Privacy Policy

The short version: UmBreak cannot see, hear, store, or share your audio, because your audio never leaves your device. Transcription and editing run entirely in your browser.

What we do not collect

Your audio or video files. They are decoded, transcribed, and edited locally. Nothing is uploaded to any server.
Accounts or identities. There is no signup, so there is nothing to tie activity to.
File contents or transcripts. These exist only in your tab's memory and disappear when you close it.

What is downloaded to your device

On first use, your browser downloads an open-source speech-recognition model (Whisper, ~40 MB) from a public CDN (jsDelivr / Hugging Face). This is a one-time download, cached by your browser afterwards. The model runs locally; downloading it does not send your audio anywhere.

What we may collect

Anonymous, aggregate usage statistics (e.g. page views) via standard, privacy-respecting analytics, if enabled. These contain no audio, no transcripts, and no personal identifiers.
Voluntary contact. If you email us, we will have your email address and message, used only to reply.

Cookies and local storage

UmBreak does not set tracking cookies. Your browser may cache the AI model files so the tool works faster on repeat visits — that cache lives on your device and you can clear it any time.

Children

UmBreak is a general-purpose utility not directed at children under 13, and collects no personal information from anyone.

Changes

If this policy ever changes in a way that affects the core promise above — your files stay on your device — the change will be noted here with a new date.

Contact