Executive Summary : | The global market for wearable devices in the medical sector is growing, allowing for the collection and monitoring of vital physical parameters. One useful information gathered is the distorted voice signal of a differently abled person and its underlying text. Triboelectric nanogenerator (TENG) can effectively collect these distorted voice signals and generate electric signals through triboelectrification and electrostatic induction. Recent developments in machine learning (ML) fields provide researchers with the opportunity to perform tasks without manual intervention. ML or DL algorithms can be applied to extract underlying texts from distorted audio signals captured by TENG.
The research aims to fabricate a flexible, contact-separation mode TENG using two triboelectric materials, capture distorted audio from patients, extract features from generated voltage pulses, train machine learning or deep learning models using extracted features, and generate the underlying text of the distorted voice using the best trained model. The fabrication process requires standardized techniques like spin-coating and metallization, and the prototype's performance is measured through electrical characterization.
The research has the potential to develop a product to generate the underlying text of a distorted voice signal. |