Mimic 3

Mimic 3 served as the voice of Mycroft AI's Mark II smart speaker: a fast, fully offline neural text-to-speech engine built for privacy-focused voice assistants on modest hardware such as the Raspberry Pi. It runs locally with low latency, offers a command line tool, a web server on port 59125, and Docker images, and supports a range of languages and voices distributed through a companion repository. The engine is open source under the AGPL-3.0 license. The project is no longer actively maintained, and the repository points to Piper TTS, from the same developer, as its spiritual successor carrying the same offline, low-resource philosophy forward. Mimic 3 still works and remains of interest to hobbyists running legacy Mycroft setups or studying compact neural TTS design, but new voice assistant projects generally start with Piper instead.