Obstruent-sonorant consonant sequences—Analysis by synthesis

Abstract

The goal of this work is to develop principles of overlapping gestures in obstruent-sonorant sequences in the word-initial position and sonorant-obstruent sequences in the word-final position. Consonant clusters such as sm in small are phonetically represented as a sequence of individual elements, but the exact perceptual representation is unclear. The modification during the production of these overlapping gestures may be driven partly by perceptual salience and partly by vocal tract aerodynamics. When two consonants occur next to each other, the same gestures may be made as for only one consonant. The aerodynamics of the vocal tract may account for the modification in the timing of the articulators during production and this modification can be incorporated as rules into HLsyn, a higher-level quasiarticulatory speech synthesizer that takes as inputs the pressures and the flows of the vocal tract. Acoustic information extracted from the speech waveform is mapped into inputs for HLsyn. This analysis by synthesis approach is a method to develop a more precise picture of the planning stage during speech production where the acoustic phonetics must be carefully planned and modified to acheive the correct target sounds. [Work funded by a grant provided by NIH.]