Atcha: AI Live Translation

Speak in your own language. Everyone in the hall reads you live in theirs, on their own phone. The audience needs no app, but can read in it.

Start a talk on your phone
Choose the language you speak, and tap Go live when you begin.

Show the QR code in the hall
Share the link, print the ticket or put it on the big screen.

The audience reads in their language
Each reader chooses a language and follows the talk sentence by sentence, as text or speech.

Read in the app
Scan the QR code with the camera, and the talk opens in your language. With Listen you hear the talk with the screen locked.

Names and terms
Add names and technical terms before the talk. Speech recognition gets them in advance, and the translation leaves them unchanged.

The same QR code every time
A room keeps its code. Print the QR code once and use it every week. Rooms that are not used for 90 days are deleted.

Put the code on the big screen
The QR code fills the screen, and the invitation changes between the languages being read.

Read aloud in the audience's language
Turn on Read aloud, and the audience can hear the translation on their phones, in the Atcha app even with the screen locked. Audio is made only for languages someone listens to.

Up to 10 languages at once
The audience chooses from 110 languages. Up to 10 of them can be read in the same talk at once.

A warning before time runs out
You are told 10 minutes before your time runs out, and can buy more without stopping the talk.

You pay for the time you speak
Every language is included, and the audience never pays. 60 minutes free for new speakers, to use within 30 days. Time you buy never expires. An hour read aloud uses 3 hours.

Privacy
We do not store audio. Audio is sent to an AI service that turns it into text. We do not keep it.
The text is deleted after 24 hours. What the audience has read is deleted 24 hours after the talk ends.
No account to read. The audience scans and reads, with no sign-in and no cookies.

How long is the delay?
The text arrives one sentence at a time, usually 2–4 seconds after the speaker has said it. On a slow network it can take longer.

Which languages can be used?
The audience can read 110 languages, and the speaker can talk in 112. The translation is made with AI and can contain mistakes, so names and technical terms can be added before the talk.