Live translation on your audience’s own phones

How it works

• Start a session. Put the phone or laptop running All Voices where it hears the speaker clearly. Start the session as usual, in the Android app or in the browser.
• Open it to the audience. Tap Audience. All Voices shows a QR code, a link you can share, and a six-digit code that people can type at allvoices.app/join. On the web you can also print a sign with the code.
• Listeners choose a language. The join page asks for one language for captions and voice. Two listeners can choose two different languages; each follows the same talk in their own.
• Read or listen. Captions appear line by line as the speaker talks. Anyone can turn the sound on to hear the translation, and turn it off again at any time.

What listeners see and hear

• Captions in the language they chose, with the speaker of each line. Lines already in their language are shown as spoken.
• "Show original" puts the speaker’s own words beside the translation, for people who understand some of the original.
• Spoken translation uses the speaker’s own voice once enough clear speech is available, with a stock voice until then. A listener can choose the phone’s own voice instead.
• Text size, and an option to show only finished sentences for calmer reading.
• Their microphone stays off. It turns on only if they raise a hand, you invite them, and they agree.

Listeners in the same room as the speaker should use earphones for sound, so that the microphone does not pick up the translation.

What you control

Your screen shows how many people are listening, in which languages, and how fast the audience is using your hours. From the listener list you can:

• pause new joins, and allow them again;
• create a new code: people already listening stay, while the old code and QR stop working;
• remove a listener;
• stop the audience while your own session continues.

Audience questions. A listener can raise a hand. When you tap Allow, they choose the language they will speak, read what happens to their voice, and agree before their microphone turns on. One person speaks at a time. The translation of what they say is spoken to you and to listeners who read another language, and the turn is saved in your meeting.

Languages and limits

• One session takes up to ten listener languages. The first six get their own live translation. Languages seven to ten are translated in short batches, so their captions can arrive a little later.
• Spoken translation for listeners is available in up to three languages per session. Listeners in further languages read along, and the page tells them so.
• Some languages are marked experimental in the language list. Their results can be less reliable.
• On the free trial, up to three listeners can join.
• Audience listening is built for the room in front of you: a class, a congregation, a team meeting or a talk. For hundreds of listeners in many languages at once, a conference interpretation platform is the better fit.

What it costs

Listeners pay nothing. Their time comes out of your balance, on top of the session itself:

• each reading listener uses 12 minutes of your balance per hour;
• from the eleventh listener, 6 minutes per hour; from the fifty-first, 3 minutes per hour;
• a listener with sound on uses double;
• your first two reading listeners are free.

Example: a one-hour talk with ten people reading along uses the hour of the session plus 96 minutes for eight listeners (the first two are free), so 2 hours 36 minutes in total. Your session shows the burn rate while it runs, and the saved meeting shows the exact minutes. Hours come in prepaid packs that never expire; see pricing.

After the talk

You keep the session as a meeting: the full transcript with each line’s original and translation, the recording, and a summary with decisions and action items. The meeting also records how many people followed, in how many languages, and the peak at one time.

Listeners keep nothing on their phones. To give them a record, share the meeting by link: each reader opens the summary, and the transcript if you include it, in the language they choose. See meeting notes in two languages.

An audience link closes one hour after your last live session. Each new talk gets a new code.

Before you start

• Tell the audience that the talk is transcribed, translated and recorded. In Switzerland and many other countries, everyone present must agree.
• This is personal translation, not official interpretation. Where the stakes require it, use a qualified interpreter.
• Live captions need an internet connection on your device and on each listener’s phone. If your connection drops, All Voices keeps the recording while it reconnects.

Dated comparison: an event translation platform.