Piper TTS

A fast, local neural text-to-speech engine optimized for Raspberry Pi 4 and low-end hardware.

Piper delivers high-quality voice synthesis by leveraging OnnxRuntime to export trained models into a compact, executable format. It supports 30+ languages and hundreds of distinct voices, achieving real-time performance on a Raspberry Pi 4 (Quad-core 1.5GHz) without requiring a GPU. The system integrates directly with Home Assistant and Wyoming protocol, making it the standard choice for privacy-focused local voice assistants.

https://github.com/rhasspy/piper