Automatic classification of consonant-vowel transitions based on subglottal resonances and second formant frequencies

Abstract

A growing body of work indicates that subglottal resonances (SGRs) divide the frequency space of vowels and consonants into distinct regions [Lulich, J. Phonetics (2009a)]. For instance, the second formant frequency (F2) of back vowels lies between the first and second subglottal resonances (Sg1 and Sg2), while for front vowels F2 lies at higher frequency than Sg2. In alveolar stop bursts, F2 lies between Sg2 and the third subglottal resonance (Sg3). In this study, we build on previous work [Lulich, POMA (2009b)] to automatically determine the SGRs from pairs of F2 measurements made in the stop burst and the vowel midpoint of 405 consonant-vowel (CV) transitions produced by an adult male native speaker of American English. Consonants and vowels are then classified into place of articulation and front/back features based only on the relations between F2 and the SGRs. The mean hit rate across categories was 77%, with a mean false alarm rate of 12%. Hit and false alarm rates in a baseline experiment (in which the F2 distribution was divided into four equal parts and then classified) were 71% and 14%. This knowledge-based approach to classifying CV transitions is potentially useful when vowel and consonant identities are unavailable.