Ultrathin Soft Dry Electrodes for Decoding Silent Speech

This project develops an unobtrusive silent speech recognition system using skin-conformable, self-adhesive, semi-transparent dry electrodes to capture high-fidelity electromyogram signals. Machine learning algorithms decode these signals into spoken words, ensuring robust performance in noisy and dark environments. The system supports real-time analysis through cloud services and MQTT methods. Its applications in augmented reality and medical services highlight its potential to enhance immersive interaction and healthcare.

This video shows the silent speech-related EMG signal is decoded in real-time to control the movements of an AR character.

This video shows the silent speech-related EMG signal is converted into spoken words for in real-time for healthcare applications.