I'm realizing that a lot of the analysis objects (like fiddle or zsa.freqpeak) might be more "accurate" if I somehow adjusted for Equal-loudness contours before. For instance when analyzing program material, it's really skewed toward bass and crash cymbals.
