Recognition
Display
Transcripts
Supported letters & limitations
Supported (static): A B C D E F G H I K L M N O P Q R S T U V W X Y — 24 letters.
Not recognized: J and Z (they are motion-based and this release has no temporal letter model — use the J/Z buttons or keyboard), numbers, and any non-alphabet handshape.
The model was trained on one contributor's hand shapes. People with different hand proportions may see lower accuracy. Common confusions observed in evaluation: M↔N/T, I↔Y, and occasionally A↔S.
Camera troubleshooting
Camera blocked? Camera access requires HTTPS (or localhost). Check your browser's site permissions — in iOS Safari: Settings → Safari → Camera; in Chrome: the camera icon in the address bar.
Tracking lost? Improve lighting, keep your hand fully in frame, avoid busy backgrounds, and hold ~30–60 cm from the camera.
Recognition feels wrong? Try raising Stability in Recognition settings, and make sure your palm faces the camera the same way the training data did (palm toward camera, upright).
Privacy & data
Camera frames are processed on-device and never uploaded or recorded. Settings and (if you opt in) transcripts are stored in this browser's localStorage only.