What stays on your device
- Your recording. Your browser reads the file from your disk. It is not uploaded.
- Your transcript. Saved in your browser’s local storage so your library works. It is not sent to us.
- Your corrections and preferences. Also in local storage.
What crosses the network
| Request | Sent to | What it contains |
|---|---|---|
| Speech engine code | cdn.jsdelivr.net | A standard download of the open-source Transformers.js library. No audio. |
| Speech model files | huggingface.co | A one-time download of the Whisper model, then cached by your browser. No audio. |
| Site analytics | Our own server | Anonymous usage events such as pages viewed, settings chosen, file type and length, and errors. Never audio, file names or transcript text. See the privacy policy. |
| Advertising | Google, when ads are enabled | Standard ad requests from content pages. See the cookie policy. |
Check it yourself
- Transcribe one short file so the model is saved.
- Turn off Wi-Fi or unplug your network cable.
- Transcribe another file. It works, because nothing needs the internet.
You can also open your browser’s developer tools, choose the Network tab and watch: no request carries your audio.
Share links
A share link carries the transcript compressed inside the part of the address after the # sign. Browsers do not send that part to servers, so we never receive it. Anyone with the link can read the transcript, so share it only with people you trust.
Deleting your data
Delete transcripts one at a time in your library, use Delete all, or clear this site’s data in your browser settings.
Reporting a security issue
Email hello@transcribela.com. We aim to reply within two business days.
Every network request the site makes
| Request | Goes to | Contains your audio or text? |
|---|---|---|
| Web pages, styles, scripts | transcribela.com | No |
| Speech model files (first use only) | Hugging Face and the jsDelivr code CDN | No |
| Anonymous usage counts (pages viewed, features used) | transcribela.com | No. No IP address is stored. |
| Ads and the consent message, when enabled | No | |
| Icons on the Summarize with AI buttons | Google’s public favicon service | No |
None of these requests carries your recording, your transcript or your file names. Transcription itself makes no network request at all once the model is saved.
How to check it yourself
- Open your browser’s developer tools (F12, or Option Command I on a Mac) and choose the Network tab.
- Transcribe a short file. You will see the model files the first time, and no request containing audio.
- Turn off Wi-Fi and transcribe another file. It still works, because nothing needs the network.
Deleting your data
Delete single transcripts in your library, or delete everything there with one button. To remove the saved speech models as well, clear this site’s data in your browser settings. Because nothing is on our side, there is nothing for us to delete and no request to send.