A new, fully local voice control application called EasySpeak has arrived on the Linux desktop, with accessibility and hands-free control for users running modern GNOME environments. Developed by Matt Hartley, EasySpeak is free, open source, and designed to turn voice commands into action.
EasySpeak should fill a notable gap in the Linux ecosystem where many existing voice tools are either proprietary, reliant on outdated X11 technology, or dependent on cloud services. This project is built from the ground up to be Wayland-native, meaning it works correctly on contemporary Linux desktops where older X11 tools often fail.
EasySpeak is designed for anyone with repetitive strain injury (RSI), specific accessibility needs, or those who simply need to operate their computer when their hands are busy. You can activate the entire system hands-free by simply saying the wake word, “Hey Jarvis.”
A major feature that sets EasySpeak apart is its commitment to privacy. The entire application runs fully locally. It uses OpenWakeWord for activation, the faster-whisper tool for transcribing commands, and the Piper engine for text-to-speech feedback. This means absolutely no voice data leaves your machine, which is a fantastic move for users who are rightly skeptical about cloud-based AI services listening in.
Once activated, EasySpeak lets users control core aspects of their desktop. One of the most detailed and impressive features is the Mouse Grid. Saying “grid” overlays a 3×3 numbered layout on the screen, similar to a phone keypad. You can zoom in on specific zones by saying the corresponding number, chaining multiple numbers together to move quickly, like “3-6-3.”
After positioning over the target, you say “click” to select it. The grid system also handles drag-and-drop actions using the commands “mark” and “drag.”
EasySpeak also includes controls for system functions and applications. You can open and close applications by name, and there are commands for managing files like “open documents” or “open downloads.” It also has MPRIS integration, letting you control playback with commands like “play,” “pause,” and “next.”
The developer is also adding more GNOME controls, including maximizing and minimizing windows. These updates will include core system commands like adjusting volume, brightness, and toggling the “do not disturb” feature.
The application currently has tight integration with the Qutebrowser web browser. This integration lets you navigate the internet using link hints, switch between tabs, browse history, and search the web using your voice. While the dictation feature is currently a work in progress, it already supports voice-to-text input in any field, complete with commands for punctuation like “comma” and “new line.”
For those who like to tinker, the system is highly extensible thanks to a simple Python plugin architecture. You can drop a Python file into the plugins directory to add custom commands and integrations for specific applications. Hartley has been working on this project for over a year and said the extensibility is a core part of the design.
EasySpeak is still in early development, so you should expect some minor bugs and changes as the project matures. It requires a Linux distribution running GNOME Shell 47 or newer on Wayland, and it currently requires Python 3.13 for installation due to dependency issues with Python 3.14.
Installation involves cloning the GitHub repository, installing several system packages, setting up a Python virtual environment, and enabling a GNOME Shell extension. EasySpeak is available right now for free under the GPL-3.0 license on GitHub.
Source: Matt Hartley/fosstodon
