A Stretchable High-Density Electromyography Mems Sensor for Efficient Silent Speech Recognition

This paper reports a novel high-density surface electromyography (HD-sEMG) MEMS sensor with high stretchability for silent speech recognition. A 21-channel annular sector-shaped electrode array is proposed with the serpentine structure to realize flexible interconnection, which can be repeatedly stretched under 40% strain without failure, perfectly adapting to the skin strain of the lower jaw and ensuring the movement of muscles without restriction. The sensor can achieve tight adhesion by modifying electrodes with self-adhesive hyaluronic acid hydrogel and the skin-electrode contact impedance is reduced six-fold compared to the dry electrode. From the results of the collection of signals, it is obvious that the envelopes have similar characteristic peaks and valleys when speaking the same word, but differ when speaking different words. This high-density electrode array improves the spatial resolution of signal acquisition to reliably provide multi-channel muscle activation information and has great potential in the application of achieving efficient silent speech recognition (SSR).