Live captions over your slides in a language the room reads — and the same words on the speaker's stage display. Two methods, the real parameters, and what to check when nothing shows.
One disambiguation first. If you are after Bible translations — switching between ESV, NIV or Folkbibeln inside ProPresenter — this is the wrong page; Renewed Vision documents that feature. This page is about the other kind: someone speaks from your platform and the words appear on screen in Spanish, Arabic or Ukrainian while they are still talking.
Tolka Live is real-time AI speech translation for live events. It listens to your speech microphone and puts the translated text in a ProPresenter Prop, on the stage display, in OBS as live subtitles, and on the audience's phones via QR code. It runs in the browser — nothing to install on the ProPresenter machine. The setup guide has screenshots of every dialog below.
ProPresenter live captions can come from two places: a hosted web page, or ProPresenter's own text engine. They differ in who draws the text and in what happens when the operator clears the screen.
https://tolkalive.se/overlay/?id=YOUR_ID. It is permanent — paste it once and it works at
every future event.Translation.&text=Test caption to the URL, show the
Prop on the real projector and judge it from the back row. Adjust size and
bottom until it reads and clears your lyrics, then remove the parameter.Why this one. A cleared Prop stays cleared — Clear All is final until a human shows it again — and captions persist across slide changes, because the Prop is its own layer. The page is hosted, so a stream team in another building can use the identical link.
Choose this when you want your own Message theme to draw the captions, or when the Prop layer is already busy.
Translation.Translation in the token box and press Enter so it
becomes a token. Style it with a Theme.localhost, or the other
computer's IP) and that port.The audience screen and the speaker's stage display are separate outputs you can drive independently.
On the ProPresenter side: Screens → Edit Layouts…, add the Stage Message element and size it where the speaker looks. Without that element, nothing appears there.
The caption page takes a few optional parameters — the first after /overlay/ uses
?, the rest use &. Preview them in a browser tab:
https://tolkalive.se/overlay/?id=YOUR_ID&size=4&bg=none
| Parameter | What it does | Example |
|---|---|---|
text | Shows fixed text instead of the live feed — for testing. | &text=Test |
size | Text size in vh (percent of screen height). | &size=4 |
color | Text colour, hex without the #. | &color=ffffff |
bg | Backdrop behind the text: a hex value, or none for fully transparent. | &bg=none |
bgopacity | Backdrop opacity, 0 to 1. Default 0.45. | &bgopacity=0.7 |
bottom | Distance from the bottom edge, in vh. Raise it to clear a lower third. | &bottom=12 |
lang | Which language channel this overlay follows. | &lang=es |
&text= to the URL
to tell a link problem from an audio problem.id — paste it again from the portal.Translation,
or has no token. Audience but no stage: the Stage Message element is missing from the layout. Yellow
status: wrong host or port, or Network not enabled.75+ target languages, source language detected automatically, and up to five running in parallel from one microphone — each with its own permanent link and QR code, so the Prop can show one language while the room reads the others on their phones, with no per-listener cost.
Every plan includes every output on this page and they differ only in hours: from €55 a month for 15 hours excluding VAT, or a one-time Event licence if you only need a single conference covered. New accounts get 15 minutes free without a credit card. Details on the pricing page.
With the Prop overlay, yes. The Prop is its own layer, so captions keep running through slide changes, and a cleared Prop stays cleared until someone shows it again. The Message route differs: every new caption shows the Message again, so a clear is undone by the next sentence.
Yes, they are separate outputs. A Send to selector decides which receives the captions — audience, stage, or both — and a character cap keeps the stage text inside the rows your layout has.
Up to five run in parallel from one microphone, and each language has its own permanent caption link. A second Prop, with a web object pointed at the second link, shows that language on screen — and the room can read any running language on their own phones.
For the Prop overlay, no. The caption link is a hosted page on tolkalive.se, so the translating computer can sit anywhere with a stable connection — which is how remote stream teams use the same overlay. The Message and stage-display route uses the local network API, so it has to reach ProPresenter over your network.
More on accuracy, privacy, the AI voice and how hours are counted in the FAQ; the rest of the setup is in the guide.