Estimating Facial Neuromotor Activity from sEMG and Accelerometry for Speech Articulation

The aim of this study is to generate a multivariate representation of speech articulation by synchronous capturing of surface Electromyographic activity on the masseter muscle (sEMG), Accelerometry on the Lower Jaw Prominence (LJP) and Microphonic Speech Signal (MSS). The purpose of this multivariate description is to validate a Neuromechanical model of the lower Jaw-Tongue System (LJTS) which may be used to help in diagnosing and monitoring neurodegenerative diseases. A general-purpose data acquisition platform is set up to capture sEMG signals from the masseter muscle, accelerometry on the chin and speech. A high degree of correlation might be observed between sEMG and the displacement of the Jaw-Tongue reference point estimated from accelerometry. A strong correlation might be observed as well between sEMG and formant kinematics derived from speech. These promising results can be used in the characterization and monitoring of neuromotor degeneration diseases.