Best Practices Question: Making a Python package heavily derived from existing code

Without getting too much into the details, here's my current situation: there's an existing repository that contains a .py script that allows you to generate a music video. You run the script with the required arguments, and it produces the video file for you. The author of the repository also has a YouTube channel where he keeps his own portfolio of videos generated by this script. There isn't any information on the repository about licenses.

What I'd like to do is revamp the code into a full-on package with additional features to maximize customizability. This way, a Python beginner can easily define their own custom functions on a notebook and pass them as arguments to an object, say, to add effects to the music video. This is one of the many additional features I'd like to add to the package.

From what I can tell, pretty large chunks of my code will be heavily derived – or even copy-pasted – from the existing repository. In addition to that, I'd like to publish a blog afterwards to promote the package (with due credit given to the original author, of course). In this case, what's the best practice? Should I contact the author before pursuing this? Or, should I go ahead and make the package, and just credit the original author in my code comments, Readme, and links on the blog?

Thanks a ton!