SignSpeak — Camera-based gesture translation

Camera

Turn on the camera to begin

Camera is off. Turn it on to start recognizing gestures.

FPS: —

Waiting for camera

«—»

Confidence

Gesture detected: none

These gestures are demonstration gestures for the prototype and are not intended to represent a complete sign language.

This prototype only covers camera → landmarks → gesture → word. These are not built yet:

Sentence generation. Turning a sequence of gestures into a natural sentence, not just single words.
Two-way communication. Speech-to-text so a hearing person can reply in text.
Custom gestures. Letting a person record and label their own signs.
On-device privacy mode. Confirmation that no frames ever leave the browser.