SignType V6 — signing workspace

GPU Required: SignType turns your camera into a local sign-to-text workspace. Click into almost any text field, start the camera, and write by signing. SignType tracks your hands, face, and body entirely on your device—no video is uploaded or recorded. Features • Hand, face, and body tracking with live framing guidance • Trainable word signing: record examples of your own signs and build a personal vocabulary • Experimental ASL fingerspelling with draft review before inserting text • Bring your own model: load supported ONNX models locally with WebGPU, load trained word models, or connect your own translator endpoint • Dedicated SEND gesture: hold both open palms toward the camera to insert and send • Draft-first editing, direct typing, backspace, word deletion, newline, and undo controls • Works with many ordinary text inputs, textareas, and compatible rich-text editors SignType is an evolving accessibility project. Word signing and fingerspelling are trainable/experimental, and continuous fluent sign-language translation requires a compatible user-supplied translator model.