Analysis of articulatory timing based on a superposition model for VCV sequences

In this study, appropriate superposition timing regarding articulatory gestures between vowels and consonants for vowel-consonant-vowel (VCV) sequences, where C is /b/, /d/, or /g/, was investigated using a superposition model. In this concept, an articulatory gesture for a VCV sequence is considered as a combination of a vowel-to-vowel transition and a consonant gesture, and each gesture pattern as a function of time is modeled with a step response of cascaded first-order systems. From a subjective evaluation test and speech recognition using Julius, the appropriate superposition timing was estimated for the test words. The results suggest that the consonant closing gesture precedes the vowel closing gesture in terms of peak velocity, and, in contrast, the vowel opening gesture precedes the consonant opening gesture in the VCV sequences, especially for /VbV/ and /VdV/.