Akhil Hansrajh, a University of KwaZulu-Natal (UKZN) engineering graduate, has developed an artificial intelligence (AI) system capable of translating South African Sign Language (SASL) into speech in real time. The system, developed as his final-year design project, uses computer vision and a custom-built AI model to track body landmarks and identify signs. Hansrajh's innovation aims to help address communication barriers faced by people who are hard of hearing.
The SASL-to-Speech Translation System was trained using a custom dataset relevant to SASL and can recognise individual letters and words before converting the interpreted signs into speech through a text-to-speech model. According to Hansrajh's supervisor, Dr Sulaiman Patel, the project was among the first assistive technology projects in UKZN's Electrical, Electronic and Computer Engineering discipline to focus specifically on the South African dialect of sign language.
SASL was recently recognised as South Africa's 12th official language, and many comparable sign-language translation systems internationally focus on American Sign Language. Hansrajh's project was motivated by his personal experience of communication barriers faced by his hard-of-hearing parents. He said his parents often encounter difficulties in everyday settings such as banks, government departments, and municipal offices.
Hansrajh's road to graduation was not without challenges. He experienced mental health challenges during the second semester of 2025 and required an extension for his first project submission. With Patel's support, including referrals to student counselling and regular wellness check-ins, Hansrajh was able to regain momentum and continue with the project, ultimately earning a highly commended award from the EECE for his final-year project.
Hansrajh plans to continue developing his sign-language translation project, focusing on improving the system's ability to recognise words and produce more cohesive and fluent speech. In the longer term, he hopes to pursue a career in software, networking, or artificial intelligence and machine learning. He credited his parents, uncle, sister, and university colleagues for supporting him throughout his journey.
The development of the SASL-to-Speech Translation System has the potential to empower people who are hard of hearing to communicate more easily and effectively. Hansrajh's innovation could enable individuals to communicate independently, without requiring the other party to have learned SASL. The system may also have applications in various settings, including education, healthcare, and government services.
Key points
- The system uses computer vision and a custom-built AI model to track body landmarks and identify signs in real time.
- SASL was recently recognised as South Africa's 12th official language.
- Hansrajh's project aims to address communication barriers faced by people who are hard of hearing.
