[sdiy] some observations : audibility of phase distortion

[Czech Martin]

During the vacations I made some experiments.
Unit pulses and a short drum sample were filtered
with a discrete all pass filter. I checked that the amplitude
of that filter was flat, but phase was very nonlinar.
The nonlinearities where arround 100-400 Hz.
I tried 2nd , 4th and 6th order.

The signals were visibly distorted, but:

I couldn't hear a difference, even if the original sample
follows imediatly after the processed with less then
1s distance. Looping serveral times and using good headphones
did not help.

What is wrong with my experiment?
I think that the time domain effect is still occlused (SP?),
i.e. the filter response is too short  to be heard.
That would mean higher order and / or higher filter "Q",
i.e. poles closer to unit circle.
E.g. if two pulses follow too close to each other you will
hear only one, with perhaps some coloring. The ear has some
time limitation.

m.c.