Please help with stable-audio-open model

[asigalov61]

Hey guys,

I was wondering if someone could help make an auto-regressive continuation script/code for stable-audio-open?

I want to see two features:

Continuation from any given audio file and a prompt
Auto-regressive continuation with a single prompt (or multiple prompts)
This would allow to continue audio files which this model should be very capable of
And it would also allow to generate long compositions auto-regressively (i.e a song based on lyrics prompts).

So if someone can help or get me started in the right direction, I would really appreciate it.

In a form of a Google Colab or HF Space, please so that I can test it :)

Thanks in advance!

Alex.