Rylo Collaborates With Nvidia on AI Sign-Language Research

Israeli startup Rylo has disclosed a collaboration using Nvidia computing resources to train artificial intelligence models for sign-language translation. The research aims to convert sign language into text or speech in real time, rather than offering an already completed translation system. Researchers are using Nvidia’s Cosmos Tokenizer to create compact representations of sign-language video for model training. Rylo, formerly known as Nagish, is pursuing the work under research chief Dr. Amit Moryossef, who said the additional computing capacity enables faster training at a larger scale.