EDIT: https://github.com/goodroot/hyprwhspr seems more powerful
A voice recording and transcription tool for Hyprland, using Whisper for speech-to-text and copying results to clipboard. It's using Faster Whisper (optimized for CPU) and runs fully locally.
- Ensure you have
uvinstalled (Python package manager). - Run
make installto build and install the daemon and CLI tools.- This creates a virtual environment, installs dependencies, builds executables with PyInstaller, and copies them to
/usr/local/bin.
- This creates a virtual environment, installs dependencies, builds executables with PyInstaller, and copies them to
- To uninstall, run
make uninstall.
- Start the daemon:
wl-voiced(run in background or as a service). - Use the CLI to control recording:
- Start recording:
wl-voice start - Stop recording and transcribe:
wl-voice stop
- Start recording:
The transcribed text will be copied to your clipboard and a notification will appear.
Add the following bindings to your Hyprland config (~/.config/hypr/hyprland.conf):
# Start recording
bind = SUPER, R, exec, wl-voice start
# Stop recording and transcribe
bind = SUPER, S, exec, wl-voice stop
Ensure the daemon is running before using the bindings. You can add wl-voiced to your startup applications.
To automatically start the wl-voiced daemon when Hyprland launches, add the following line to your Hyprland config (~/.config/hypr/hyprland.conf):
exec-once = wl-voiced
This ensures the daemon is running in the background whenever you start Hyprland.