Findings 4
Phase:
Resynthesizing simple K5 waveforms using only amplitude from the fft
doesn't always do the trick.
The result sometimes looks and sound different or
the result looks different, but sounds the same or
looks similar but sounds different.
So what about phase? Does it matter?
This is one cycle of the K5 sound SWWOOD
This is Igors analysis of SWWOOD. Observe that harmonics 5, 9, 10,
12.... are almost -pi out of phase(green). The others are reasonably within
phase.
This is the Igor and K5k resynthesis proposal(they agree it seems).
The result looks and sound different.
So, observing the phase relationships and knowing that the K5k is phase
linear I removed harmonics 5, 9, 10... since they would actually contribute
in a "inverted" way to the wave.
The result sounds and looks more like the original. Hmmmm...
Resynthesizing a whole sound using phase information from just the
first cycle results in a sound more natural and more like the original.
Phase matters!
<<--Go back!