To differentiate between solo and non-solo music, I decided to see if could find the sparse pieces of the spectrogram. But what exactly does it mean to be sparse? To simplify the picture a little, let's look at a couple time slices of the spectrogram during a solo period:
and during a non-solo period:
As you can see, the frequency magnitude really stays small for most frequencies in the solo time slice. Just look at the percentage of frequencies that are below -20 dB in the solo section versus the non-solo section. I figured that maybe I could differentiate between solo and non-solo music by looking at each time slice and calculating the percentage of frequencies that are small.
<<Prev Up Next>>Christine Smit