A congress followed in fourteen languages
The speaker talks from the lectern. Everybody in the room listens in their own language on their own phone — no booth, no receivers, nobody handing anything out at the door.
A congress with an international audience has two usual ways out: pay for booths and interpreters per language and per day, or let half the room follow the talk at half strength. This is a third, and it does not replace a human interpreter for everything — that is said below, where it belongs.
Where the audio comes from
From the room's desk or the lectern. A congress already has a sound engineer and a desk: an auxiliary output is taken and sent. If there are parallel rooms, each one is its own room in the panel, with its own languages and its own link.
How it is set up
- 1
The engineer takes an auxiliary output from the desk to the box, or the audio of the software already streaming the event is connected.
- 2
One room is created per physical room, with the languages for those sessions.
- 3
The code goes in the programme, on the screen before the session starts, and on the congress website.
- 4
Afterwards, each session's transcript is available and doubles as minutes.
Why it fits a congress
The cost does not grow with the room
Thirty people listening or a thousand, it is the same: what is billed is the speaking, not the listeners. With booths, each extra language is two more interpreters.
Parallel rooms
Each room is independent, with its own languages and link. No multiplying equipment or staff per room.
Decided the day before
Adding a language is a change in the panel, not a hire. If a delegation confirms at the last minute, you add it and that is that.
The text stays
Each session leaves its transcript with timings, in every language that was open. Good for minutes, for subtitling the video, and for the congress site's own search.
When you need a real interpreter
In a negotiation where one word changes a contract, in a hostile press conference, or when the speaker improvises, interrupts himself and jokes with the room. A human interpreter understands intent and decides; a machine translates what was said. For a prepared technical talk the difference is small; for a tense panel it is not.
Questions people ask
- Does it work if the speaker is fast or has an accent?
- With an accent, well: it is trained on real voices rather than studio ones. In a hurry, also — though the more speech runs together the more it blurs, which happens to a person listening too.
- What if the speaker switches language halfway?
- A room has one source language configured. If somebody switches, that part comes out worse. For sessions where it is known to happen, the sensible thing is one room per speaker language.
- Can the audience ask questions and have them translated?
- Yes, if the question goes through a microphone connected to the desk. What does not reach the desk is not heard.
- What does a two-day congress cost?
- It is worked out on speech: two eight-hour days with five hours of actual talking each is ten hours of speech. In four languages, those ten hours are 20.5 hours of credit. The pricing page has the sum with each pack's numbers.