"Bob Marcus" wrote in message
news:blBQa.60322$H17.19111@sccrnsc02...
"Harry Lavo" wrote in message
...
Wait a minute, wait a minute. Only sound? There is no music until the
brain has processed the sound and interpreted it as music. And that is
the
primary reason music reproduction is not simply electrical and physical
engineering. Their is no way to measure *music*. Ultimately whether
the
music strikes our brain as right, or the brain tells us something is
amiss,
is not "objectively" measurable. The only way to objectify it is by
allowing humans to interpret it as music, and then to develop tests to
try
to record that *subjective* response in ways that can generate some
*objective* results in the statistical sense. And there is the rub.
It
demands context for the brain to interpret what is going on. Let me
give
you a non-musical example that I use here before...about a year and a
half
ago, I think.

Hmmm, I smell a bad analogy coming on...

Always at your service to amuse and please, sir............


Suppose you hear a split second of a car crash. That's all...half a
second
of indecipherable noise. You wouldn't even know what it was. However,
if
you heard a recording of street sounds, and auto approaching, a squeal
of
tires, and then the crash, you would know what you were hearing. And if
you
heard it through two different systems you could probably which one
sounded
"most real".

Actually, you probably couldn't, unless one of the systems was
grotesquely bad. That's why ABX and ABC/hr tests NEVER ask subjects to
compare a sound to something they've remembered. ALL comparison sounds
are immediately available. And that's why this analogy is wrong.


Ever been 50 yards from a car crash. I have been, twice. Their are
elements of the sound you are not likely to forget (burned in by trauma, no
doubt, and also pretty unique among everyday sounds). Morever, these
elements are not likely to be handled well by many systems (dynamic
response, high frequency transients, low frequeny power). I suspect I could
make a judgement.

However, if all your heard were two snippets of sound of the
crash itself, my guess is your brain would be trying so hard to make
sense
of what you were hearing you couldn't evaluate anything in the way of
which
sounded most "real" because you didn't know what "real" was.

Let's keep in mind this concern of Harry's about what sounds most
"real," because he stumbles over it later on.

No, lets keep the focus on what this means; if the brain is trying hard to
make sense out of something other than intepresting the musical reproduction
"as music" then it is likely to botch the job of evaluating/identifying the
musical significance of what it is hearing. Like perhaps focusing on
"difference" and using relatively short excerpts, rather than more relaxed
monadic evaluation.


Something similar happens with music but even more complex. Because
scientist now know that the brain is hardwired to respond to this thing
we
call "music", both rhythmically and emotionally. Further the work done
by
Oohashi et al (The Journal of Neurophysiology Vol. 83 No. 6 June 2000,
pp.
3548-3558) indicated that this emotional response took place over time,
as
much as twenty seconds of time, from the time of the sound.
Presumable this
is the time it takes the brain to recognize and interpret the music as
pleasurable, unpleasurable, rhythmically coherent or incoherent, etc.

I think you're stepping a bit beyond Oohashi's findings here, Harry,
and you really don't need to to make your point.

Oohashi himself made this same speculation. But the advances in scientific
investigation of human response to music are apparently real and documented.


This fundamental fact means that you cannot measure "sound" and
determine
its impact as "music". The factors affecting how we respond to music
are
apparently very subtle and "time-based" and "harmony-based" and not
static.
But a short burst comparison without much in the way of context for
recognition, relaxation, and response (which is the way most
short-interval
testing is done) tends to short-circuit the process. This is the
objection
must "subjectivists" have to ABX'ng in practice and why they question
"null
results" that seem so at odds with so many people's otherwise fairly
clear
perceptions of differences.

Oohashi also says that the subjects were not consciously aware of the
difference, so it's a little hard to see the connection between his
results and subjectivist claims.


Not at all. Subjectivists have all along stressed that fact that "test
anxiety" alone may interfere, and have argued that the best test is the
serial monadic approach, ie relaxing and listening to set pieces on
equipment in a familiar system, in familiar surrounds. Then doing the same
with an alternative piece of gear. Taking notes on both. Going back and
repeating, but in a relaxed fashion which still enjoying the music. Taking
careful note on the emotional responses (or lack thereof) and rythmic
responses (or lack thereof) elicited by the DUTs. And finally have a clear
preference emerge and 'reasons why" in musical terms. Very, very close to
what Ooashi et al did but they did it even better in that it was a blind
test and the subjects didn't know much of anything. And yet what emerged
was statistical significant results in favor of one variable.

Oohashi et al indicate in the quoted article that they have confirmed
this
speculation. That is, they have used short-interval comparisons of
music,
and found "no difference" in ratings in line with accepted believe. But
when using "long-intervals" using the same stimuli and sequentially
monadic
ratings, the achieved statistically significant differences in response
to
the two stimuli.

I don't think so, Harry. What they did was to do ABX-type tests using
short snippets of the music they were using, and found no difference.
So far as I can see, however, they did not use short snippets in their
monadic ratings.


You are absolutely right...but that is my point. Most of the objections
raised in this forum have been to the heavily promoted use of abx testing
and the supporting claim that it is the most sensitive test for evaluation
and shows no differences in most cases in audio gear (speakers and
cartridges excepted). Subjectivists feel there are much better ways of
testing that focus more naturally and fully on musical evaluation and
believe that if differences in preference occur, the fact that there has to
be a difference in some factor of reproduction is a given. Not all
subjectivists here are anti-dbt. Most are anti-abx, it seems to me.

You "amps is amps" people seem to want to ignore this finding, which is
pretty earthshaking and has nothing to do with whether you think the
"ultrasonic" portion of this test was done correctly or not, since
presumable the same stimulus was used in the precursor test (although to
be
fair this should have been better documented in the article rather than
treated as almost a passing reference). This finding alone, if
substantiated by others, would rule out much of the abx and possibly
most of
the dbt's done to date.

Huge leaps here. First, as I just noted, they don't seem to have done
the key comparison, which is applying the same test to both long and
short snippets. (That's not a criticism of them, by the way. You're
the one who's trying to make more of this research than is really
there.)


I agree from the standpoint of methodological rigor. However, in the real
world a comparison of blind protomonadic testing using full samples of
music, vs. one that apes a standard abx test with shorter snippets, and
which consequentially shows the former to give statistically significant
results of preference (and therefore difference) while the other shows the
accepted "null hypothesis" is highly significant. It suggests that critics
of abx testing "as practiced" appear to be more right than wrong.

Second, there are a few huge gaps in the theory here, as Oohashi & co.
concede. In particular, there's the little matter of how this section
of the brain gets its information, since the normal hearing mechanism
cannot supply it. Also, there's the little matter of the data. Where
is it?


The summary tables are there, but not the individual results. However, the
statistics are so overwhelmingly significant that it is probably less
critical than if the results were marginal. I would guess that Oahashi et
al would provide the raw data to critics and other researchers if asked; my
impression is that most journal articles do not provide the raw data from
such research but instead provide relevant summary statistics as they have
done.

Then, as I also noted, the relevance of this to the experience of
subjectivists, listening sighted, is open to serious question. If
anything, this study confirms the necessity of DBTs, potentially
challenging only their methodologies.


It certainly was an excellent approach, IMO. I don't think you would find
many subjectivists objecting to this type of dbt'ng. It is, as noted, a
more rigorous approach to what many already do.

Finally, I would note that Oohashi provides absolutely no evidence
that his subjects could tell which of the two samples was closer to
"real." One might interpret his results to suggest that they found one
more appealing than the other, but even that is hard given the lack of
data. For that matter, they have not excluded the possibility that the
inclusion of ultra-high-frequency noise, as opposed to harmonic
information, might be responsible for their results, or cause similar
ones.


I won't dispute the fact that "real" was not the evaluative criteria here.
My focus on "real" is a) one part for accuracy (after all I was raised in a
"hi-fi" family), and 2) one part reproduction of a live musical event. The
assumption is that most of us listen to music to feel good, and we tend to
play the kinds of music that we have heard live most often, experiences we
usally have a "pleasure response" to. So for me, the "real-er" a sound is,
the more likely the music performed on my system will give me pleasure. I
do understand your point, but whether "real-ness" or "positive emotional
response" is the evaluative criteria, I believe the brain has an important
role to play that negates the "all we are measuring is sound differences"
simplistic approach to component evaluation.

In short, I'd say this study is interesting, and opens up some
possibilities that need to be explored more fully. But as I've made
clear before, I'm no expert in this field, so I'd defer to the
opinions of those who are. I recommend you do the same, Harry.

bob


Thank you for your thoughtful critique, Bob.