Quote:
The set up you describe nick_charles is hard though as he has no reference for A and B. When I think of double blind I think of the foobar ABX plugin - you get an A, B, X and Y sample and need to match A to X and B to Y or A to Y and B to X correctly. Also the time between trials will be large while he gets ushered in and out of the room, audio "memory" is shorter than that no?
Still, I hope some kind of experiment goes ahead, I'm very interested
It is less of a problem than you think. The
guinea pig subject can listen sighted as long as he/she wishes
before the test. This allows the subject to be familiar with the
sound of the two stimulae. I should have mentioned this but it is common practice in audio tests to allow listeners to have sighted experience first - sorry, my bad. So before the blind tests the listener has a conception of how the two items sound and the blind part is to confirm that their conception is accurate when other clues are removed.
I agree there are issues with delay and audio memory and a random ABX box is a
much better solution in general.
However, most DACs have only 1 USB input and so even if you have a switching box with 2 USB in and one out you have an extra USB cable so you would need two of the super-duper cables otherwise an intervening bog-standard cable would be deemed to nobble the super cable, then you have the issue of feeding two identical signals into the box so you would need two computers running the same software/music and carefully time-aligned on each trial, even a slight misalignment would easily identify A from B when compared with X.
If you had a DAC with two USB inputs you could switch between them , but then you do not have an X just A and B, you still have the sync problem even if you have one computer running two instances of the same media player through different USB ports (if this is even possible)
You could do rapid cable swaps behind a curtain (Single blind) but then you have the Clever Hans problem where the invigilator can consciously or unconsciously give clues as to the identity of the cable, perhaps one is harder to fit or takes more effort to remove.
One way you might manage it would be to have two identical DACs fed from two identical computers carefully sync'ed, feed the fixed line-level outputs (using identical analog cables) to an ABX box which sends A or B to an amp depending on the switching.
Personally how I would do it would be to digitally record the analog line-level outs from the DAC fed by both cables, run the results through a spectrum analyzer (run the math on them) or set up Foobar blind tests between two carefully trimmed and aligned samples, but then the objection is that the digitizing process obscures the differences between the cables even if it is accurate to 16 or 20 bits.
