Add Tortoise

[behnaz-motavali]

Tortoise is an expressive, multi-voice text-to-speech program created by James Betker. It uses both an autoregressive decoder and a diffusion decoder, which contributes to its slower speed. Released in 2023, the author claims the model is open source. Given its functionality, it would be a good idea to add it to audio generation models.

https://github.com/neonbjb/tortoise-tts

[liesenf]

Couldn't find the OS claim?
Otherwise looks good to go

[DBlankvoort]

Bottom of the arXiv abstract I think: https://arxiv.org/abs/2305.07243

[liesenf]

“ All model code and trained weights have been open-sourced at
https://github.com/neonbjb/tortoise-tts.” Also a gray zone I reckon.