WayVoice

WayVoice is a local AI dictation app for Linux users who want to enter text by speaking in their desktop applications. It processes speech on your machine, with no cloud service required for recognition.

• Local speech recognition: Use Faster-Whisper or whisper.cpp with prepared local models to transcribe speech offline.
• Text delivery: Send transcriptions to the Wayland clipboard or use optional automatic paste in the active application; text remains in the clipboard if insertion fails.
• Language selection: Choose a fixed recognition language or let compatible models detect the language automatically; the interface also supports multiple languages offline.
• Dictation controls: Choose your microphone, apply personal text replacements, and use recording sounds and a compact status indicator to track recording.

Initial runtime and model preparation requires an internet connection. Once the models are ready, recognition works offline. The dictation shortcut doesn't trigger downloads for missing models.

GNOME on Wayland is the primary tested desktop. KDE Plasma has shortcut integration, but full dictation and paste compatibility still needs validation; wlroots also needs community testing. Automatic paste depends on permissions and the optional ydotool helper, while manual clipboard paste remains available.

WayVoice is open source under AGPL-3.0. Linux packages include Debian/Ubuntu and Fedora packages, plus an x86_64 AppImage that requires glibc 2.39 or newer.