Surface Electromyography for Silent Speech Recognition: Applications in Physiotherapy and Assistive Technologies

This study explores the classification of electrical signals from facial muscles, captured through non-invasive surface electromyography (sEMG), for the purpose of silent speech recognition. The research is particularly relevant to the field of physiotherapy, where such technology could provide innovative solutions for individuals with speech impairments resulting from conditions like Multiple Sclerosis with Dysphonia. By converting muscle activity into text without the need for vocalization, this technology offers a discreet, privacy-preserving method of communication. The project aims to advance the existing methodologies in sEMG-based silent speech systems by improving model performance on key metrics such as word- error rate (WER) and inference time. Additionally, it explores the potential applications of silent speech systems in rehabilitative practices, offering new avenues for assistive communication technologies in physiotherapy. Through the development and open- source dissemination of an enhanced sEMG model, this research contributes to both the technological and therapeutic landscapes, providing a foundation for future work in the intersection of physiotherapy and human-computer interaction.