[andrew3726]

Sure, that's called audio super resolution, there's a few papers/projects doing that. Haven't really seen models which are robust and have good generalization though.