Piper is a fast, local text-to-speech system optimized for Linux, and Raspberry Pi (it can also be run on Windows via Python). It features pretrained voices in over 30 languages and can accept text input or JSON to synthesize high-quality audio. Piper is open-source and written in C++ and Python, with binaries available for desktop Linux, Raspberry Pi, and more. You need to provide a onnx model and config file to generate speech from text; some example usages are provided. The code, documentation, and samples are available on GitHub under an MIT license.
