Bob Lee (QSC)
In case you missed it, I work for QSC Audio!
- Jul 3, 2001
- 12,668
- 2,036
- 67
- Disclosures
- Former Technical Communications Developer, QSC Audio
Originally posted by MKS
As a qualified statistician I would be very pleased to help design a blinded test of these cables.
Unfortunately it can't be double blind unless you can make two different power cables look identical. Double blind means that the person hearing the results and the person switching the cables does not know which cable is which.
My thought would be that you need to have two or more testers / listeners who will listen to and rate the results. You ought to use one standard test rig with an identical signal going through the rig each time - this could either be a looped signal or a standard test signal. Presumably you would have a some kind of standard signal that covers a variety of frequencies... Then you need to listen to this test signal a number of times (10+?) while switching the cables *randomly* for each listening but making sure that each cable gets used the same number of times (viz. Cheap cable 5 times, Tara Labs cable 5 times). The person switching the cable should not be one of the listeners and if possible should not be visible by the listeners (so that they don't know if that person "switches" between one cable and the same one. i.e. that the same cable is used in two adjacent tests). The listeners decide whether what they are hearing is the Tara Labs cable or the cheaper cable. They don't *have* to rate what they like about the sound etc. Simply A or B will be enough.
Once the ratings have been completed it is a really simple statistical test to test if the listeners are guessing which cable is which and if they can hear a difference.
If you would like to take more in-depth measurements or qualitative measurements, then I'd be happy to provide input to the design and analysis.
Hi Mike,
We have QSC ABX comparators at our disposal, and they track results for up to 25 trials in a session.
For the sake of completeness, I would plan also to do "same-different" tests. Tom Nousaine has used those for situations where ABX testing was not practical.
You are correct that it first has to be established either that there is an audible difference, or that no evidence exists that there is. Only if an audible difference can actually be demonstrated can anyone truthfully judge whether one sounds "better" than the other.
A major fallacy of most comparative listening "tests" is the a priori assumption that the change will cause an audible difference, leading to an abundance of false positives. That's why the stories of audio "experts" claiming to hear great differences when they think the speaker cables have been changed, but in fact the cables haven't been, ring so true with anyone who's been involved with comparative listening tests.

This just makes your whole logic seem deeply flawed!