Desktop Reliability
Reliable Voice Dictation on Real Linux Desktops
Recent Vocalinux beta releases focused heavily on the hard parts of Linux desktop dictation: Wayland text injection, IBus lifecycle behavior, suspend/resume recovery, keyboard layouts, non-ASCII text, and settings compatibility.
Suspend and resume recovery
Vocalinux restarts recognition and shortcut backends after wake so laptop sleep cycles do not require a manual app restart.
Safe engine switching
Recognition stops before engine changes, preventing crashes while moving between whisper.cpp, Whisper, VOSK, or Remote API.
Keyboard layout preservation
IBus activation preserves the current XKB layout so dictation does not unexpectedly switch to US keyboard behavior.
IBus runtime recovery
Readiness probes, scoped activation, and runtime recovery improve text injection without restarting the whole app.
Non-ASCII fallback
ydotool can fall back to clipboard paste for accented and non-ASCII characters on paths that cannot type them directly.
Tray and settings polish
Bundled resources, lower dialog height, and an explicit close button improve behavior across desktop environments.
Reliability Work by Release
v0.12.0-beta
Thread safety hardening for Remote API, IBus, and text injection plus better settings behavior.
v0.11.0-beta
IBus readiness probes, runtime recovery, final speech preservation, installer hardening, and distro compatibility fixes.
v0.10.2-beta
Non-ASCII ydotool fallback, IBus detection on Wayland, startup fixes, and Pop!_OS/Ubuntu dependency coverage.
v0.10.1-beta
Suspend/resume recovery, safe engine switching, keyboard layout preservation, tray resources, and push-to-talk reliability.
Why This Matters
- Dictation should survive laptop lid closes, docking changes, and USB re-enumeration.
- Text should appear in the focused app without breaking existing IBus engines or keyboard layouts.
- Accented characters should be handled even when a low-level injection backend cannot type them directly.
- Settings should fit smaller displays and window managers with different decoration behavior.
Where to Debug Issues
If dictation works in one app but not another, start with the display server and input method path. Wayland sessions often depend on IBus setup, while X11 paths typically use xdotool or compatible fallbacks.
Open troubleshooting guide