dylanrothhuck/clip2speech

clip2speech is a small collection of shell scripts that reads whatever text you have copied to your clipboard out loud. Press a keyboard shortcut, hear the copied text spoken, and press the same shortcut again to stop it immediately. It runs entirely offline using a local text to speech engine called Piper, so no internet connection or cloud service is needed and nothing you copy leaves your computer. It is built for Linux systems running Arch Linux with the Hyprland window manager, though the author notes it should work on any Linux setup that has the wl-clipboard, piper-tts, and aplay tools installed. Setting it up means installing those three dependencies, downloading one or more voice model files for Piper from an online voice repository, placing the scripts in a folder on your command line path, and marking them as runnable. Once installed, you can run the tool to read your regular clipboard, to read text you have highlighted but not copied, or to speak a specific sentence typed directly into the command. A companion script speaks in English while the main script defaults to a Spanish voice, and both can be switched to any of the many voices Piper supports by changing a setting in an optional configuration file. The README also shows how to wire the tool into a keyboard shortcut, how to trigger a small desktop notification when it starts or stops speaking, and how to add a clickable icon for it to a status bar tool called Waybar. Running the command a second time while it is already speaking stops it right away, which the author designed specifically so it is safe to bind to a single key without causing overlapping audio. This project is aimed at Linux desktop users, particularly those using Hyprland, who are comfortable editing configuration files and running commands in a terminal. It is released under the MIT license.