A model of optimal speech production planning integrating dynamical constraints to achieve appropriate articulatory timing

Ralf Winkler (1, 2) , Liang Ma (1) , Pascal Perrier (1)

1 GIPSA-MAGIC - GIPSA - Machines parlantes, Gestes oro-faciaux, Interaction Face-à-face, Communication augmentée
2 ZAS - Zentrum für Allgemeine Sprachwissenschaft [Berlin]

It has been often suggested that, similarly to skilled arm or limb movements in humans, the production of speech gestures could be based on an optimal planning in the central nervous system. This planning would use internal representations of the speech production apparatus to determine the motor command patterns allowing the achievement by the speaker of the desired speech communication goals with the minimum of effort. In this context major issues are related to the nature of the speech communication goals (targets or spatiotemporal trajectories) and to the definition of minimum of effort (minimum motor command change, minimum velocity peak, trajectory smoothness, minimum of jerk...). GEPPETO, the speech production model presented in this paper, has been designed within this general theoretical framework. It is based on a motor control model involving optimal planning to shape a biomechanical model of the vocal tract coupled with a harmonic acoustical model of speech production. In this paper it is shown how dynamical constraints can be taken into account in order to achieve appropriate articulatory timing.

Mots clés: Speech production; Motor Control; Movement planning; Optimal planning; Speech Timing

Domaines: Neurosciences; Robotique [cs.RO]; Autre; Linguistique

Ralf Winkler, Liang Ma, Pascal Perrier. A model of optimal speech production planning integrating dynamical constraints to achieve appropriate articulatory timing. CPMSP2 2010 - Cognitive and Physical Models of Speech Production, Speech Perception and Production-Perception Interaction - Part III Planning and Dynamics, Sep 2010, Berlin, Germany. pp.44-48. ⟨hal-00531573⟩