🚀 KittenTTS – Ultra-Tiny, Supurr-Expressive Open Source TTS Just spotted an awesome open-source TTS release for builders, hackers, and anyone who loves running clever AI on any device. KittenTTS is a new, super compact text-to-speech model from Kitten ML. 8 expressive voices (4 female, 4 male) Preview model: ~15M params (under 25MB!) Runs anywhere: Raspberry Pi, your phone, ancient laptop—zero GPU required Open source and free for all English (for now), with multi-language support on the roadmap Great for tinkering, edge devices, indie products, or just geeking out over efficient models! 🛠️ GitHub: github.com/KittenML/KittenTTS Huggingface: huggingface.co/KittenML/kitten-tts-nano-0.1 Full (80M param) model coming next week. Give it a try!

Whoa, 25MB TTS?! That's truely awesom. Running complex AI models on low-power devices has always been a pain, so this whole "runs on a Raspberry Pi" thing is kinda genius imo. So excited to try it out on my old laptop – what's the latency like on something that low-powered, though?

KittenTTS

SOTA Text to Speech Model | 25 MB | Runs in Browser

SOTA Text to Speech Model | 25 MB | Runs in Browser