Wylie Williams wrote:
I recall in olden days Consumer Reports tested speakers with a method
that compared speakers under test with their reference(AR3a). It was an
elaborate system that they considered scientific, to see how closely other
speakers could mimic the sound of the AR, but I wondered how it was able to
distinguish a speaker with better resolution than the AR3a.
This if prefatory to a conversation with an audiophile friend today. I
mentioned that I was not quite pleased with my system's sound. He asked
"Haven't you listened to those Tara Labs cables the rep loaned you? ( I have
some expensive ones on loan, and some much cheaper ones I own and use) I
replied that I had but hadn't heard enough difference to be sure it was
real. Certainly no significant improvement to my ears.
His reply was that the problem is probably that my speakers are too low on
the food chain to discern the difference, and offered to loan me his when he
goes out of town soon. He says I'll hear the difference then.
This raises an interesting point about all those listening tests we hear
about. Maybe the speakers in the tests have too little resolution to allow
differences to be heard.


Actually it's one of the handful of same old point 'audiophiles' use in
such arguments, but it doesn't get around the need for bias controls.

Also, tests of audibel difference can and have been done using the claimant's
own system, where they'd *already* claimed they heard a difference, sighted.
The famous 'Sunshine Audio test is one such example. And ABX tests
incorporate the idea of switching back and forth between A and B in the
belief that you hear a difference (if you don't believe you hear a
difference 'sighted', then continuing the test can only reveal whether
there was a bias towards perceiving 'sameness') . Then you test whether this
difference is real, by testing whether 'X' is identifiable as A or B.



--
-S.