"GEPPETO": A target-based model of speech production including optimal planning and physical modeling

International audienceGEPPETO is a model of speech production aiming at generating speech movements and speech acoustical signals from a sequence of phonemes, with some prosodic constraints expressed in terms of stress and speaking rate. GEPPETO includes (1) a model of speech sequence planning, (2) a model of muscle force generation mechanisms and (3) a model of the physical speech production apparatus. The physical model is a two-dimensional biomechanical model of the vocal tract, in which the tongue is represented as a deformable Finite-Element mesh attached to the mandible and interacting with the vocal tract boundaries. It includes 7 muscles responsible for the main deformations and displacements of the tongue in the mid-sagittal plane of the head (Perrier et al., 2003). This model is coupled to an acoustic analog of the vocal tract (Story, 2005).Muscle force generation mechanisms are implemented according to the lambda-model proposed by Feldman in the context of the Equilibrium Point Hypothesis (Feldman, 1986). For each muscle a control variable lambda is specified for each phoneme of the sequence. Thus the movement between two phonemes is depending on the values of the control variables at the phonemes, without any specification of the trajectory between the phonemes: in that sense, GEPPETO can be seen as a "target-based- mode l of speech production". The model of sequence planning selects an optimal sequence of control variables at the different phonemes of the sequence, aiming at reaching for each phoneme specific target-related acoustical characteristics. The search for optimality involves the use of internal models of the relations between motor commands and the acoustical characteristics of the speech signal (Perrier & Ma., 2008). - References - *Feldman, A.G. (1986). Once more on the equilibrium-point hypothesis ("Lambda" model) for motor control. Journal of Motor Behavior, 18(1), 17-54. *Perrier, P., Payan, Y., Zandipour, M., & Perkell, J. (2003). Influences of tongue biomechanics on speech movements during the production of velar stop consonants: A modeling study. Journal of the Acoustical Society of America, 114(3), 1582-1599. *Perrier, P. & Ma., L. (2008). Speech planning for V1CV2 sequences: Influence of the planned sequence. Proceedings of the 8th International Seminar on Speech Production (ISSP 2008) (pp. 69-72). Université de Strasbourg, France *Story, B.H. (2005). A parametric model of the vocal tract area function for vowel and consonant simulation. Journal of the Acoustical Society of America, 117(5), 3231-3254