Post by JerrySI don't want to turn this into an argument because I just don't know
enough about it. My point is that if the brain knows that A and B are
different file types then it is going to be working pretty hard to
detect audible differences. If it finds such differences then it is
easier to discern them in X because the brain now knows where to listen
for them (different phrases, difference notes, different instruments
etc). If on the other hand the first two plays in triangle testing are
from the same file, then the brain has not prelearnt the cues to pick up
the differences on the third play. It would not be surprising if the
sensitivity and specificity of ABX and triangle testing are different
because of this prelearning.
I think that "pre-learning" often takes place long before anyone sits
down to a particular ABX, triangle, or any other type of comparative
listening session. Indeed, there are people who consciously train
themselves to identify the differences in lossless versus lossy, and
even more casual comparative listeners learn to pick out certain
instruments that reveal differences. And it seems there are a few who
train themselves in identifying the differences between lossless and
lossless (misguided in my opinion, but that's just my opinion).
I don't see why having those cues about where the listener _thinks_ they
hear a difference between A and B is a bad thing at all. It most
certainly does not bias the test results. Keep in mind that often there
are differences that exist but that listeners cannot hear (see
Archimago's recent thread and blog on comparing lossless and lossy). If
they really can hear a difference, then that will be borne out in their
assignments of X to A or B.
------------------------------------------------------------------------
aubuti's Profile: http://forums.slimdevices.com/member.php?userid=2074
View this thread: http://forums.slimdevices.com/showthread.php?t=98828