Nothing kept
No transcript is saved anywhere. Lines show for a few seconds and go.
Live captions
Turn on captions in a call and words appear as people speak. Each person's own device turns their speech into words, and nothing is kept.
Captions are in English for now
Press CC beside React in a call. Captions are yours alone, so turning them on changes nothing for anyone else.
Each line shows for a few seconds, then goes, so it reads like the conversation it is.
A speech model runs in the Grumble app on each person's computer. It comes from Grumble's own server once, then stays on your device.
Words go only to the people who want them, sealed like everything else in the call. Someone on a phone is captioned by the device of whoever asked, from their voice as it arrives.
Private by design
No transcript is saved anywhere. Lines show for a few seconds and go.
The speech model, Moonshine from Useful Sensors, runs on your own device. Your voice never goes to an outside service.
Words travel between devices in the call, encrypted like the call itself. Grumble's server can't read them.
A device only turns speech into words when someone in the call has captions on.
Captions show for whoever turns them on, whatever device the speaker is on.
Live captions are in every call, with nothing to buy.
How to
Any voice channel or call with friends.
It's beside React in the call. The first time, the speech model loads once.
Words appear under the call as people talk.