UKZN engineering graduate creates real-time South African Sign Language translator

A University of KwaZulu-Natal engineering graduate has created an AI system that translates South African Sign Language (SASL) into speech in real-time. Akhil Hansrajh, who developed this SASL-to-Speech Translation System as part of his BSc Engineering final-year design project, aims to alleviate communication challenges faced by the hard of hearing community.

The computer vision-based system utilizes a custom-built artificial intelligence model and relies on an established machine learning framework to track body landmarks and identify signs in real-time. The technology has been trained on a dataset specific to SASL and can recognize individual letters and words before converting them into speech via a text-to-speech model.

Dr Sulaiman Patel, Hansrajh's supervisor, noted that this project is among the first assistive technology initiatives in UKZN's Electrical, Electronic, and Computer Engineering discipline to concentrate on the South African dialect of sign language. SASL has recently been recognized as South Africa's 12th official language, while many similar sign-language translation systems globally focus on American Sign Language.

Hansrajh's motivation to develop this technology stems from his personal experience, as both his parents are hard of hearing and have encountered communication barriers in everyday situations. He hopes that a fully developed version of the system will empower individuals to communicate more independently, without requiring the other party to be familiar with SASL.

Despite facing mental health challenges during his studies, Hansrajh persevered with the support of his supervisor, Dr Patel, and ultimately earned a highly commended award for his final-year project. Currently, Hansrajh plans to further refine the translation system, focusing on its ability to recognize words and produce more fluent speech.

In the long term, he aspires to pursue a career in software, networking, or artificial intelligence and machine learning, crediting his family, friends, family, and university colleagues for their unwavering support throughout his journey.