Kitten TTS announced, real-time AI speech on minimal hardware

Kitten TTS is an ultra‑lightweight, open-source text-to-speech system designed to run in real-time on modest CPUs and even mobile devices. The smallest model counts just 14–15 million parameters and ships as a sub‑25 MB file.

The system also comes with additional 40M and 80M variants that still stay comfortably under 100 MB. Users can select from a set of preset voices and adjust generation speed via a public demo on Hugging Face.

Despite its tiny size, Kitten TTS delivers reasonably natural and expressive speech, handling tricky disambiguation sentences. This includes the likes of ‘The record producer refused to record the band’s new single” or “The wind was too strong to wind the kite string around the spool.’

The project provides full instructions for local installation in its GitHub repository, emphasizing CPU‑only and mobile‑friendly operation. While it doesn’t yet support custom voice cloning, it offers a combination of small footprint, real-time performance, and open licensing.

This makes Kitten TTS an appealing alternative for those needing a building block for embedded voice assistants and offline applications.