NVIDIA's Signs turns any webcam into an American Sign Language tutor

American Sign Language is the third most prevalent language in the United States, but it is almost invisible to AI: far fewer tools are built with ASL data than with English or Spanish. The gap matters most for families — most deaf children are born to hearing parents, who need a reliable way to learn signs to communicate with children as young as six to eight months old.

NVIDIA, the American Society for Deaf Children and Hello Monday/DEPT answered with Signs, a free interactive web platform. A 3D avatar demonstrates each sign from a library validated by fluent ASL users and interpreters, and an AI tool watches the learner's webcam to give real-time feedback on hand and finger positions. Users can learn on any device, and fluent signers can contribute videos, turning every session into data collection for a growing public ASL dataset.

The platform launched with 100 signs and a target of 400,000 video clips representing 1,000 words, validated as both a visual dictionary and a training resource for accessible AI — including video-conferencing sign recognition and digital-human applications. At the 30th Annual Webby Awards, Signs won in three categories, including one People's Voice Award, as an example of technology built with an underserved language community rather than for it.