communication at the speed of a gesture
Say it now
Without typing.
Without a voice.
AirSign turns a small set of starter hand gestures into spoken phrases for quick, high-stakes moments — a pharmacy counter, a reception desk, a loud room. It is a helper, not an interpreter.
sentence strip
starter gestures · not sign language
Three beats, one phrase.
01 — how it works
Make a starter gesture
Seven built-in handshapes map to phrases, and two-hand pairs unlock combo phrases. Starter gestures, not ASL.
Hold to lock
A ring fills while the pose is steady. Pairs need a longer hold, so a resting second hand will not fire.
Speak it
A natural voice says the phrase. Speak-on-lock is on by default; switch it off to confirm first.
Fast where it matters.
02 — engineering
Video stays on-device
privacyHand tracking runs in the browser. Only the confirmed phrase text is sent to the server.
The camera follows your hands
trackingThe preview eases toward the union of both hands so a small gesture or a pair still fills the frame.
Prewarmed voice
voiceStarter phrases are generated and cached when the workspace opens, so the first lock is near-instant.
Keyboard fallback
resilienceKeys 1–9 fire the same events as the camera. The demo survives bad lighting or a denied camera.
03 — limits
Honest by design.
AirSign does not translate sign language. The built-in poses are arbitrary starter gestures that we map to phrases, and the interface says so.
Nothing is spoken until a gesture locks. With speak-on-lock on, the phrase is spoken as soon as it locks — switch it off when you want an explicit confirm step.
It is a helper for small moments where no interpreter is available. The honest next step is co-design with Deaf users, not a claim that this replaces one.
Open the workspace.
Camera on, hand up, phrase spoken. Keys 1–9 work without a camera.