Get started on Windows
Scribe0 for Windows is a native desktop app. Transcription runs on your PC — consult audio is never uploaded.
Requirements
Section titled “Requirements”- Windows 10 (version 2004) or Windows 11, 64-bit
- ~1 GB free disk space (app + speech model)
- An internet connection for the one-time model download and for the optional note-enhancement step
Install
Section titled “Install”- Download Scribe0-Setup.exe.
- Run it. Windows SmartScreen may show “Windows protected your PC” — this is expected for a new, unsigned app during the beta. Click More info → Run anyway. (You can verify the download against its SHA-256 checksum.)
- The app installs per-user (no administrator rights needed) and opens automatically, with shortcuts on the Desktop and Start menu.
On first launch, Scribe0 downloads the Parakeet speech model (~631 MB) into
%LOCALAPPDATA%\Scribe0\models. This happens once; after that, transcription
works fully offline.
Your first note
Section titled “Your first note”- On the Workbench tab, press Record (Ctrl+R) and speak. With live transcription on (the default), text appears as you talk.
- Press Record again to stop.
- Pick a prompt from the dropdown and press Enhance (Ctrl+E) to draft a structured note. The note streams into the pane as it generates and renders with headings and lists.
- Copy (Ctrl+Shift+C) the result into your EMR or letter.
Every recording and enhanced version is saved locally — find them on the History tab.
Updates
Section titled “Updates”The app updates itself: new versions download in the background as a small delta and apply the next time you close and reopen Scribe0. Check the current status any time in Settings → About → Updates.
Where your data lives
Section titled “Where your data lives”Everything stays on your PC, per-user:
| What | Where |
|---|---|
| Transcripts & notes | %APPDATA%\Scribe0\scribe0.db (SQLite) |
| Speech models | %LOCALAPPDATA%\Scribe0\models |
| Recordings (WAV) | Documents\Scribe0 |
| Logs | %LOCALAPPDATA%\Scribe0\logs |
The one exception is the optional Enhance step, which sends the transcript text to a cloud model — by default via an Australian-sovereign Amazon Bedrock endpoint. See Privacy & data handling.