I’ve been playing with plotting various EN analysis quantities against one another. I thought that pitch vs loudness or pitch vs segment duration might turn up something interesting, but visually at least, there’s not much of interest. Then I tried loudness vs duration, and wow! Some pretty distinct distributions. Parker playing Ornithology is fairly consistent, whereas our 10 versions of Autumn Leaves by ten different ensembles are more varied. Almost all of the tracks seem to have mostly short, loud segments. That might just be the nature of segments — they’re distinct events, so very quiet moments probably don’t end up as independent segments…
Ornithology:
Autumn Leaves:
