Re: The Value Of DBT

[EMAIL PROTECTED] (Nousaine) wrote in message news:<[EMAIL PROTECTED]>...

Snip previous discussion:
Let's begin at the end. Tom said:

> So Ludovic; what if I ABX speakers and get positive results? Will you 
> then provide a corresponding bias-controlled experiment that confirms 
> the validity of your method?

                     I have no method "valid" for you, me and all and
sundry. I never had. I don't believe you have it. I don't believe one
exists or could exist.
               Let's agree what we mean by a "method". This is RAHE-
not the JAES or J.of Psychoacoustics Research (if there is such a
thing)
                The "method" must be a fool-proof technique usable by
ordinary audio consumers to establish differences between comparable
components. (With view to choosing one over the others of course
because... why else would one bother?)
                 I repeat what I said: "ALL the available listening
PANEL tests comparing ANY roughly comparable audio components ended in
 "they all sound the same",  negative verdict by majority of
panelists. But all showed widely divergent individual performances."
(References rerererequoted below).
                   The results show unequivocally that most of us
*when ABXing do not hear GROSS differences such as 1,75 db. volume
difference and anything less than 2% distortion.
                  Your objection that the differences between
individual performances are not statistically convincing should be
addressed to the tests organisers and reporters: Clarke, Masters,
Greenhill,- "objectivists" to a man.  If their test methods were
flawed then so were their conclusions: namely "they all sound the
same". They were wasting their time. Have you got anything better to
quote or propose? Something with *unanimous* agreement between
panelists.
                  Your objection that Krueger and Clarke once reported
amplifier differences would be valid if anyone took Krueger and his
collaborator Clarke for a *broadly based* audiophile panel. Come on,
you must be joking. Who ever denied that highly trained ABX test
developers would hear differences, EVEN when ABXing between
electronically different amps?
                   Bias exists. In this market-oriented,
advertising-deafened society it is worse for some than for the others.
My own bias is against high-price: brand-names, bells and whistles,
imposing and pretty facades. I buy my watch at RS and my T shirts at
Penney's.
                   To prevent bias do whatever suits you: cover
brand-names, do a long or a short listen, whatever. I prefer my
*simultaneous* blind right-left comparison method ( see below) because
repeats of successive A then B then X snippets scramble my brain but I
claim no "validity" for my results over any one else's. It seems to me
though that ABX for use at home complicates things unduly for no
proven benefit.
                  When you look at what is playing you're biased and
may hear differences where there are none. When you ABX you may not
hear differences that you would hear otherwise. Take your pick.
                  It is a delusion peculiar to audio that a "test"
exists valid for Tom who never listens to unamplified instruments and
Harry who listens to nothing else. I'm not interested in knowing that
Tom hears no differences when solo violins are playing and he's not
interested in my opinion about electric guitar reproductions. Which is
as it should be.
Ludovic Mirabel

I said:
 > >       And how are they to "show us" that to your satisfaction?
You
> >don't say but let me venture a guess: would it not be by the ABX/DBT
> >method.
Nousaine said:  
> That would be a perfectly acceptable method. BUT, I'll accept ANY method that
> simply compensates for known human listening bias. You pick. 
> 
         Gladly. The only bias- controlling method that suits me sends
your camp into what the Victorians used to call conniption fits. I
like one component on the left, the other one on the right with random
changes from side to side while I'm blinded. Random changes eliminate
room, speaker etc. differences. Unless I get a 100% score I'm out.
          Why? Because I am no good at storing dificult acoustic
information like that: First store A, then B, then listen to X and
compare with A and B.
After 2 such trials my brain reels.
           And you know what: same goes for most people. I rerequote
the evidence below. Perhaps this time you will comment on it. You have
not done it in the past.In 1983 Greenhill cable test 80% of the
"expert audiophile" panelists couldn't tell 1,75 db. volume difference
between thin and thick cable.
You said: "....> What's funny is that Ludovic won't see that the
reason for this is because there were no differences." (between
cables, amps, preamps, cdplayers and dacs)
You said something similar before. So I quoted this to you on Nov 20,
'02 in the "Here it is - a SOTA recommendation" thread:
"In '89 a rather elaborate listening test for audibility of distortion
was  performed     .(Masters and Clark, St. Review, Jan. '89).
Various types of distortion with different signals were tested.. There
were  15 TRAINED listeners -? Gender?. At  high  distortion level
(2db), playing "natural music", the correct "hits " varied from 1/5 to
5/5.  The "average" was 61% (barely above the minimum statistically
significant level of 60%)
Similar discrepancies were observed in phase shift recognition.: from
zero score to the perfect 5/5.  The subjects underwent further
education for repeat testing but when their scores added up still fell
below "average" when guessing phase shift in "natural music" signals
with scores varying between zero and 5/5
Authors' conclusion:  Distortion has to be very gross and the signal
very simple for it to be  noticed by the "average""
NO COMMENT WAS MADE AND NO ATTENTION PAID TO THOSE WHO KEPT SCORING
BETTER OR WORSE THAN THE AVERAGE. It would appear that those
performances were fairly consistent but the authors did not think
this deserved any attention".
If 1,75 db volume and 1, 99% distortion difference is not different
enough for you Tom, please do tell what is.