Latency
Text within half a second
Words appear as they’re spoken — under 400 ms behind the voice — so following the room in text feels live, not delayed.
Live captions for the room you’re in.
Open a tab, press record, and speech becomes readable text as it happens — for the attendee who needs captions today, in the room as it actually sounds.
- Speaker-labelled, as you speak
- Holds up in a loud room
- Audio is never kept after a session
Captioning shouldn’t require a booking
Professional captioners are excellent and scarce: booked in advance, budgeted per hour, and out of reach for the Tuesday seminar, the walk-in meeting, the volunteer-run event. The result is that most rooms simply go uncaptioned.
Automatic live captions are not a replacement for professional access services where they’re required — but they are radically better than the nothing most rooms currently offer.
What you get
Used by accessibility services, event teams and educators.
Real rooms
Noise doesn’t end it
A seminar room fan, corridor noise, an audience — the captions keep going in conditions that stop tools tuned for quiet.
Afterwards
The transcript remains
Captions vanish; this leaves a speaker-labelled, timestamped transcript — the same session doubles as notes for everyone.
Questions
Are automatic captions accurate enough for accessibility?
For many rooms they’re transformative; for formal accommodations, follow your institution’s requirements — automatic captions supplement rather than replace professional services where those are mandated. Accuracy is honest-to-goodness dependent on audio: a close microphone helps enormously.
How do I show the captions to the room?
The transcript renders live in a browser tab, so anything a tab can do works: full-screen it on a laptop for one attendee, or put the tab on the projector for everyone.
What’s the delay between speech and text?
Words land in under 400 milliseconds — fast enough that captions track the speaker rather than trailing a sentence behind.
Further reading
Also good for