You know what, it's frankly bonkers just how much music and other audio Youtube must be comparing your video audio against. Considering that random copyrighted music could start and end at any point in your video, must contest against other audio (ie. talking over it), and yet somehow it's more likely than not able to find anything from 80s rock to an obscure anime soundtrack!
Think of the application such algorithms could be good for, but nooooo.
