Show HN: Three new Kitten TTS models – smallest less than 25MB

A Show HN post introduced three new Kitten TTS models, with the smallest under 25MB — 30% lighter than the runner-up in efficient TTS models. The largest model, at 150MB, still offers voice synthesis on local hardware, a rarity in 2026 when cloud TTS dominates. These models achieve 95% phoneme accuracy, outpacing many 500MB+ alternatives. This represents a quiet win for offline, low-latency applications, though it trails Astral's 964 upvotes in community impact.