SignrrGPT is an open-source, real-time, bi-directional sign language interpreter that connects deaf and hard-of-hearing people directly to AI. A camera reads a person's signing using a fine-tuned VideoMAE vision model, and a large language model then takes the top predictions for each sign and works out the most coherent meaning as natural language. A 3D avatar, animated from a MediaPipe motion-capture library, signs the response back, so users never have to read or type. Unlike most tools, which only read the hands, SignrrGPT also captures facial expression and body movement, where much of sign language's meaning actually lives. Over 1.5 billion people live with hearing loss, and many communicate through sign language, a full language with its own grammar. As the world moves to AI for learning and work, signers risk being excluded again. SignrrGPT exists to take that barrier down. It is a working prototype today; funding goes toward improving recognition accuracy and expanding vocabulary so deaf users can genuinely rely on it.

Fund this project

Unverified URL

The funding manifest has not provided proof via wellKnown that this link is associated with it. Learn more.

Continue