Comment by zhyder

Comment by zhyder 2 days ago

2 replies

V v cool: first time I've seen such expressiveness in TTS for laughs, coughs, yelling about a fire, etc!

What're the recommended GPU cloud providers for using such open-weights models?

JonathanFly 2 days ago

> first time I've seen such expressiveness in TTS for laughs, coughs, yelling about a fire, etc!

The old Bark TTS is noisy and often unreliable, but pretty great at coughs, throat clears, and yelling. Even dialogs... sometimes. Same Dia prompt in Bark: https://vocaroo.com/12HsMlm1NGdv

Dia sounds much more clear and reliable, wild what 2 people can do in 3 months.