Privacy

Your audio is never uploaded. It is recorded, read and transcribed on your device.
An account is optional. If you sign in and press Save, the text of that transcript is stored in your account — never the audio.
Analytics is cookieless and counts named steps, never your files, words or titles.
One advertisement, at the foot of the home page, loads only when you scroll to it.

This page describes what the site actually does. Most of it you can check yourself in your browser’s developer tools, and the paragraphs below say where to look.

Your recording

A recording made here comes from your microphone, through your browser, and goes no further. A file you open is read where it already is. Either way it is decoded and resampled to the 16 kHz the speech model expects, then handed to a Web Worker on the same device. No request carrying any part of your audio is made. Open the network panel before you press Transcribe and watch: the recording does not appear there, at any size, at any point.

The microphone

The page asks your browser for the microphone at the moment you press Record, and at no other moment. Nothing reads your list of input devices on load, and nothing opens a stream early. While recording, the page shows the word REC, a lamp, a counter and a level bar. Stop or Discard releases the microphone.

What is written to this device, and what is not

Nothing is kept automatically. Not the audio, not the transcript, not your corrections or labels, not the matter or date you type onto a document. Reload the page and an unsaved transcript is gone, so download it or save it to your account first.

One thing of yours can be kept, and only by asking. Pressing Keep on this device writes that recording — the audio, its name, its date and its length — into this browser's own storage on this computer, in an IndexedDB database belonging to this site. You press it once per recording; there is no default. It survives closing the tab, it is never sent anywhere, and it can be played by anyone who can use this computer: it is not encrypted or behind a password.

The Kept on this device list has a Delete button for each recording and a Delete everything button that removes all of them, each with a second press. Clearing this site’s data in your browser removes them too.

Every file this page writes for download is opened again and measured in this tab before it is offered; that check makes no request. Playback makes no request either: the sound is played from the page’s memory through the browser’s audio engine, with no media element and no address for your audio.

What we keep on your device

One thing, and it is ours rather than yours: the speech model. About 71 MB of runtime and model files are served from this site and kept by your browser so the next run does not download them again; the larger model adds 80 MB, only if you choose it. In developer tools, under Application → Cache Storage, the entry named transformers-cache holds the model files. They are the same for every visitor and contain nothing of yours.

If you sign in: your account

An account is optional; everything above works without one. It exists to keep transcripts you choose to save and to share them by link.

Sign-in is with Google. We ask for the scopes openid email profile. Google sends your name, email address and picture; we keep your name and email and do not keep the picture. Our use of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements.
What a saved transcript holds. When you press Save to my account, we store that transcript’s title, its lines of text, their times, the speaker labels and marks you typed, the language, which model produced it, and the recording’s length. The audio is never sent, and neither is the file name or the matter and date you type onto a document.
Where and how. Saved transcripts are stored in Amazon Web Services’ Frankfurt region (eu-central-1), encrypted at rest. They are not end-to-end encrypted: the site’s operator can technically read them, and does so only to keep the service running or when the law requires it. They are never used to train any model.
What you are responsible for. A transcript can contain other people’s words and special-category information such as health details. Save only what you are entitled to keep. This service offers no business associate agreement: do not store protected health information here.
Share links. A link you create lets anyone who has it read that transcript — never the audio — until it expires after 7, 30 or 90 days, or until you turn it off. The link carries a random token after #, so it is not sent in the request line or as a referrer. The shared page loads no analytics and no advertising.
Activity. When you visit signed in, we note that your account was active that month, in a marker that expires after 14 months, and add one to a monthly count that names nobody. That is how we learn whether accounts come back.
Pro, which is not open. If you press I’d pay for this while signed in, we record which feature, which price was shown, the date and that nothing was charged. If you tick Email me when Pro opens, that consent is stored on the same record and used for that one email.

Cookies and storage when you sign in. Three cookies, all first-party: __Host-sf-quiettranscript is your session (HttpOnly, 12 hours); __Host-sf-quiettranscript-hint is a flag with nothing in it that tells the page you are signed in; __Host-sf-quiettranscript-state lives for ten minutes during a sign-in to prove it started here. During a sign-in the tab’s session storage briefly holds the sign-in state and the page to return to; each is removed once used. If you pressed Save first, it also holds the transcript waiting to be saved: that is removed once the save is confirmed or you press Start over, and if the sign-in does not finish it stays in that one tab, until the tab is closed, so the page can put your transcript back. Separately, while a transcription runs, the tab’s session storage holds a marker saying one is in progress, so that if the page reloads mid-run it can tell you the run was cut short; it is removed when the run ends.

Deleting. Delete any saved transcript or share link from your account page. Delete my account removes your profile, every saved transcript and every share link immediately. Point-in-time backups of the database keep its state for up to 35 days, after which deleted data is gone from them too.

There is no payment processor on this site: nothing can be bought here.

Analytics

This site counts page views and named steps with a self-hosted, cookieless Umami collector at analytics.pagefabrica.com. It stores no identifier in your browser, cannot recognise you on another visit and cannot follow you to another site. It honours Do Not Track: with that setting on, the collector is never contacted. It records a page address (never the part after #, and of the part after ? only the five campaign tags utm_source, utm_medium, utm_campaign, utm_content and utm_term; every other parameter is dropped before sending), a referrer shortened the same way, a screen size, a browser and a country.

Named steps carry only closed lists of values, so they cannot carry your file name, your words, a title, a speaker label, an exact length, an account or a share token. The two the fleet counts everywhere are work_started and work_completed.

Every step this site counts, and every value it may carry

Steps: record_pressed, record_started, record_refused, file_chosen, file_read, file_refused, transcribe_pressed, work_started, work_completed, work_failed, work_interrupted, engine_fallback, model_changed, language_changed, scope_changed, live_draft_toggled, playback_started, transcript_edited, speaker_labeled, mark_added, export_clicked, export_completed, export_refused, keep_pressed, save_file_pressed, save_pressed, transcript_saved, save_refused, saved_opened, saved_updated, share_created, share_copied, sign_in_clicked, sign_in_completed, pro_lock_seen, pro_pressed, pro_intent, faq_opened, hint_opened, cta_clicked and page_error.

Values: model (small or larger), engine (graphics, one CPU thread or several), format (the export format), band (a recording’s length in five coarse bands, never the length itself), source (recorded, live, file, video, kept or saved), reason (why something was refused or failed, from a fixed list), target (which button, question or option), from (which Pro feature), lang (the language, as one of eleven common languages or “other”), tool (which of this site’s pages the step happened on), cause (a failure or refusal in one coarse word: memory, browser, engine, stalled, stopped, format or size) and layout (whether a PDF had numbered lines).

Advertising

Your recording, your transcript and your corrections are not part of this. They are made and read on your own device and the audio is never uploaded; that is untouched.

What advertising changes is that your visit is now seen by Google. This site carries advertising from Google AdSense, under the publisher account pub-3015567014793107, in one box at the foot of the home page. The dedicated tool pages carry no ad box or ad loader. When the box is filled, your browser asks Google for an advertisement, and that request carries your IP address, your user agent and the address of the page you are on. Google may set or read advertising cookies in your browser, and for everything it receives that way Google is its own controller.

The home page’s security policy is wider to allow it: it may run Google’s advertising and consent code and load images and connections from any HTTPS address, because advertisers’ files come from changing domains. The tool pages use a stricter isolated policy that admits no advertising host.

Nothing about a recording can travel with any of it. Not a word of a transcript, not a file name, not a duration, not the language it was in: the advertising runs from a separate file that is never given them, and no code on this site passes anything from one to the other.

None of it happens unless you scroll to it. Google’s library is fetched only as the box nears your screen. A visit that never reaches the bottom makes no request to Google. Block advertising entirely and the tool works exactly the same, because the whole engine is already on your machine.

As this build stands the box carries no ad unit, so nothing on this page fetches Google’s advertising code and no request reaches Google. Advertising is approved and switched on separately from this release, and this paragraph goes when it is.

The consent choice

In the European Economic Area, the United Kingdom and Switzerland, Google shows a consent message from a Google-certified consent platform before it personalises advertising. We do not run a banner of our own. What that choice controls is whether Google may use cookies and identifiers to personalise and measure advertising; it does not decide whether an advertisement appears. Your answer is stored by Google, not by us. To change it, use the control on Google’s message, or clear this site’s cookies and answer again. Cookies sets this out in full.

What our server can see

Asking for a page and downloading the model are ordinary web requests: our server sees that a request arrived, from which network address and for which file. None of those requests contains your recording. If you sign in, the account requests described above go to the same site, under /api/.

Email and contact

If you write to support@quiettranscript.com we keep your message so we can answer it. Ask at the same address and we will delete the correspondence. The site is operated by AIErudit; the same address reaches us about anything on this page.