When an international event has no interpretation, half the room takes home half of every talk. But hiring two interpreters per language, renting booths and handing out receivers is out of reach for far more events than it is within. This is the preparation sequence for doing it with live subtitles instead.
1. Decide the languages first
The most common mistake is switching on as many languages as possible. Every language you put on screen takes floor space from the others, and past a certain point the type is too small to read from the back row.
Take the country and language split from your registration data, and for anything under about 5% of attendees, serve it by audience QR rather than on the screen. Three or four languages on the wall and up to eight on people's phones is the arrangement that works in a real hall.
2. Decide how you will declare the spoken language
If a speaker stays in one language, naming that one language is the most accurate thing you can do. When a talk genuinely moves between languages — a Korean presentation with English terms and whole English sentences in it — you can select several spoken languages.
Selecting several adds a language-identification step, which costs some accuracy and some latency. The detected language is shown live as a flag on the original panel, so if the flag flips to the wrong country during rehearsal, cut the number of spoken languages.
3. Screen layout and seating
The screen divides itself according to how many languages you have. The original-language panel always gets a distinguishable position and carries the audio waveform, so the audience can see where the sound is coming from.
- Three languages — four panels including the original, side by side
- Four languages — the original tall down the middle, two on each side
- Six languages — the original full-height on the left, a 3×2 grid on the right
- Seven languages — a 4×2 grid
With only one screen, show the original plus your two biggest languages and move the rest to QR. A subtitle nobody can read is not a subtitle.
4. Decide where the QR code appears
Put it in three places along the way in: the registration desk, the seat cards, and the opening slide of each session. The presentation screen has a QR button that fills the display, and holding that up for one minute before a session starts is enough to get most of the room connected.
Attendees choose a language in the browser — no app, no account. Someone who arrives late picks up from the most recent lines.
5. What the rehearsal has to cover
- 1Have the speaker talk into the actual microphone for two minutes and check that confirmed sentences keep accumulating without stalling.
- 2Read the subtitles from the back row and adjust the type size (the +/− buttons at the top of the screen).
- 3Check how names and technical terms come out. If one is reliably wrong, add it to the note glossary and tell the speaker which words to enunciate.
- 4Scan the QR with two phones, pick different languages on each, and see for yourself how many seconds behind it runs.
- 5Test on the venue Wi-Fi. Results on your office network tell you nothing.
6. After the event
The whole programme is left as a transcript. Summarise each session with AI, export to Word, PDF or Markdown, and the post-event pack and the press release draft are already written.
The translations are stored alongside, so the follow-up email to overseas attendees can quote them directly.