Sign to build a sentence, then speak it. Replies are captioned live.
Camera off
Sentence builder
Words recognized from your gestures collect here
No words yet — sign a gesture to begin.
Transcript
Both sides of the conversation, in order
chat_bubbleNothing said yetYour signed sentences and their spoken replies will appear here.
Classroom mode
Large, high-contrast captions built for the front of a room.
Waiting for a gesture or a spoken reply
Turn on classroom mode to project captions here.
Why classroom mode
Text scales up, colors simplify to a single high-contrast panel, and a raised-hand signal lets a student
flag that they want to speak next without saying a word — useful when a teacher is addressing the group
and a student is composing a longer thought.
Emergency phrases
Fixed phrases that speak instantly. No network call, no delay.
wifi_offWorks fully offline — these phrases never depend on a network connection.
Automatic trigger
A hands-free way to call for help
Holding a closed fist steadily in view of the camera for about two seconds speaks the top emergency phrase
automatically, even if this tab is showing another screen. A ring around the camera preview fills as it counts down.
Custom gestures
Teach SignBridge a personal gesture and the phrase it should mean.
Add a gesture
Saved gestures are checked before the built-in library
Enter a phrase, then press and hold the record button while making the gesture in front of the camera for two seconds.
Saved gestures
0 saved this session
front_handNo custom gestures yetGestures you record are kept for this session, on this device.
Settings
Language, sensitivity and privacy controls.
Language
Conversation languageUsed for spoken output, live captions, and composed sentences
VoiceText-to-speech voice for composed sentences
Recognition
Confidence thresholdGestures below this confidence are flagged as unsure instead of guessed
60%
Hold time to confirmHow long a gesture must be held before it is accepted
550ms
Camera
Camera accessNot started
Privacy
Hand landmark detection and gesture matching run entirely in this browser tab. Video frames are never
transmitted anywhere. The only data that ever leaves this device is the short list of recognized words,
and only when you choose to compose an AI sentence — emergency phrases and instant-speak gestures never
make a network call at all.