Short answer: every LiveInterpret.AI event has
a caption page made for screens: your event address followed by
/display.html. Open it full-screen on the projector computer,
or add it as a browser source in OBS or a Web Browser input in
vMix with a transparent background, and the translated captions appear as a
lower third. No plugin, no app, no extra login. It shows nothing at all
while the event is not live, so it can stay in your scene permanently.
theme=transparent as an OBS/vMix overlay.theme=dark full-screen on a projector.Start from the address your audience QR code opens, add
/display.html, then the room and the language:
Leave lang out and the page shows a language picker once;
after you pick, the address in the browser bar already contains your
choice, so copy that one into OBS, vMix or a bookmark. Every setting below
is optional.
| Setting | Values | Default | What it does |
|---|---|---|---|
room | your room name | the only room | which hall, if your venue runs more than one |
lang | language code, e.g. en, es, uk, pt-BR | asks once | caption language (any of the 39) |
theme | dark, transparent, light | dark | black projector screen, overlay for a stream, or white |
lines | 1–3 | 2 | how many caption lines are on screen |
chars | 20–80 | 42 | characters per line (42 is the broadcast norm; use ~24 on a phone) |
size | 2–12 | 5 | text height in % of screen height |
side | target, source, both | target | translation, the original speech, or one line of each |
align | bottom, center | bottom | lower third or the middle of the screen |
hold | 1–30 seconds | 4 | how long the last line stays after the speaker pauses |
font | sans, serif, mono | sans | typeface |
margin | 0–20 | 4 | safe-area margin in % of screen height |
theme=transparent, for
example …/display.html?room=main&lang=en&theme=transparent.theme=transparent, size 1920×1080.theme=dark and key the black out.theme=dark (or light for a bright room)
in Chrome, Edge or Firefox.ProPresenter: there is no native integration today. Teams run the caption page on a separate output, or bring the browser window into their mix the way they capture any other browser. NDI output is on our roadmap.
A caption screen is a listener. If someone in the room is already listening in that language, the screen costs nothing extra. If not, the screen starts that language and it uses minutes like any listener while the event is live. The page says so in a short notice the first time it starts, so nobody is surprised.
Pastors' and guests' names are what a congregation notices. In the
operator dashboard under Names & terms, list names one per line,
or wrong => Right for a fix. They are corrected in captions,
on caption screens, in the event transcript and subtitle downloads.
Spoken translated audio is not changed by this list.
Nothing. The page stays blank while the event is offline, and if the translation pauses or reconnects it clears rather than freezing on old text. Every caption block carries a small "AI-generated translation" label, as EU rules require for AI output shown to the public.
Yes. Open the caption address once per screen; each one can use its own theme and size. Your operator dashboard shows how many caption screens are connected.
One translation language per screen. side=both adds a line
of the original speech above it; for a second translation language, open a
second screen.
No. It is a normal web page; OBS and vMix load it like any other browser source. A free rehearsal in practice mode lets you set the scene up before a real service.
Author: Ondřej Plevka, builder of LiveInterpret.AI. Questions about your setup: reply to any of our e-mails.