Skip to main content

Troubleshooting SigmaShake Voice

Hotkey does nothing​

Check the app is running — look for the tray icon. If it's not there, launch the app.

Check the hotkey is configured — right-click tray → Settings. If the Hotkey field shows "Click to record", no hotkey is set.

Linux Wayland — no input group — the app sends a notification at startup if it can't read /dev/input. Fix:

sudo usermod -aG input $USER
# Log out and back in, then restart the app.

Linux Wayland — modifier combo — modifier combos (Ctrl+F9) don't work in the hotkey field on Wayland. Use a bare key (F9) or bind sigmashake-voice --toggle in your OS shortcut settings.

Hotkey works from the tray menu but not from the keyboard​

The tray → "Record" menu item triggers recording; the keyboard shortcut is separate. Common causes:

  • Wrong hotkey format — on Wayland, modifier combos silently fail. Switch to a bare key.
  • Key is claimed by another app — try a different key.
  • AppImage path changed — if you moved the AppImage, update any OS-level shortcut binding that calls sigmashake-voice --toggle.

Silent recording / empty transcript​

  1. Open Settings → Input Device. Select your microphone explicitly (don't rely on System default if you have multiple devices).
  2. Speak more clearly and closer to the mic.
  3. Check that your mic isn't muted at the OS level.
  4. Check the waveform bars in the HUD indicator — if they don't move when you speak, the wrong input device may be selected.

Text doesn't paste after transcription​

The transcript is placed in your clipboard regardless. Check:

macOS — missing Accessibility permission:

  1. System Settings → Privacy & Security → Accessibility
  2. Add SigmaShake Voice if it's not listed, or enable its toggle.

Linux X11 — missing library:

sudo apt install libxtst-dev    # Debian/Ubuntu
sudo dnf install libXtst-devel  # Fedora

Linux Wayland paste — requires zwp_virtual_keyboard_manager_v1 (supported by COSMIC, Sway, and most compositors). If paste fails but clipboard works, your compositor may not support virtual keyboard input. Use auto-paste disabled and press Ctrl+V manually.

In all cases the text is in your clipboard — paste manually with Ctrl+V if needed.

"Whisper model loading" on first use​

The model initializes in the background and takes 1–3 seconds. Wait a moment and try the hotkey again. The notification disappears once the model is ready.

If this persists beyond 10 seconds, check the log file (~/.config/sigmashake-voice/voice.YYYY-MM-DD.log on Linux) for errors.

"Microphone unavailable" notification at startup​

The app couldn't open your audio device. Possible causes:

  • Another app has exclusive access to the mic (common with hw:CARD=... direct ALSA devices — switch to PipeWire or System default in Settings).
  • No microphone is connected.
  • On Linux, the audio system (ALSA/PipeWire) is not running.

Open Settings → Input Device and try a different selection.

"Transcription unavailable" or "Model not found" notification​

The bundled Whisper model failed to load. Try:

  1. Reinstall the app — the model file may have been corrupted.
  2. Check disk space — the model needs ~600 MB.
  3. Check the log file for a more specific error.

macOS: "App can't be opened"​

Right-click the app in Finder → Open → Open. This one-time confirmation bypasses Gatekeeper for unsigned apps.

Windows: SmartScreen warning​

Click More info → Run anyway. The installer is not yet code-signed.

Checking logs​

Logs are at:

  • Linux: ~/.config/sigmashake-voice/voice.YYYY-MM-DD.log
  • macOS: ~/Library/Application Support/sigmashake-voice/voice.YYYY-MM-DD.log
  • Windows: %APPDATA%\sigmashake-voice\voice.YYYY-MM-DD.log

The Settings page footer shows the exact path for your system.

# Linux/macOS: tail the current day's log
tail -50 ~/.config/sigmashake-voice/voice.$(date +%F).log

Still stuck?​

Open an issue at github.com/sigmashake/sigmashake-voice/issues and include:

  • Your OS and version
  • The relevant lines from your log file
  • What you tried