How to estimate the number of repeated musical phrases in audio?

Question

Mathematics

The music is based on repetitions. In the classics, the repeating parts are long and there are few repetitions. In pop, a chorus 8–16 bars long can be repeated N times. In electronic music, there are continuous loops, but usually short, only 1-4 measures, but 100 times each. It is
interesting to analyze the recordings for the number and length of repeated phrases - parts of one or several instruments. Given that, over the repetitions, there can be, say, unique vocals.
Something like autocorrelation and Fourier transform. Or wavelet analysis.
How to practically approach the problem?
upd. found a study of a similar problem by the Finns: Music Structure Analysis by Finding Repeated Parts (pdf)
upd.2 is the so-called. VAMP plugins– developments of universities and research groups – there is both autocorrelation and selection of notes and segmentation of the track into choruses / verses, selection of tempo and measures.