A Highly Flexible Self-Powered Triboelectric Sensor Array for Silent Speech Recognition and Swallowing Motion Analysis.

The growing prevalence of speech and swallowing disorders necessitates the development of advanced, non-invasive technologies for effective communication and rehabilitation. Conventional silent speech recognition (SSR) methods, including vision-based, ultrasound, inaudible acoustic, and surface electromyography (sEMG) approaches, suffer from limitations such as sensitivity to lighting conditions, occlusions, motion artifacts, and reliance on external power sources, restricting their applicability. Similarly, gold-standard swallowing assessments, including videofluoroscopic swallowing study (VFSS) and flexible endoscopic evaluation of swallowing (FEES), are invasive and unsuitable for continuous monitoring. To address these limitations, we introduce a highly flexible, self-powered tactile sensor array based on triboelectric nanogenerator (TENG) for SSR and swallowing motion analysis. The sensor comprises a microstructured polydimethylsiloxane (PDMS) layer and an electrospun Nylon 6/6 nanofiber film optimized for triboelectric charge generation and mechanical stability. Integrated within a 2×2 matrix, the TENG sensor array accurately captures lip and laryngeal movements. Machine learning analysis enables accurate silent speech-based user authentication (97.06%) and high-precision classification (98.04%) of critical swallow rehabilitation maneuvers, including the supraglottic swallow, Mendelsohn maneuver, and super-supraglottic swallow. This TENG-based sensor array offers a robust, non-invasive, and self-sustaining solution for real-time speech and swallowing analysis, establishing a foundation for next-generation wearable assistive technologies bridging clinical diagnostics and rehabilitation.