"Steven Sullivan" wrote in message
Actually it's from a 2007 AES conference, not the JAES,
so I'm not sure it's peer reviewed
Conference papers aren't reviewed at all - only the title and abstract have
been seen by the session chairman.
...but anyway:
"Which of the Two Digital Audio Systems Best Matches the
Quality of the Analog System?"
http://www.hitech-projects.com/hera/...s/aar07pu4.pdf
It appears that the formats compared to a live feed
(analog) were DXD (353.8 kHz/24) and
44.1/24, both in surround, using a blind comparison
protocol.
It appears that the program material is best described as being "sound
effects".
Also, every time some high resoution advocate cites this paper, ask them to
reflect on how closely it resembles listening to music for pleasure in a
residential listening room. ;-)
Two additional listening conditions were tested
: one where the A/D signal bandwith was 100 kHz (thanks
to special microphone 'super-tweeters') and the other
where bandwidth was limited to 20kHz. The authors say
their results show that listeners 'more often than not'
identify the hi-rez audio (and not
the 44.1 kHz audio) as being similar in quality to the
analog feed...but only when the bandwidth is limited!
Let's cut to the chase and look at the raw data on page 19.
As I read it, there were 10 subjects and 6 listening sessions for which
individual responses were required. IOW, 60 trials.
I pasted their matrix into Excel and tried to do some quick sums. I came up
with Test Condition 1 = 23/60 and Test Condition 2 = 31/60. IOW, one test
produced an outcome that was worse than random guessing, and the other was
random guessing.
Results that are worse than random guessing may cause some head scratching,
but they are not all that unusual in experiments like this where
communication between the listeners can affect the outcome.
The most probable explanation for worse-than-random results is that that
some of the listeners were basing their results on their perceptions of what
other listeners were perceiving, and the total number of independent
responses was far less than what you get from a naive count of the actual
responses.
IOW, the actual situation was not 23 independent responses of 60 trials, but
maybe more like 4 independent responses of 10 independent trials, so the
true numbers were so small that statistics doesn't really apply.
I think these results would pretty well explain themselves to just about
anybody, were they reproduced any place within the actual paper.
Short answer - the outcome was random guessing, and both the test itself and
the statistical analysis were greviously flawed.
I guess we can chalk this paper up as yet another "Proof by complex
statistical analysis" which defies common sense.