From discrete linguistic units to continuous speech characteristics—a functional formulation of articulatory and phonatory dynamics

Abstract

Formulation of the process of conversion from discrete linguistic units to the continuous acoustic characteristics of speech and vice versa is crucial to both speech synthesis by rule and automatic speech recognition. This paper will present a review of recent works on this problem, with an emphasis on the functional modeling of the processes whereby the segmental and suprasegmental features of speech are realized through the dynamic control of the articulatory and the phonatory systems. The trajectories of acoustic or articulatory parameters of speech are to be interpreted as consequences of the control of the articulatory system by a sequence of discrete phonemic commands, while the trajectory of the voice fundamental frequency is to be regarded as a consequence of the control of the phonatory system by a sequence of discrete prosodic commands. Speech analysis and synthesis based on these models are shown to lead to a better understanding of the dynamic processes of speech production. [Work supported by Ministry of Education Grant‐in‐Aid for Scientific Research No. 239005.]