Bio-inspired Dynamic Formant Tracking for Phonetic Labelling

Abstract

It is a known fact that phonetic labeling may be relevant in helping current Automatic Speech Recognition (ASR) when combined with classical parsing systems as HMM's by reducing the search space. Through the present paper a method for Phonetic Broad-Class Labeling (PCL) based on speech perception in the high auditory centers is described. The methodology is based in the operation of CF (Characteristic Frequency) and FM (Frequency Modulation) neurons in the cochlear nucleus and cortical complex of the human auditory apparatus in the automatic detection of formants and formant dynamics on speech. Results obtained informant detection and dynamic formant tracking are given and the applicability of the method to Speech Processing is discussed.