1: Consonant polyphonic version:
2: Dissonant polyphonic version:
3: Consonant monophonic version:
4: Dissonant monophonic version:
This figure plots dissonance (y-axis) versus time (x-axis) for each
synthesized melody. Notice that the difference between predicted dissonance
for the regular monophonic (black) and dissonant monophonic (magenta)
melodies is much greater
than the difference between the consonant (blue) and dissonant (red)
harmonizations. In fact, the red line sometimes crosses the blue line, because
the model apparently considers the traditionally dissonant minor 9th interval
less dissonant than the traditionally consonant minor 6th. It does
consider the minor 2nd to be extremely dissonant, as one would expect, but
surprisingly not that much more than the consonant intervals of a major or
minor 3rd.
In case using such a small number of harmonics was to blame for the slightly
surprising results above, I also tried using voices with 6 harmonics. They
sound a little brighter, but the results are comparable.
1: Consonant polyphonic version:
2: Dissonant polyphonic version:
3: Consonant monophonic version:
4: Dissonant monophonic version:
