Whisper and other TTS support

[MerijnHendriks]

Hey, thanks for the hard work as always!

I was looking at adding TTS to my setup, and I see whisper being supported.
On the repo (https://huggingface.co/koboldcpp/whisper/tree/main) I see .bin models, which makes me wonder

• How did you produce them?
• How can I produce my own?

I want to run whisper-large-v3 (https://huggingface.co/openai/whisper-large-v3) at fp16 and possibly at other quants.

In addition, I saw that parakeet (https://huggingface.co/nvidia/parakeet-tdt-0.6b-v3) seems like a meaningful innovation, considering that whisper is already quite old.

Is it possible for it to be supported as well?

[LostRuins]

Support is provided by whisper.cpp, models are converted by ggerganov as part of the whisper.cpp project. You can browse all supported models here https://huggingface.co/ggerganov/whisper.cpp/tree/main