About, and what happens to your audio.
FreeTranscribe runs a speech-recognition model inside your browser using WebGPU. The file you choose is read from disk into memory on your own machine and processed there. There is no upload step, because there is no server to upload to.
| What | Where it goes |
|---|---|
| Your audio or video file | Nowhere. It stays in the browser tab and is released when you close it. |
| The transcript | Nowhere, until you copy it or download it yourself. |
| The model weights | Downloaded from our servers to your browser cache once, then reused. |
| Page analytics and ads | Standard third-party cookies from Google, subject to your consent. See the privacy policy. |
How the site is paid for
Advertising, and nothing else. There is no paid tier, no upsell and no account, because there is no server cost to recover. Transcription happens on your computer, so a ten-hour file costs us the same as a ten-second one.
The model
Speech recognition uses OpenAI's open-source Whisper models, released under the MIT licence, converted to run in the browser with transformers.js and ONNX Runtime Web. The default model is the small, fast one. A larger, more accurate model is offered when your machine has the graphics memory for it.
What it needs
A recent Chrome or Edge on a desktop or laptop with WebGPU. Other browsers and phones are being worked on. If your browser cannot run it, the page tells you before anything is downloaded.
Who runs it
FreeTranscribe is an independent project run from Europe. It is not affiliated with OpenAI.