Autodetecting the limited core materials

[LifeIsStrange]

Hi,
someone on hackernews made the following comment:

Most of this writeup is just a reinvention of Schenkerian analysis, and suffers from the same problem, in that you exercise a lot of editorial judgement in deciding which parts are the core/structural ideas and which parts are embellishment. That undermines the whole idea that this is automatic composition, because you are deciding a heck of a lot upfront. https://en.wikipedia.org/wiki/Schenkerian_analysis

I think you did a great blog and this project is fascinating. However, the graal to me would be to produce a software that can take a music as an input and "easily" allow to generate variations of the composition.
As such I'm wondering wether machine learning techniques, such as a neural network would be useful, not to generate the music but to help your rule based program to generate it by classifying/detecting in a song, which notes are the most likely to be parts of what you call "limited core materials" AKA the primitives to remix.

[flujoo]

@LifeIsStrange Thanks for your comment. I totally agree that the autodetection functionality you mentioned is essential, which my package currently does not have. I think deep learning is very promising in doing this job, and I should spend some time studying it before considering how to Implement this functionality.