Closed sbenthall closed 5 years ago
While there is now a non-breaking setup for a 'style transfer' from one fragment to another, the 'transfer' is not audibly convincing.
This is likely due to the weakness of the current model architecture, which probably is not capturing content and style in an appropriate way.
With #18 this seems to be working...
Using the song recognition model from #10, and given two fragments, perform a style transfer from the second fragment to the first.