Sigma-Lognormal Modeling of Speech

Abstract Human movement studies and analyses have been fundamental in many scientific domains, ranging from neuroscience to education, pattern recognition robotics, health care sports, beyond. Previous speech motor models were proposed understand how is produced the resulting varies when some parameters are changed. However, inverse approach, which muscular response subject’s age derived real continuous speech, not possible with such models. Instead, handwriting field, kinematic theory of rapid human movements its associated Sigma-lognormal model applied successfully obtain parameters. This work presents a kinematics-based that can be used study, analyze, reconstruct complex kinematics simplified manner. A method based on describe parameterize asymptotic impulse neuromuscular networks involved as neuromotor command. The carry out transformations formants observation also presented. Experiments carried (English) VTR-TIMIT database (German) Saarbrucken Voice Database, including people different ages, without laryngeal pathologies, corroborate link between extracted aging, one hand, proportion first second required applying movements, other. results should drive innovative developments modeling understanding kinematics.