The blazing fast, offline, and privacy-respecting Text-to-Speech tool for Linux read-aloud experience.
curl -sSL https://raw.githubusercontent.com/elkwaet/piper-tui/main/net-install.sh | bash
⚡ Then restart your terminal (or run source ~/.bashrc) and launch with piper-tui.
Listen to heavy outputs from ClaudeCode, Antigravity, QwenCode or Ollama while multitasking in your terminal. Zero LLM token costs, zero API latency.
Highlight and listen to lengthy API documentations or code tutorials while keeping your eyes focused on your actual code in the IDE.
Listen to long discussions on Slack, Discord, Telegram or heavy email threads in the background while working on something else.
Absorb long PDF reports and Word documents without getting stuck scrolling. 100% offline text-to-speech processing.
Not at all. Unlike many heavy local AI tools, Piper is highly optimized (using ONNX runtime) for CPU inference. It requires zero GPU, less than 100MB of RAM, and runs smoothly on almost any 64-bit processor.
Yes. After the initial installation and downloading your preferred voices (which are fetched directly from HuggingFace), all text-to-speech processing happens locally on your machine. Zero telemetry, zero cloud.
Currently, the built-in UI catalog supports over 45 high-quality voices across English (US/GB) and French. Piper engine itself supports dozens of other languages which can be manually configured.
Once you trigger playback via your keyboard shortcut, a minimalist floating widget will appear. You can pause, resume, or entirely kill the playback directly from it.
It works seamlessly on GNOME, KDE Plasma, XFCE, Cinnamon, MATE, LXQt, and Pantheon. Setup instructions for window managers (like i3 or sway) are available in our wiki.