After you start, open Apps β Recovery Translator in your Zoom meeting and tap
Start captions. Captions begin as soon as Zoom's audio connects.
Everyone gets live English and Spanish captions, whatever language the speaker uses.
Listeners can also hear the languages marked "audio" read aloud.
Session live
Share this code
Listeners scan the QR code or go to
Microphone: startingβ¦
English
EspaΓ±ol
Join a session
Connected
Privacy Policy
Last updated: October 3, 2026
Recovery Translator was built by and for the recovery community, where anonymity matters. This page explains, in plain language, what happens to what is said in a meeting while Recovery Translator is running.
The short version
No accounts, no names, no cookies, no ads, no analytics.
We do not record or save meeting audio, captions, or translations.
Meeting audio is sent to Google's Gemini API to be translated. We use Google's paid service, so Google does not use it to train or improve its products.
Google keeps what is sent and returned for up to 55 days, only to detect abuse of its service, such as hate speech or harassment. If its automated systems flag something, authorized Google staff may review it.
Anything said aloud while captions are running is sent for translation, so share with that in mind.
What we never collect
Recovery Translator does not ask for your name, email, phone number, or any other personal details. Listeners join with only a session code. Session leaders use an access code, which is not tied to anyone's identity.
What happens to meeting audio
While a session is running, audio from the leader's microphone (or from the Zoom meeting, through the Recovery Translator Zoom app) passes through our server straight to Google's Gemini API. Google returns the English and Spanish captions and spoken translation, and our server passes them straight on to the people in the session. Nothing is written to disk along the way, and nothing remains on our server after the session ends.
How Google handles it
Recovery Translator uses the paid tier of the Gemini API. Under Google's terms for paid services:
Google does not use the audio, captions, or translations to train or improve its products.
Google processes the data as a data processor, under its Data Processing Addendum.
Google logs what is sent and what comes back for up to 55 days. It uses these logs only to detect and prevent violations of its Prohibited Use Policy, and for any legally required disclosures. It does not use them to train AI models, other than the systems that enforce that policy.
Google's automated systems scan for the kinds of misuse its Prohibited Use Policy lists. For example:
Hate speech or harassment: using the service to attack, threaten, or bully people.
Sexually explicit or dangerous content: using it to produce explicit material, or instructions that could cause serious harm.
These rules target misuse of the service. They are not aimed at people sharing honestly about their own experience.
Flagged content may be reviewed by authorized Google employees.
The data may be stored briefly or cached in any country where Google or its agents have facilities.
Google does not receive anyone's name or account from us; it receives the audio itself. But a voice, or details someone shares, could still identify them.
Protecting anonymity in your meeting
Leaders: please let the group know at the start that the meeting is being translated live by an AI service, and briefly explain what that means.
Members: as you would in any open meeting, consider using first names only and leaving out details that could identify you or others, such as last names, workplaces, or places.
Everything said aloud while captions are running is sent for translation. In Zoom, the leader can tap Pause to stop sending audio for part of a meeting, or end the session.
What our server keeps
During a session: the session code, and which screens are connected to it. This is held in memory only and deleted when the session ends.
Technical logs: our hosting provider keeps routine logs, such as session codes, connection and error events, and Zoom stream IDs. These logs never include audio, captions, translations, or names.
Zoom app: when the Zoom app is installed, Zoom sends us an authorization token. We use it only to finish the installation, and we don't keep it.
What stays on your device
Some pages remember small settings in your browser, such as caption text size and display colors. The Zoom app also remembers the leader's access code. These stay on your device. The access code is sent to us only when you start a session, to check it, and is not stored. We do not use cookies.
Other services
Like any website, our hosting provider can see the IP address of each visitor. The pages load a QR-code library from Cloudflare's cdnjs, and the Zoom app loads Zoom's own app software. When used inside Zoom, Zoom's privacy policy also applies.
Changes
If how Recovery Translator handles data changes, we will update this page and the date at the top.
Tap I am a session leader and enter your access code.
Choose who can listen to audio: Spanish, English, or both. Everyone always gets English and Spanish captions.
Choose where the audio comes from β this computer (a microphone, or system audio through BlackHole) or a Zoom meeting β and tap Start session.
For a Zoom meeting, open Apps β Recovery Translator in the meeting and tap Start captions. You can also start the session from the Zoom app directly and post the join link to the meeting chat.
Share the QR code or session code. Optionally open the big display screen for a room projector, or the captions-only pop-out window for a simpler view.
Speak normally, in English or Spanish β both caption boxes update live.
Tap End session when the meeting is over. Sessions from a Zoom meeting also end on their own when the meeting does.
Listeners
Scan the QR code or tap I am a listener, enter the session code, choose your language, and tap Join.
Captions appear automatically. If the session offers audio in your language, tap π Listen in English or π Escuchar en espaΓ±ol to hear it.
If you switch apps or lock your phone briefly, reopening the page should reconnect you automatically.
About Recovery Translator
Recovery Translator provides real-time spoken language interpretation and live captioning for recovery meetings, area committees, and service assemblies.
How It Works
The leader opens a session and shares a QR code. Whatever language someone speaks, the room gets live English and Spanish captions side by side. Depending on the session, listeners can also hear a live spoken translation in Spanish, English, or both.
Getting Access
Leaders need an access code. Listeners only need the session code. To request an access code, contact dynontricia@gmail.com.
Built for Recovery
Built by and for the recovery community, with anonymity and accessibility in mind. Free to use and open source.