Data & privacy
Parolo is a free, real-time speech-translation service for community events — schools, universities, faith communities, libraries, bookstores and lectures. We built Parolo so that a room full of people who speak different languages can simply understand one another. This notice explains, in plain language, what happens to your voice and your words, how we keep them safe, and how to reach us.
The short version
- Parolo does not record or store your audio. Translation happens live, and only the written transcript of an event is kept.
- Only the event organizer can see the saved transcript of their own event.
- We do not use advertising, tracking, or analytics, and we do not sell or share your data.
- If an organizer protects their room with a password, that password is stored encrypted.
- Organizers stay in control — they can review, export, or delete their transcripts whenever they choose.
How Parolo handles your speech
When a speaker talks, their microphone sends audio to OpenAI, which translates it in real time and returns translated audio and captions. Those captions and the translated voice are delivered to the room's listeners using LiveKit.
Translation happens in the moment — we never record or store your audio. There is no audio file of your event sitting on our servers, and there never will be. The only thing Parolo keeps is the written transcript (the original and translated text), saved in our Neon database, so the organizer can review or share it afterwards.
What we collect and store
- Event transcripts — the written original and translated text of a session. This is the heart of Parolo, and it belongs to the organizer.
- Organizer accounts — event organizers sign in through Neon Auth. Listeners do not create accounts and do not log in.
- Room details — the room name, languages, audio settings, and (if chosen) a listener password.
- Team membership — if an organizer invites co-hosts, we keep their name and email so they can help run the event.
- Custom vocabulary — any special terms or pronunciation hints an organizer adds to improve translation.
- Device history — which microphone and speaker an organizer has used, so the next event is set up faster.
We also keep lightweight, anonymous operational records — such as how many listeners joined and how long a session ran — purely to keep the service healthy. These records do not contain the content of your speech.
What we don't do
- We do not record or store audio — not even for a moment.
- We do not use advertising, tracking, or analytics. Parolo contains no ad trackers.
- We do not sell, rent, or trade your content or personal data.
- We do not collect payment information — Parolo is free, with no subscriptions.
- We do not require listeners to identify themselves — they join without an account.
Who can see your content
Within Parolo, only the event organizer can access the saved transcript of their own event — to review it, export it, or delete it. Invited co-hosts can help run the room, but they see event details, not the full transcript text.
Listeners receive the live translated audio and captions during the event, and nothing more. They cannot see the organizer's saved transcript history, and they leave no account behind.
Our own administrators see only operational metadata — such as session status, duration, and listener counts — and never read the content of your transcript. Your words stay between you and your event.
The services that help us
Parolo relies on a small set of trusted providers, each doing one specific job:
- OpenAI — processes the speaker's speech in real time to produce translated audio and captions, and helps generate pronunciation hints for custom vocabulary.
- LiveKit — delivers translated audio and captions to listeners during the event.
- Neon — stores transcripts and application data in a managed Postgres database, and powers organizer sign-in through Neon Auth.
- Render — hosts and runs the Parolo application.
- jsDelivr and Google Fonts — deliver the small code libraries and fonts that make the pages load quickly and look right.
These providers process only what is needed for their role, and they are not permitted to use your data for advertising.
How we protect your data
- Your audio is never stored. Because translation is live, there is no recording that could ever leak.
- Listener passwords are encrypted at rest using AES-256-GCM, a modern encryption standard, with a unique key. We never store them as readable text.
- Co-host invitations are secure. Invite links are single-use and expire after seven days, and only a fingerprint (never the link itself) is stored — so a shared link can't be reused or guessed.
- Access is scoped. Listeners receive audio but can never broadcast or alter a session. Transcript access stays with the organizer.
- Private by default in the important ways. API responses are never cached, and browser requests are checked to prevent cross-site abuse.
We take these measures seriously and continue to harden Parolo. No system is perfect, but protecting your events is a commitment, not an afterthought.
Data retention & deletion
Transcripts are kept until the organizer chooses to delete them. We do not delete them automatically, and we do not set them to expire — your event's text stays available exactly as long as you want it.
The organizer can export a transcript as plain text, a Word document, or a PDF to keep a personal copy, and can delete any transcript they no longer need. Deleting a room also removes its transcripts, settings, vocabulary, and member list.
If you attended an event and have a question about its transcript, the event organizer is the person to ask — they control it. You are always welcome to reach us using the contact below.
Privacy contact
If you have a question about your data or how Parolo handles it, please get in touch — we're glad to help.
Nikolaï Clavier
clavierclassics@gmail.com