Skip to content

matx104/shout

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

shout

Simple speech-to-text for Wayland/Hyprland. Press a key to record, press again to transcribe and press Enter.

Works great with AI coding agents like Claude Code - they're forgiving of typos and excellent at semantic reasoning, so transcription doesn't need to be perfect.

demo.mp4

Installation (Arch Linux / AUR)

Prerequisite: A Vulkan driver for your GPU (install one):

# AMD (often pre-installed)
sudo pacman -S vulkan-radeon

# Intel
sudo pacman -S vulkan-intel

# NVIDIA (usually comes with nvidia drivers)
sudo pacman -S nvidia-utils

# No GPU / fallback (slower)
sudo pacman -S vulkan-swrast
yay -S shout

Uses the tiny whisper model (75 MB) - fastest transcription, good enough for AI agents.

Setup

Add this keybind to ~/.config/hypr/hyprland.conf:

bindd = SUPER, R, shout, exec, shout

Then reload: hyprctl reload

Usage

  1. Press Super+R to start recording
  2. Speak
  3. Press Super+R again to stop, transcribe, and type the text

Troubleshooting

File conflicts with libggml-git

If you see errors like libggml-git: /usr/include/ggml.h exists in filesystem (owned by whisper.cpp):

# Remove old whisper.cpp that bundles ggml
yay -Rns whisper.cpp

# Fresh install
yay -S shout

This happens when upgrading from an older whisper.cpp that bundled ggml internally.

License

MIT

About

Fast Speech to Text for Omarchy (or Arch/Hyprland)

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Shell 100.0%