Questions

Live interpretation for events. Someone speaks into one microphone, and everyone in the room hears them in their own language, as audio and captions on their own phone, a few seconds later.

No. Speech is transcribed, translated and voiced by AI services, sentence by sentence. That is what makes it fast and able to cover languages few events could staff — and it also means it will sometimes mishear, mistranslate or mispronounce. Don’t rely on it where a mistake could cause real harm.

About one sentence. Each sentence arrives a few seconds after the speaker finishes it — around three seconds in our measurements — much as a human interpreter waits to hear where a sentence is going.

Yes — watch one speaker reach three languages at once on the homepage. Everything on the phones there is exactly what an attendee receives.

No. They scan the event’s QR code or enter its code at infinispeak.com/live, and it opens in the browser they already have. No app, no account, no email address.

A phone with a web browser, earbuds or headphones, and a connection — the venue’s Wi-Fi or their own data. They tap Listen once; the audio keeps playing with the phone locked in a pocket.

About 11–13 MB for an hour of listening. Anyone paying for data can tap “Read the captions instead” on the listening screen: captions alone use under 1 MB an hour.

Yes, when the organiser turns on “Let attendees talk”. A Talk button appears on everyone’s phone: hold it, speak in your own language, and the room hears you in theirs. One person at a time; others raise a hand and are called in turn, or the organiser chooses. Speakers can use any of 71 languages, from Arabic, Hindi and Swahili to Japanese and Portuguese; listeners can choose any language the event offers.

Yes, with Listen for me — at any talk, service, lecture, tour or meeting where someone is speaking a language you don’t understand. Your phone hears them and interprets into your earbuds. It works best when you can hear the speaker clearly, because your phone can only interpret what it hears. Your first 15 minutes are free; after that it’s $9 a month for 3 hours, or $15 for 6 hours to use within 30 days. It stops by itself after two hours, or after ten minutes of silence. Try Listen for me.

Tap “More languages” and search. Depending on how the event is set up, you can add it yourself and it starts within seconds, or ask for it and the organisers see your request.

A laptop running the operator page, with a feed from your sound desk or a good microphone, and a reliable internet connection. Press “Check everything works” at least ten minutes before you start. Running a session in Help walks through the rest.

Attendees don’t have to use the internet at all. In Venue Mode, InfiniSpeak runs from a laptop in the room and phones join a local Wi-Fi network; only the laptop needs a connection, and a phone’s hotspot is enough. Talk to us and we’ll set it up for your venue.

No. One translation per language serves everyone who chooses it, so five listeners or five hundred cost the same and put the same load on interpretation.

Yes. Add them to your glossary as words not to translate — people’s names, your organisation, programme titles — and they are kept exactly as written in every language. Or upload the talk notes (PDF or Word) and we suggest the names for you to tick.

You can choose male or female voices for an event. A few languages offer only one; those use their default.

Captions and audio travel separately, so if a voice fails the words still arrive as captions. One language having trouble doesn’t affect the others, and running out of hours never stops a session that’s already live.

101 languages to listen in, 98 of them spoken aloud; the rest arrive as captions. A speaker can use 71 of them. That includes 16 languages most voice services leave out — Yorùbá, Hausa, Nigerian Pidgin, Akan, Oromo, Kinyarwanda and more — which we voice ourselves.

You pay for languages being listened to, not for attendees. One language-hour is one language being listened to for one hour, and a language nobody selects costs nothing. Plans start at $79 a month, and event packs cover one-off events with no subscription. See pricing.

It is sent to AI services to be transcribed, translated and voiced — they are named in our privacy policy. The interpretation audio is deleted within 24 hours, and we do not store transcripts.

Almost nothing. They have no account, their language choice stays on their own phone, and we keep a count of listeners, not a list. Their microphone is never used unless they hold Talk at an event that allows it — and even then nothing is recorded.

Leave feedback — one tap is enough, and it’s anonymous unless you add your email. Attendees can also use the “How is it going?” link under their captions.

Help is organised by what you’re seeing, for listeners, operators and organisers. Or write to hello@everytongue.net — a person reads every message.