What it is
Voxtype turns speech into text on Linux. Hold a hotkey while you talk, let go, and it types what you said wherever your cursor is: an editor, a browser, a chat window. Transcription runs on your own computer with speech recognition models such as Whisper and Parakeet, with no cloud service or subscription.
It handles punctuation, can pause your music while you dictate, and has a meeting mode that transcribes continuously and exports the result.
Why it’s here
- Local. Your audio stays on your machine.
- Works in any app. It types into whatever window has focus.
- Fast on a plain CPU. No GPU required, though it can use one.
On CoreDesktop
Voxtype is installed from the AUR, and CoreDesktop leaves its setup to you. voxtype setup checks its dependencies, including the tool it types with, voxtype setup model picks and downloads a speech model, and voxtype setup systemd runs it in the background from login. GNOME shortcuts fire on a key press but not on its release, so use toggle mode: add a shortcut in Settings › Keyboard › Keyboard Shortcuts that runs:
voxtype record toggle
Press it once to start dictating and again to stop. update keeps Voxtype current with the rest of the AUR packages.