Kitten-TTS-2 ships 4MB Apache-2.0 TTS models in 16 languages
ElectricalBar7464 · reddit · 2026-11-20
KittenML open-sourced Kitten-TTS-2 (Apache-2.0 code and weights): Pico (7M params, 4MB, 8 languages), Nano (17M, 8.8MB, higher quality) and Mini (80M, 48MB, 16 languages). Pico runs faster than realtime on a single smartphone CPU thread and even on a Raspberry Pi. The new pico is 6x smaller than the old 25MB nano, with a big quality jump and new multilingual support. GitHub, HF weights and a live demo are available.
Related event: KittenML Open-Sources Ultra-Lightweight Kitten-TTS-2 Models(2 posts)→
More from Infra
- Alphabet plans $195B–$205B in 2026 CapEx, renting external GPU clusters to bridge Q4 demand — Money-String4165 · 2026-11-20
- NVIDIA general counsel files to sell 30,000 shares, SEC filing shows — churchkey · 2026-11-20
- Google Cloud makes the case for Envoy as the open standard for AI and agent traffic — rseroter · 2026-11-20
- QuixiInfer author: tell Claude 'defeat is not an option' to make it write kernels for your GPU — QuixiAI · 2026-11-20
- Volcano launches Agent Cloud: self-healing deploy platform built for coding agents — rohanpaul_ai · 2026-11-20
- Running Krea-2-Turbo on a GTX 1650: 18 minutes per image cut to 30 seconds — A_NASHEX · 2026-11-20