Black TikToker Develops Game-Changing Algorithm: Translating Kenyan Sign Language To Audio
A Black TikToker has built a machine-learning algorithm that translates Kenyan sign language into audible language.
What Has She Developed?
The TikToker, under the username @Mrembo, uses mathematical instructions to translate different sign language dialects.
She took to TikTok to show a quick demonstration of the process.
“First things first, I had no idea about anything to do with machine learning,” she explained, as she was initially unfamiliar with the type of AI.
With the help of YouTube and a machine-learning course, the creator learned to build her machine-learning model.
“We were able to get our hands to get recognized and finally translate words and sentences from Kenyan Sign Language,” she said.
According to the World Federation of the Deaf, more than 300 sign languages are spoken by more than 70 million deaf people worldwide.
In the US alone, American Sign Language is the third most commonly used language after English and Spanish.
Other Sign Language Innovations
25-year-old inventor Roy Allela from Kenya invented smart gloves called Sign-IO to make communication with the Deaf community effortless.
The gloves use flex sensors on each finger to convert sign language movements into audio speech as each sensor quantifies the bend of the fingers and processes the letter being signed.
A fourth-year computer science student specializing in data science at the Vellore Institute of Technology, Priyanjali Gupta, also went viral on LinkedIn after using AI to translate American Sign Language.
She made the model using Tensorflow Object Detection to translate a few American Sign Language signs to English using transfer learning from a pre-trained model by Australian data scientist Nicholas Renotte.
Renotte told Business Insider he created a program that can also read a person’s sound waves and facial expressions as they speak to determine their feelings.
Researchers in Ireland estimate that 70% of sign language comes from facial expressions.