I agree it is difficult. But then, look at the typical home session.
"You know I have these superduper extra expensive cables here that I really feel gives a much tighter feeling to the whole rythm session. Listen for yourself and you will hear it"
Errors:
1 - you tell the listener what he is supposed to hear
2 - you give the listener only one chance to hear the change or even none
3 - you as test designer will be disappointed if the person does not hear what you want him to hear
and so on.
Try this instead.
1 - make sure the listener cannot see which cable is beeing used. Screen, behind the boxes, whatever.
2 - write 10 small paper pieces with A on them and 10 with B on fold, fold them and put them in a jar.
3 - disconnect both cables.
4 - draw a paper from the jar (A or B) and connect the corresponding cable. Again so that the listener cannot see. Note out of sight for the listener which it was.
5 - play the same song (maximum first 30 seconds or so)
6 - ask the listener to keep a protocol for the difference between this listening and the previous: better / same / worse.
7 - repeat at least 7 times, preferrably more.
After the test has been done it is possible to apply statistical principles to the results obtained. Look at the change form A to B (as example) if this happened four times and the listener answered 2 times better and 2 times worse we are talking pure chance. Not much difference really with 3 to 1 as this could be pure chance again. In order to get a valid result we need a larger sample or a 4 to zero score.
If you want to make the test even better, invite a sceptic like me to conduct the test and several people to listen and hidden to each other keep the protocol. Next time I am around your part of the world I volunteer.
Remember that bias happens every time we as humans are asked to compare things. This is proved again and again and again in science and in practical tests in marketing.
// Gunnar