Support Text to speech models

[gianpaj]

e.g. https://deepinfra.com/canopylabs/orpheus-3b-0.1-ft

Also, it would be great if it works with streaming and there's an example on how to make this work in React (Next.js API + React front-end)

Thanks