalent and the frequencies of the first three formants are extracted from this spectrum, using the peak-picking mechanism. The central processing stage performs a two-formant approximation. The first and second formant frequencies of the two-formant model are taken to be equal to the first and second formant frequencies extracted at the peripheral processing stage plus a perturbation term which accounts for the interaction effects of the neighboring formants. The perturbation caused by each of these neighboring formants is inversely proportional to its separation from the main formants. This model compares favorably with previous models in its prediction of the formant frequencies obtained from the vowel identification task.