A new approach for speech dynamics studies

A new dynamic model for vocal tract configuration based on a theoretical quantification of dynamic area function A(x,t) into distinctive regions having areas S(r,t) is proposed. Articulation is transcoded into changes of these S(r,t) through three modes [Mrayati et al., Speech Comm. 7, 257–286 (1988)]. A dynamic control strategy of S(r,t) is extremely simple to apply and correlates well with human speech organ musculature movements. This strategy transprojects the classical longitudinal displacement of the constriction into a transversal area function variation of S(r,t). This dynamic strategy can incorporate easily context‐dependent allophonic variations, such as: coarticulation, reduction, assimilation …. The distinctive regions, along the vocal tract, are associated with specific consonants and vowels. These new “phonetic regions” concept can be used to vectorially code target configurations for consonants and vowels. The new concept relates, in a simple fashion, changes in the acoustic domain to chang...