I've been participating in forums in the US, the UK, France, Spain and Portugal for almost 15 years now.
Some people like "coloured" gear. Is it that too difficult to accept?
It's not just valves or vinyl that people like; some people put their speakers near the side-walls with no toe-in; some people power hard to drive low-sensitivity loudspeakers using 5 or 10W amps with high output impedance; some people like Redbook over NOS DACs; et cetera.
Many signal-correlated distortions increase the sense of "spaciousness", a bit like reverb. Others make the tonal balance perceptually "warmer". Others still mask problem up- or downstream...
All these cases
may be (I don't claim they necessarily
are) the result of sighted bias. Where is the scientifically gathered data showing that people prefer these things under conditions in which sighted bias has been removed?
Anyway, for the sake of furthering the discussion, I've used RePhase to create a couple of impulses (download
here) that add high-Q pre- and post-ringing, with
no effects in the amplitude domain. You can convolve these in software with a recording of your choosing and then use Foobar's ABX comparator, for example, to determine whether you can reliably detect a difference between the filtered and unfiltered signals.
I chose a centre frequency of 2.5kHz as it is within the range where our ears are most sensitive. The impulses are generated at three different sample rates: 44.1kHz, 48kHz, and 96kHz, to give people some flexibility in running the tests.
Note that these impulses include
both pre- and post-ringing, and therefore are more severe than the kind of ringing you've been describing
@tuga, since in the cases you're describing there is only post-ringing. Unfortunately, I couldn't think up a way to do this in software such that only post-ringing would be generated while the amplitude response would remain unaffected and the group delay would remain constant.
Perhaps someone else here has some ideas on how this might be done?
Anyway, if you run this test and get a positive result, this leaves open the possibility that it is the post-ringing that is audible (but note of course that the research is quite clear that the pre-ringing is far more likely to be what is audible).
So yes, a flawed test, but at least one that's biased towards detection rather than non-detection.
Here are some graphs generated using REW showing the behaviour of the filter at a sampling rate of 96kHz (but in any case, within the audio band, all three impulses are identical).
Amplitude response:
Step response:
Waterfall (only post-ringing visible):
Spectrogram (only post-ringing visible):