A Major Project On "Human-Computer Interaction using Neuromuscular Signals"
Subvocal speech or internal articulation is a form of non-voiced speech that is voluntarily spoken. It is generated alongside the micromovement of the articulatory muscles, imperceptible to others. However, the faint sEMG (surface Electromyography) signals can still be detected and analyzed to predict the internally articulated speech. This research project attempts to study this very phenomenon and its possible use case in human-computer interaction. By extracting sEMG signals from several of these articulators and processing the extracted signals, prominent features of a particular utterance can be isolated. They can be used to train a machine-learning model. After training the model on several such utterances, accurate predictions of the utterances can be made which can be further utilized to perform a predefined action on a remote computer. This research also explores improving traditional speech recognition models by possible augmentation of both approaches.
Rabin Nepal https://github.com/Bot-Ro-Bot
Rhimesh Lwagun https://github.com/riderman10000
Sanjay Rijal https://github.com/zovelsanj
Upendra Subedi https://github.com/upendrasubedi
@InProceedings{IOEGC-12-082-12118,
author = {Rhimesh Lwagun and Sanjay Rijal and Rabin Nepal and Upendra Subedi and Dinesh Baniya Kshatri},
title = {{Silent Speech Recognition in Nepali}},
pages = {628 -- 635},
booktitle = {Proceedings of 12th IOE Graduate Conference},
year = {2022},
volume = {12},
month = {October},
organization = {Institute of Engineering, Tribhuvan University, Nepal}
}