-black){kind=icon}
🇺🇸 English | 🇫🇷 Français | 🇩🇪 Deutsch | 🇪🇸 Español
PiperRead is a lightweight automation solution designed to bring high-quality neural text-to-speech (TTS) to Linux desktops.
🌐 Official Website: piperread.davalan.fr
Unlike cloud-based solutions, PiperRead operates entirely offline (locally) thanks to the Piper engine. It acts as a bridge between your desktop environment (Clipboard/Mouse) and the synthesis engine.
It allows you to read aloud any text selected with the mouse or copied to the clipboard, without requiring a complex screen reader.
- Accessibility: Quick reading of content for people with mild visual impairment or eye fatigue.
- Productivity: Listening to articles or documents while performing another task.
- Proofreading: Hearing your own text read by a third-party voice to detect errors.
- Total Privacy: 100% local processing. No data is sent to any cloud.
- Zero Latency: Instant playback suitable for real-time usage.
- Universal Compatibility: Automatically detects and adapts to Wayland (Debian 12/13) or X11.
- Smart Selection: Prioritizes mouse selection (primary) and switches to clipboard if no selection is active.
- Isolation: Runs in its own Python virtual environment to avoid polluting your system.
Before installing, ensure your system has the required audio and clipboard tools.
# System update
sudo apt update
# Install Python, Audio, and Clipboard tools
# (Installs both wl-clipboard for Wayland and xsel for X11 to ensure compatibility)
sudo apt install -y python3 python3-venv python3-pip alsa-utils wl-clipboard xselSince this project relies on large voice models and a specific virtual environment, you must initialize the project after cloning it.
mkdir -p $HOME/git/piper
cd $HOME/git/piper
git clone https://github.com/RonanDavalan/PiperRead.git
cd PiperReadThis step creates the Python isolation and downloads the neural voice model (French - Siwis Medium).
# Create virtual environment
python3 -m venv piper-env
# Install Piper TTS Engine
./piper-env/bin/pip install piper-tts
# Download voice model
mkdir -p voices
cd voices
wget https://huggingface.co/rhasspy/piper-voices/resolve/main/en/en_US/lessac/medium/en_US-lessac-medium.onnx
wget https://huggingface.co/rhasspy/piper-voices/resolve/main/en/en_US/lessac/medium/en_US-lessac-medium.onnx.json
cd ..chmod 700 read.shTo launch PiperRead like a native application:
# Create local applications folder
mkdir -p $HOME/.local/share/applications
# Copy desktop file
cp Ressources/PiperRead.desktop $HOME/.local/share/applications/
# Update menu database
update-desktop-database $HOME/.local/share/applications- Highlight text in any application (Browser, PDF, Editor).
- Click the PiperRead icon in your menu (or use your custom keyboard shortcut).
- The text is read immediately.
- Copy text (Ctrl+C).
- Launch PiperRead.
Current Alpha Limitation: To stop reading while it is playing, run pkill -f aplay in a terminal.
This project was born from a discovery: the impressive quality of the Piper engine for a fully free and local solution.
- Natural Vocal Rendering: The choice of this neural technology allows for fluid and poised reading, making listening comfortable over time.
- Lightweight Architecture: PiperRead is not a heavy application, but a minimalist orchestrator. It connects your desktop and the audio engine with an almost non-existent system footprint.
- Clean Installation: The strict use of virtual environments (venv) ensures that the software remains confined and does not modify your main system libraries.
The impetus for this project comes from my brother, a historic Debian user, who identified Piper as a useful solution for local TTS.
The PiperRead project is the result of a hybrid Human-AI collaboration:
- Ronan Davalan: Project Lead, Chief Architect, Quality Assurance (QA).
- Google Gemini: AI Architect, Planner, Technical Writer.
- Claude (Anthropic) & Perplexity: AI Technical Consultants (Code review and optimization).
- Core Engine: Piper TTS by Rhasspy.