I think it's a common mistake, especially in classical music, to think about musical traditions as "rooted in clear rules and numerical relationships." It's analogous to thinking that language is rooted in clear rules and structures.
The truth is just the reverse, the rules, numbers, and structures are rooted in music. They exist to describe an organic, emergent cultural mechanism that is continuously changing. Ask yourself: which came first, music or theory? Language or grammar? There are plenty of musical styles with no formalized structure or numeric relationships, just as there are plenty of languages with no formalized structure or even spelling. Grammar and music theory/notation are fingers pointing at the moon, and we are looking at the finger.
And when we try to engineer systems to help people point at the moon, we should be focused on the finger. Musical structure and numeric relationships are the way people communicate about music, and the tools should speak their language. Unfortunately they have to grapple with the painful inconsistencies in that language and those structures. Computer music comprehension has problems analogous to computer language comprehension - emergent complexity so high it took neural nets to finally achieve real utility.
PS - it's true that Western music has, near the bottom, some relationships that were mathematically derived by none other than Pythagoras, among others. Relationships of fourths, fifths, octaves, and equal temperment all had numeric justification... But that justification was still only there to describe the practice which had already become common, an explanation of "why this sounds harmonious" as well as a proscription for "how to sound harmonious". That some rules are internalized by some generations and broken by others illustrates exactly the problem with the system.