Privacy policy

Last updated: 2026-07-29

The short version

Your audio and video files are never uploaded. They are opened by your browser and processed by a speech model running on your own device. This site has no server and no upload endpoint, so there is no mechanism by which we could receive a file even if we wanted to. Your transcripts are equally never transmitted.

You do not have to take that on trust — here is how to verify it yourself, including the simplest test: disconnect from the internet once the model has cached, and watch the tool keep working.

What we do not collect

• Your audio or video files, or any part of them.
• Your transcripts, edits, speaker labels or exports.
• File names, durations, formats or any metadata about what you transcribed.
• Microphone audio — recording happens in the page and is never streamed.
• Accounts, email addresses, names or payment details. There is nothing to sign up for.

What is stored on your device

Three things, all local to your browser and all removable by clearing this site’s data:

• Speech model weights in Cache Storage — between 45 MB and 81 MB per model tier you have used. This is what makes the second visit instant and offline use possible. It contains no data of yours. You can delete it from the Options panel in the tool.
• Site files cached by a service worker so the site loads offline.
• Your light/dark theme preference — a single entry in localStorage. It is a display preference, not tracking, and it is not transmitted.

Your recording and your transcript live only in the tab’s memory while you work and are discarded when you close it. They are never written to the cache.

Analytics

If analytics is enabled, it is Cloudflare Web Analytics: a cookieless page-view counter that records which pages were viewed, from which broad region, and which referrer sent you. It sets no cookies, builds no profile and follows you nowhere. It cannot see your files, your transcripts, or anything you type — it only knows a page was loaded.

Because it is cookieless and stores nothing on your device, there is no consent banner. That is the point of choosing it.

Hosting and logs

The site is served as static files by Cloudflare Pages. Like any web host, Cloudflare processes standard request information — IP address, user agent, requested URL — in order to deliver the page and protect against abuse, and that is governed by Cloudflare’s privacy policy. This is the same information any website you visit necessarily receives, and it says nothing about the content of what you transcribe, because that content is never sent.

Third parties

There are none in the processing path. The speech models and the inference runtime are served from this domain rather than from a model hub or a CDN — deliberately, because loading a model from a third party would tell that third party your IP address and what you were doing every time you used the tool. The automated tests fail the build if any cross-origin request appears.

Advertising

There is no advertising today. If display advertising is added later, this page will be updated before it goes live, and if a network requires consent for personalised ads a proper consent banner will appear at that point. No advertising arrangement will ever involve access to your files or transcripts — the architecture makes that impossible, not merely prohibited.

Children

This is a general-purpose utility, not aimed at children, and it collects no personal information from anyone.

Your rights

Rights such as access, correction and erasure apply to personal data a service holds. We hold none: there is no account, no stored file and no transcript on any server. Anything stored in your own browser can be deleted by you at any time by clearing this site’s data, or by using the “Delete cached models” button in the tool’s Options panel.

Changes

Material changes will be reflected here with an updated date. The commitment that recordings are processed only on your device is architectural — changing it would require rebuilding the site as something fundamentally different, not editing this page.

Contact

Questions: hello@transcribelocally.com