Does anyone REALLY believe in 'cable sound'?
Stewart Pinkerton
whatever reasons, let's try another tack.
Is there *anyone* out there who really *does* believe that one cable
sounds better than another?
The prize pool is somewhere around the $2500 mark, and all you have to
do is demonstrate that you can hear the difference between any two
cables which meet a very simple level-matching criterion.
The basic rules are as follows:
Cables must be level-matched from 20Hz to 20kHz, within +/- 0.1dB at
the speaker terminals. Any conductor and insulation materials are
allowed, and any cable construction is allowed.
The test protocol will be double-blind ABX, implemented in whatever
way is both practical for the circumstances, and acceptable to all
interested parties. The test will comprise 20 trials, the 'pass'
criterion being correct identification in 16 out of the 20 trials.
The rest of the system (source, amplifier, speakers and listening
room) is at the choice of the test subject, as is the selection of
music, and the duration of the individual trials.
****************
We have all seen the numerous claims made for 'high end' cables, and
many posters claim to hear 'huge' differences, so there should be no
problem in beating this straightforward challenge in the poster's own
'reference' system, with his own choice of music, right?
Who will be the first to put his ears where his mouth is, and claim
the prize?
--
Stewart Pinkerton | Music is Art - Audio is Engineering
Paul Guy
.....[stuff deleted].......
>The basic rules are as follows:
>
>Cables must be level-matched from 20Hz to 20kHz, within +/- 0.1dB at
>the speaker terminals. Any conductor and insulation materials are
>allowed, and any cable construction is allowed.
>
.......[stuff dleted]......
I am assuming that "level matching" here means that the levels
refer to the input and output sides of the cable, and not the
comparative levels of the the outputs of the left and right channels.
If that is so, then any cable that influences frequency response
over the audio range would be compensated for (which is a worthy
assumption). This would be a very tight test, which would eliminate
the effects of cable inductance and resistance, especially at 20 kHz.
Thin wire cables (#22AWG and smaller) were the only cables that I
was able to reliably identify in listening tests. They would probably
be difficult, if not impossible, to discern, once the levels and
frequency responses were compensated for.
I could can think of a cable that might qualify for the prize
money, but it would be a perversion. If, in addition to the main pair
of wires, you ran multiple pairs of wires in the jacket of the cable
in a "bifilar" fashion, then hooked them up in series, and stuck a
shottky diode across the output of this parasitic connection, you
would probably introduce some audible distortion. The pairs would need
to be spaced relatively far apart in order to get enough inductance
for the whole thing to act like a step-up transformer. Since most of
the bad behaviour would only occur at high frequencies, it would be
very difficult to hear the distortion. If I had several hundred 2-hole
ferrite cores that I could string the wire pairs through, it would
give me enough transformer action at the lower frequencies to be more
noticeable! Even the saturation effects of the ferrite cores might
cause enough bad behaviour to be audible! Maybe, it would be wacky
enough for me to market this kind of cable to the audio fringes!
In your contest, how would you avoid nut cases like me who would
try to dream up a cable that should sound WORSE (well, to some)?
-Paul
-----------------------------------------------------------
*** Steppin' on your toes ***
Paul J Guy work:902-742-0770
gu...@burridge.nscc.ns.ca
PO Box 900, Yarmouth, Nova Scotia, B5A 4A9, Canada
Gordon Airport
can't deny the experience of the many of people that DO hear
differences between cables; they're in the best position to
report on this. I believe the point is that it seems largely
(if not entirely) the result of knowing that the cable is
there by other means. I suggest the real question, and what
the proposed test gets at, is "Can anyone identify a cable
by its sound only?"
Obviously it wouldn't be enlightening to compare a system's
sound with and without the cable in place, so you need a
reference cable to swap in when the cable in question isn't
connected. The result of the proposed test would indicate,
with whatever level of certainty, that someone can or cannot
distinguish between the cable and the reference without
obvious means other than system sound.
Whether a negative result would be disheartening to someone
who values the cable more than the reference is a question
of how much
knowing-the-cable-is-there-by-means-other-than-its-sound
matters to the person, which seems pretty much a matter of
taste.
Gary Eickmeier
>
> I think you might want to shift your emphasis a little. You
> can't deny the experience of the many of people that DO hear
> differences between cables; they're in the best position to
> report on this. I believe the point is that it seems largely
> (if not entirely) the result of knowing that the cable is
> there by other means. I suggest the real question, and what
> the proposed test gets at, is "Can anyone identify a cable
> by its sound only?"
The "by its sound only" part is what is meant by double blind test
conditions.
Gary Eickmeier
Gordon Airport
> conditions.
>
Right, I was concerned about the title "Does anyone REALLY
believe in 'cable sound'?"
There's no doubt that people believe in cable sound because
they hear REAL differences between cables. I was pointing
out that the contention is really over the reason for the
differences, not their existence. It starts these arguments
off on the wrong foot if it sounds like you're denying
people's experience.
Gary Eickmeier
> Right, I was concerned about the title "Does anyone REALLY
> believe in 'cable sound'?"
> There's no doubt that people believe in cable sound because
> they hear REAL differences between cables. I was pointing
> out that the contention is really over the reason for the
> differences, not their existence. It starts these arguments
> off on the wrong foot if it sounds like you're denying
> people's experience.
Perhaps you are relatively new to this debate. I will let Mr.
Pinkerton answer for himself, but in my opinion the answer is that we
ARE denying people's experience. That's the whole point. Now, if there
are any "people" who disagree with us, they can apparently try for a
$2500 prize if they can hear the difference between two cables.
Gary Eickmeier
Stewart Pinkerton
<dispose...@yahoo.com> wrote:
>> The "by its sound only" part is what is meant by double blind test
>> conditions.
>>
>
>Right, I was concerned about the title "Does anyone REALLY
>believe in 'cable sound'?"
>There's no doubt that people believe in cable sound because
>they hear REAL differences between cables.
Sorry, that is a mere assertion. Where is the proof that there are
such 'REAL' differences? This is the whole point of the challenge.
Please note that most of this money has been on the table for 3 or 4
*years*, yet none of the highly vocal 'subjectivists' who post to the
audio newsgroups has even *attempted* to collect it. Does this tell
you something about how 'real' are those differences?
>I was pointing
>out that the contention is really over the reason for the
>differences, not their existence.
Wrong.
> It starts these arguments
>off on the wrong foot if it sounds like you're denying
>people's experience.
I am denying that it is based in physical reality. To suggest that
'cable sound' even *exists* is an extraordinary claim, yet we see
absolutely *zero* evidence in support of that claim, just wild
speculation and pseudo-scientific technobabble from the so-called
'high end' cable makers.
Stewart Pinkerton
<dispose...@yahoo.com> wrote:
>I think you might want to shift your emphasis a little. You
>can't deny the experience of the many of people that DO hear
>differences between cables; they're in the best position to
>report on this.
No, that is the whole *point* of the challenge! I and many others
assert that those people are not 'hearing' anything which exists in
the real physical world - it is a mere illusion.
> I believe the point is that it seems largely
>(if not entirely) the result of knowing that the cable is
>there by other means. I suggest the real question, and what
>the proposed test gets at, is "Can anyone identify a cable
>by its sound only?"
>Obviously it wouldn't be enlightening to compare a system's
>sound with and without the cable in place, so you need a
>reference cable to swap in when the cable in question isn't
>connected. The result of the proposed test would indicate,
>with whatever level of certainty, that someone can or cannot
>distinguish between the cable and the reference without
>obvious means other than system sound.
That is *exactly* what is proposed. Do you not understand the nature
of an ABX comparison?
Stewart Pinkerton
>pat...@dircon.co.uk (Stewart Pinkerton) wrote:
>
>.....[stuff deleted].......
>>The basic rules are as follows:
>>
>>Cables must be level-matched from 20Hz to 20kHz, within +/- 0.1dB at
>>the speaker terminals. Any conductor and insulation materials are
>>allowed, and any cable construction is allowed.
>>
>.......[stuff dleted]......
>
> I am assuming that "level matching" here means that the levels
>refer to the input and output sides of the cable, and not the
>comparative levels of the the outputs of the left and right channels.
That is correct. Levels are matched for each channel and for each
cable, at the speaker terminals. This is usually a formality for
cables.
> If that is so, then any cable that influences frequency response
>over the audio range would be compensated for (which is a worthy
>assumption). This would be a very tight test, which would eliminate
>the effects of cable inductance and resistance, especially at 20 kHz.
Not really - do the math. Only when you talk about very extreme cases
such as electrostatic speakers dipping to one or two ohms at 20kHz,
and exteremes of construction such as Naim NACA5 and Alpha-Core MI, do
you see noticeable differences related to cable inductance.
> Thin wire cables (#22AWG and smaller) were the only cables that I
>was able to reliably identify in listening tests. They would probably
>be difficult, if not impossible, to discern, once the levels and
>frequency responses were compensated for.
Quite so. No maker of 'high end' cables has ever related the 'sound
quality' of his cables to mere resistance or inductance!
> I could can think of a cable that might qualify for the prize
>money, but it would be a perversion. If, in addition to the main pair
>of wires, you ran multiple pairs of wires in the jacket of the cable
>in a "bifilar" fashion, then hooked them up in series, and stuck a
>shottky diode across the output of this parasitic connection, you
>would probably introduce some audible distortion. The pairs would need
>to be spaced relatively far apart in order to get enough inductance
>for the whole thing to act like a step-up transformer. Since most of
>the bad behaviour would only occur at high frequencies, it would be
>very difficult to hear the distortion. If I had several hundred 2-hole
>ferrite cores that I could string the wire pairs through, it would
>give me enough transformer action at the lower frequencies to be more
>noticeable! Even the saturation effects of the ferrite cores might
>cause enough bad behaviour to be audible! Maybe, it would be wacky
>enough for me to market this kind of cable to the audio fringes!
> In your contest, how would you avoid nut cases like me who would
>try to dream up a cable that should sound WORSE (well, to some)?
That's a fair point. Perhaps the challenge should relate only to
cables which are commercially available, since the 'cheapies' are
usually referenced to Radio Shack 'Gold' interconnects and 'zipcord'
speaker cables. OTOH, you do admit that this would be a perversion, a
'Formula 1' evasion of the rules in order to win the prize, but
completely unrelated to the purpose of the challenge.
A simpler method would be to exclude any cable containing nonlinear
devices (such as your diodes), since they clearly have no place in
*any* high fidelity sound system!
Howard Ferstler
Gary. Remember that in a previous post I indicated that I
would be more than willing to chip in a dollar of my own.
So, the pot is really at $2,501. That buck shows just how
serious I am about this situation, because to me a buck is a
big deal.
Howard Ferstler
Malcolm McAvoy
matching the cables from 20hz to 20khz.In real life situations I dont
believe this is a practical test. The only matching that should be necessary
would be db level because this is what goes on in real life situations.
Personally I dont get a new cable go home and match it to my old cable.
Cables will sound different...unless you match them...then of course they
are going to sound the same because at that point they will esentially be
the same cable, if not physically. So the argument should be that if cables
are switched in a normal system with volume levels matched will there be a
difference? yes.
Malcolm McAvoy, Why argue? just enjoy the music.
"Stewart Pinkerton" <pat...@dircon.co.uk> wrote in message
news:aou0c5$acp$1...@bourbaki.localdomain...
Mkuller
>Is there *anyone* out there who really *does* believe that one cable
>sounds better than another?
>
types of cables they provide it is apparent a lot of audiophiles (probably
*most* audiophiles other than the rigid objectivists here) believe one cable
can sound better than another. Personally, I have heard cable differences and
have selected my cables on the basis of their audible superiority.
What I don't believe in is the use of blind A/B tests to *prove* subtle audible
differences exist between audio components.
Regards,
Mike
Steve G
(...)
>
> The test protocol will be double-blind ABX, implemented in whatever
> way is both practical for the circumstances, and acceptable to all
> interested parties.
Although this is reasonable, the incumbent might be better going for a
2-interval forced-choice design, since there's some evidence that this
leads to higher estimated sensitivity cf. ABX (e.g., [1], [2]). 2IFC
is also a simpler procedure for inexperienced listeners, assuming the
task is easily conceptualised as 2IFC.
> The test will comprise 20 trials, the 'pass'
> criterion being correct identification in 16 out of the 20 trials.
20 trials provides inadequate power for testing small - but better
than chance - 'genuine' performance levels. The hapless listener would
be better to push for >100 trials, if they could stand this.
The listener should also have trialwise accuracy feedback.
(...)
>
> Who will be the first to put his ears where his mouth is, and claim
> the prize?
I predict no-one with a vested interest in (selling) cables will,
ever. I would happily take part - but I doubt cables 'sound'
different, making the social context of the experiment a whole lot
less interesting, eh?
Steve.
Harris (1953). J Acoust Soc Am 24, 417
Hautus & Meng (2002). Percept Psychophys 64, 89
Uptown Audio
electrically, then what would you actually be testing, assuming again that
the differences heard would be those differences that you have decided to
"match"? That sounds sort of like me saying that you could not tell the
difference in weight between two bags of sand as long as I could weigh and
equalize the bags. Who cares?
- Bill
www.uptownaudio.com
(540) 343-1250
"Stewart Pinkerton" <pat...@dircon.co.uk> wrote in message
news:aou0c5$acp$1...@bourbaki.localdomain...
Richard D Pierce
Paul Guy <pg...@klis.com> wrote:
>pat...@dircon.co.uk (Stewart Pinkerton) wrote:
>
>.....[stuff deleted].......
>>The basic rules are as follows:
>>
>>Cables must be level-matched from 20Hz to 20kHz, within +/- 0.1dB at
>>the speaker terminals. Any conductor and insulation materials are
>>allowed, and any cable construction is allowed.
>>
>.......[stuff dleted]......
>
> I am assuming that "level matching" here means that the levels
>refer to the input and output sides of the cable, and not the
>comparative levels of the the outputs of the left and right channels.
Neither. What must be matched are the levels of the two cables
being compared.
> In your contest, how would you avoid nut cases like me who would
>try to dream up a cable that should sound WORSE (well, to some)?
In some audiophiles "dreams," there would be manufacturers who
PURPOSELY dream up cables that sound worse, i.e., deliberately
alter the sound.
For the rest, the nightmare is that some manufacturers REALLY do
this.
--
| Dick Pierce |
| Professional Audio Development |
| 1-781/826-4953 Voice and FAX |
| DPi...@world.std.com |
Arny Krueger
news:aoust7$smb$1...@bourbaki.localdomain
> I think you might want to shift your emphasis a little. You
> can't deny the experience of the many of people that DO hear
> differences between cables; they're in the best position to
> report on this.
Actually, these reports are generally pretty easy to assign an
appropriate cause to. It is extremely well known that most listeners
will perceive an audible difference if they believe that there is a
good reason to believe that there is an audible difference, even if
there is in fact no audible difference.
> I believe the point is that it seems largely
> (if not entirely) the result of knowing that the cable is
> there by other means.
Right. And this means that unassisted listeners are NOT in the best
position to report on audible differences due to cables.
> I suggest the real question, and what
> the proposed test gets at, is "Can anyone identify a cable
> by its sound only?"
This seems like the most important thing, right?
> Obviously it wouldn't be enlightening to compare a system's
> sound with and without the cable in place, so you need a
> reference cable to swap in when the cable in question isn't
> connected.
There may be another choice, which is to position the power amp(s)
very close to the speakers (say, 2 monoblocks) and compare cables of
a normal length with cables of virtually zero length.
>The result of the proposed test would indicate,
> with whatever level of certainty, that someone can or cannot
> distinguish between the cable and the reference without
> obvious means other than system sound.
This seems like the most important thing, right?
> Whether a negative result would be disheartening to someone
> who values the cable more than the reference is a question
> of how much
> knowing-the-cable-is-there-by-means-other-than-its-sound
> matters to the person, which seems pretty much a matter of
> taste.
I wonder why knowing that the cable is there (without any audible
improvement in sound quality) would be assigned any value at all.
Dr Bukkake
>The basic rules are as follows:
> The test will comprise 20 trials, the 'pass' criterion being correct
identification in 16 out of the 20 trials.
I've completely missed this thread, so I can only imagine the diatribes that
have been flying around. What I find most interesting in this challenge is not
the particulars of test (level matching, et al) but the selection of the number
16 as a "passing" grade. I was wondering if you had examined the significance
of this number, or if you just thought it seemed impressively high enough a
success rate to justify awarding the prize. Certainly, you and others here
possess significant technical acumen, but did you, in fact, determine the
probability of someone simply guessing 16 out of 20? It would seem a good idea,
at least if my money were at stake.
Mind you, this may have been addressed already (in which case, a thousand
pardons) but I decided to see for myself just how challenging a proposition it
is that you offer. Employing the rather cumbersome binomial probability
theorum-
p(x)=n!/((n-x)!x!)*p^x*q^n-x, I've come up with the following-
the probability of randomly selecting 16 or more correct answers out of 20 is
0.0059. This represents an aproximate odds against of 168 to 1. While this
makes it a reasonably difficult test to pass, I wouldn't plan on opening a
casino. If enough people took your test (including the deaf), you would have
to pay off much sooner than you may have expected. Of course, the "winner"
will have proven nothing, as he will have accomplished no more difficult a feat
than FAILING to properly identify the cable 16 or more times out of 20. In
other words, your test is hard, but it's not Lotto, and it doesn't even cost a
buck.
I would hate to see someone I so fully agree with (at least in the matter of
cables) lose a wager when the loss would not, in fact, be dictated by any
fallacy in his basic premise regarding the ability of a listener to identify
cables. I positively couldn't stand the response from those who would suggest
that the success of Murray Schwartz of Bethesda, MD- he of golden ear and $2500
check- stands as proof positive for all time that those uppity engineers don't
know what the hell they're talking about.
I offer the above not because I think you are wrong, but because I think you
are correct. For purely arguments sake, if you are correct then the applicant
will have no choice but to offer a random opinion, and, as such, the control of
the random aspect of the experiment should be your primary concern. Again, 168
to 1 is no easy task, but people routinely play (and win) the "number" at odds
of 1000:1. A fifteen dollar entrance fee would, theoretically make your
scenario fair and equitable. Otherwise, what the hell, put me down for a
listen! I'll even bring along my 5 year old, who will have as good a chance of
winning as I do.
More numbers? Just shy of 50% of those tested would be expected to select 9,
10, or 11 cables correctly. Almost 3/4 would choose between 8 and 12 properly,
and over 88% would rightly identify between 7 and 13 cables. Only 17.6% would
get 10 right and 10 wrong, which will seem incomprehensible to those not
familiar with probability.
Good "luck" with your test, because that's exactly what you may be needing.
The applicants certainly will.
Music is Art
Audio is Engineering
Math is God
Richard Wall
Your test seems a little strange, surely the main differences between the
"sound" of cables are how they affect the frequency response at the speakers
? There is no such thing as a speaker that is truly un coloured from 20Hz
to 20Khz is there ? What about room effects on the frequency response at
the listening point ? I thought part of the reason that people bought
different cables was to match the sound of the system to the sound they
wanted in their room.
If you want to set up a test in the south east of the UK I am sure that the
New Ash Green Hi-Fi club would be delighted to try the test what ever the
restrictions you impose. Perhaps we might even persuade Jimmy Hughes (a
member) to attend.
Let me know
Regards Richard
"Stewart Pinkerton" <pat...@dircon.co.uk> wrote in message
news:aou0c5$acp$1...@bourbaki.localdomain...
jjn...@sonic.net
> On Mon, 21 Oct 2002 00:56:50 GMT, Gordon Airport
> <dispose...@yahoo.com> wrote:
>>> The "by its sound only" part is what is meant by double blind test
>>> conditions.
>>>
>>
>>Right, I was concerned about the title "Does anyone REALLY
>>believe in 'cable sound'?"
>>There's no doubt that people believe in cable sound because
>>they hear REAL differences between cables.
> Sorry, that is a mere assertion. Where is the proof that there are
> such 'REAL' differences? This is the whole point of the challenge.
> Please note that most of this money has been on the table for 3 or 4
> *years*, yet none of the highly vocal 'subjectivists' who post to the
> audio newsgroups has even *attempted* to collect it. Does this tell
> you something about how 'real' are those differences?
They are real to the person experiencing them. On the other hand,
that doesn't necessarily mean that they are audible acoustic events.
> I am denying that it is based in physical reality. To suggest that
> 'cable sound' even *exists* is an extraordinary claim, yet we see
> absolutely *zero* evidence in support of that claim, just wild
> speculation and pseudo-scientific technobabble from the so-called
> 'high end' cable makers.
Certainly subjective effects that have no basis as an audible acoustic
event have chemical processes in the brain that are a physical reality.
Just a little clarification.
Gordon Airport
> there are such 'REAL' differences?
I'm backing up my assertion with the copious number of
anecdotal (hold on) reports of people hearing 'jaw dropping'
differences between cables, describing their sound, how they
match with their system, etc. The usual subjectivist stuff.
We cannot be uncharitable and assert that they are lying
when they report these perceptions, just so they can then
spend ridiculous sums on these cables. (further, stronger
considerations below)
The point I'm making (and I think we agree on this) is that
in all the anecdotal cases (that I've heard of anyway), the
person already knew that one cable or another was in the
system, and that this is the control that causes failure in
DBT's.
I don't believe competently constructed cables sound
different than each other either - with the caveat - when you
don't know what you're listening to.
> This is the whole point of the challenge. Please note
> that most of this money has been on the table for 3 or 4
> *years*, yet none of the highly vocal 'subjectivists' who
> post to the audio newsgroups has
> even *attempted* to collect it. Does this tell
> you something about how 'real' are those differences?
I would love to see someone take the test (which I fully
expect they would fail), if only to move the debate on to
how important it should be to your enjoyment of music that
you know what components are in your system (which isn't
grounds for a real debate at all.)
>> I was pointing out that the contention is really over
>> the reason for the differences, not their existence.
>
> Wrong.
>
That IS what I was pointing out ;-)
> I am denying that it is based in physical reality.
To the extent that psychology has no basis in physical
reality? Still, whatever your views on that, AFAICT it is a
psychological effect: the interaction of knowledge and
perception. Remove the knowledge and you change the
perception. I'm just clarifying:
WITH AND WITHOUT the knowledge of the components in play, we
have to believe people's reports of their perceptions. If we
dismiss the subjectivist's anecdotal reports of their
perceptions, then when someone reports perceiving no
difference in a DBT nothing would follow from it.
> To suggest that 'cable sound' even *exists* is an
> extraordinary claim, yet we see absolutely *zero*
> evidence in support of that claim, just wild speculation
> and pseudo-scientific technobabble from the so-called
> 'high end' cable makers.
People do *perceive* (hear) differences, but for the cable
manufactures and marketers to claim that this is a result of
the properties of the cable's properties alone (with
absence of psychological considerations) is disingenuous at
the very least.
Alan Murphy
> Sorry, that is a mere assertion. Where is the proof that there are
> such 'REAL' differences? This is the whole point of the challenge.
> Please note that most of this money has been on the table for 3 or 4
> *years*, yet none of the highly vocal 'subjectivists' who post to the
> audio newsgroups has even *attempted* to collect it. Does this tell
> you something about how 'real' are those differences?
>
Perhaps $2500 isn't enough to tempt anybody.
I beleive I can hear distinct differences in cables and if you up the
ante to say $100,000 I'd be prepared to pursue your challenge.
Alan :-)
Stewart Pinkerton
wrote:
>It seems to me that people that are hearing differences in cable arent
>matching the cables from 20hz to 20khz.In real life situations I dont
>believe this is a practical test. The only matching that should be necessary
>would be db level because this is what goes on in real life situations.
In almost all cases, there is no need to 'match' anything, the levels
will naturally match to the required accuracy. This is a red herring.
>Personally I dont get a new cable go home and match it to my old cable.
>Cables will sound different...unless you match them...then of course they
>are going to sound the same because at that point they will esentially be
>the same cable, if not physically. So the argument should be that if cables
>are switched in a normal system with volume levels matched will there be a
>difference? yes.
That's exactly what will happen, except in the case of a very few
extreme constructions (not the most expensive, BTW). So far, no one
has ever shown that there is an *audible* difference.
Stewart Pinkerton
wrote:
>"Stewart Pinkerton" <pat...@dircon.co.uk> wrote in message
>news:ap0ad2$fst$1...@bourbaki.localdomain...
>> Sorry, that is a mere assertion. Where is the proof that there are
>> such 'REAL' differences? This is the whole point of the challenge.
>> Please note that most of this money has been on the table for 3 or 4
>> *years*, yet none of the highly vocal 'subjectivists' who post to the
>> audio newsgroups has even *attempted* to collect it. Does this tell
>> you something about how 'real' are those differences?
>Perhaps $2500 isn't enough to tempt anybody.
>
>I beleive I can hear distinct differences in cables and if you up the
>ante to say $100,000 I'd be prepared to pursue your challenge.
Excellent! I'll be happy to accommodate you, at even odds.
--
Stewart Pinkerton
<Ric...@richardwall.freeserve.co.uk> wrote:
>Dear Stewart
>Your test seems a little strange, surely the main differences between the
>"sound" of cables are how they affect the frequency response at the speakers
>?
That's not what the cable makers claim.
> There is no such thing as a speaker that is truly un coloured from 20Hz
>to 20Khz is there ? What about room effects on the frequency response at
>the listening point ? I thought part of the reason that people bought
>different cables was to match the sound of the system to the sound they
>wanted in their room.
And how do you suppose that will happen? Have you ever measured just
how small are such effects, compared to the several dB differences
typical among speakers?
>If you want to set up a test in the south east of the UK I am sure that the
>New Ash Green Hi-Fi club would be delighted to try the test what ever the
>restrictions you impose. Perhaps we might even persuade Jimmy Hughes (a
>member) to attend.
You've seen the requirements, and I would *love* to see the famously
'tweaky' Jimmy Hughes put his ears on the line. Care to place a
sidebet that he won't go within a mile of such a test?
Stewart Pinkerton
wrote:
>If what you appear to be saying, that different cables would be "matched"
>electrically, then what would you actually be testing, assuming again that
>the differences heard would be those differences that you have decided to
>"match"?
When did you *ever* see 'high end' cables claiming that their sound
quality had *anything* to do with basic LCR parameters. In the vast
majority of systems, 12AWG zipcord will match to any of those $100 a
foot 'designer' cables with no problem. Don't attempt to build a
strawman out of the level matching requirement.
Richard D Pierce
Uptown Audio <uptow...@rev.net> wrote:
>If what you appear to be saying, that different cables would be "matched"
>electrically, then what would you actually be testing, assuming again that
>the differences heard would be those differences that you have decided to
>"match"? That sounds sort of like me saying that you could not tell the
>difference in weight between two bags of sand as long as I could weigh and
>equalize the bags. Who cares?
But the assertion that some cable proponents is that the
differences in sound are NOT explained by differences in level
and frequency response. In other words, they claim that the RLC
parameters of the cable are NOT sufficient to account for these
differences. The idea to equalizing level and frequency response
differences was to remove those differences due RLC effects,
leaving only the differences due to these "other" mechanisms.
TonyP
> >Stewart Pinkerton wrote:>
> >Is there *anyone* out there who really *does* believe that one cable
> >sounds better than another?
> >
> From looking at the number of successful cable manufacturers, and the various
> types of cables they provide it is apparent a lot of audiophiles (probably
> *most* audiophiles other than the rigid objectivists here) believe one cable
> can sound better than another. Personally, I have heard cable differences and
> have selected my cables on the basis of their audible superiority.
Although I would agree on the success of a lot of "hi end" cable people (how much
CAN wire really cost??), I know that I have heard differences in cables. Some were
very subtle. Some, I couldn't really tell the difference.
But I do remember comparing at the time, MIT shotgun everything to my Monster
Powerline 2 and Interlink Reference A. I had nothing to prove and wanted to know if
there was a difference in sound, since the price difference was vast. Well, my
buddy who bought the MIT stuff left my home a little sadden by the experience. At
least in my system (at the time, Acoustat 1+1's, Counterpoint SA 3 pre amp, SA 2
head amp, VPI/MMT/Koetsu analog), the MIT didn't "sound" as good as the Monster
stuff I had.
We listened to a variety of music that we were both familiar with at different
sound levels.
No ABX matching. Just listening to music.
horizon
news:ap20v...@enews3.newsguy.com...
> Although I would agree on the success of a lot of "hi end" cable people
(how much
> CAN wire really cost??), I know that I have heard differences in cables.
Some were
> very subtle. Some, I couldn't really tell the difference.
My experience as well. I have definitely heard differences in cables, but
I'm not at all sure that the expensive cables sounded "better" than
inexpensive cables. In fact, all of my current interconnects are of the
quite inexpensive variety. But I have heard serious and substantial
differences (many of which I did not like in my system, and hence returned
to the inexpensive cables) between interconnects as varied as Kimber PBJ,
Transparent Music Plus, MIT Terminator 2, Canare L-46, Radio Shack Gold,
etc. I would assume that this result could be easily replicated, although
it might take a properly designed study that would involve blind listening
of the same music over an extended duration -- since I find that I begin to
hear things 40-60 minutes into my listening, rather than within the first
few minutes. Of course, at some point in this approach the fatigue factor
will set in, but I can't see any other way to do it, at least based on my
experience.
> But I do remember comparing at the time, MIT shotgun everything to my
Monster
> Powerline 2 and Interlink Reference A. I had nothing to prove and wanted
to know if
> there was a difference in sound, since the price difference was vast.
Well, my
> buddy who bought the MIT stuff left my home a little sadden by the
experience. At
> least in my system (at the time, Acoustat 1+1's, Counterpoint SA 3 pre
amp, SA 2
> head amp, VPI/MMT/Koetsu analog), the MIT didn't "sound" as good as the
Monster
> stuff I had.
I also had Powerline 2 at one point in my system, then with Apogee Calipers.
I quickly heard that this definitely "dark & fat sounding" cable was not a
proper match for a rich-sounding speaker like Apogee full range ribbons, and
went back to the Apature zip cord that Innovative Audio gave us when we
originally bought our system. The effect was not subtle in any way, shape
or form.
Matt C
Arny Krueger
> On 21 Oct 2002 16:39:13 GMT, "Uptown Audio" <uptow...@rev.net>
> wrote:
>
>> If what you appear to be saying, that different cables would be
>> "matched" electrically, then what would you actually be testing,
>> assuming again that the differences heard would be those
>> differences that you have decided to "match"?
> When did you *ever* see 'high end' cables claiming that their
> sound quality had *anything* to do with basic LCR parameters.
How about denials of that obvious fact or suggestions that other
factors are more important?
http://www.stereotimes.com/cables031102.shtm
"I have not taken measurements, but the dielectric constant should be
close to 1 [the ideal]. Similarly, I have not measured actual LRC
(inductance, resistance, capacitance) values, though, given the
geometry of the ribbons, those can easily be derived and are very
low. Still, they have nothing to do with the sound of a cable.
[Bravo! If only the pocket-protector types would sit down and
listen!]"
http://www.tnt-audio.com/clinica/spkcbl_e.html
"I would propose, that the usual RLC parameters (as with Resistance
[R] , Inductance [L] and Capacitance [C] ) should be viewed as first
order effects, though not in all situations each parameter carries
the same weighting."
"As a result the Capacitance (C) is a prevalent characteristic with
Resistance (R) and Inductance (L) relegated usually BELOW most second
Order Effects in their magnitude of sonic impact."
http://www.stereophile.com/showarchives.cgi?84
"In my view, RLC advocates take too simplistic a view of audio-signal
transmission-there are other factors that affect a cable's electrical
performance."
http://www.blackdahlia.com/tipindex/tip_25/tip_25.html
"The decisive mistake engineers make is to ascribe ALL sonic
differences to a cable's electrical parameters."
"Put a gun to their head, and they will maintain till their dying
breath that the RLC paradigm is the truth, the whole truth, and
nothing but the truth. In the past ten years, a handful of
investigators have shown that the RLC paradigm just doesn't go far
enough, and that there are other factors that do indeed affect signal
transmission."
Stewart Pinkerton
wrote:
>I have definitely heard differences in cables, but
>I'm not at all sure that the expensive cables sounded "better" than
>inexpensive cables. In fact, all of my current interconnects are of the
>quite inexpensive variety. But I have heard serious and substantial
>differences (many of which I did not like in my system, and hence returned
>to the inexpensive cables) between interconnects as varied as Kimber PBJ,
>Transparent Music Plus, MIT Terminator 2, Canare L-46, Radio Shack Gold,
>etc. I would assume that this result could be easily replicated,
Need we deconstruct 'assume' one more time? :-)
> although
>it might take a properly designed study that would involve blind listening
>of the same music over an extended duration -- since I find that I begin to
>hear things 40-60 minutes into my listening, rather than within the first
>few minutes. Of course, at some point in this approach the fatigue factor
>will set in, but I can't see any other way to do it, at least based on my
>experience.
The problem is that *everyone* has the same story of 'substantial'
differences, but *no one* has yet replicated this when they don't
*know* which cable is connected. Been there, done that, very
irritating - but educational!
Stewart Pinkerton
That may indeed be true, since MIT shotgun is notoriously 'dull',
perhaps due to extremely high inductance.
>We listened to a variety of music that we were both familiar with at different
>sound levels.
>No ABX matching. Just listening to music.
IOW, you do not *really* know if there was a difference in the
physical sound field.
Darryl Miyaguchi
wrote:
>pat...@dircon.co.uk (Stewart Pinkerton) wrote in message news:<aou0c5$acp$1...@bourbaki.localdomain>...
>(...)
>>
>> The test protocol will be double-blind ABX, implemented in whatever
>> way is both practical for the circumstances, and acceptable to all
>> interested parties.
>
>Although this is reasonable, the incumbent might be better going for a
>2-interval forced-choice design, since there's some evidence that this
>leads to higher estimated sensitivity cf. ABX (e.g., [1], [2]). 2IFC
>is also a simpler procedure for inexperienced listeners, assuming the
>task is easily conceptualised as 2IFC.
How many trials does one have to perform in a 2IFC to get the same
type 1 error as 16/20 correct in ABX? Raw test sensitivity should be
traded against the probability of tester fatigue. ABX is already
pretty inefficient and takes a lot of effort and trials. 20 trials at
full concentration is very tiring.
>> The test will comprise 20 trials, the 'pass'
>> criterion being correct identification in 16 out of the 20 trials.
>
>20 trials provides inadequate power for testing small - but better
>than chance - 'genuine' performance levels. The hapless listener would
>be better to push for >100 trials, if they could stand this.
100 trials is far too many to be practical, IMO. However, I can see
the value (ideally) in a large number of trials. The test offer as it
stands is too short to allow for a small type II error. So it would
not be correct to say, if one failed to achieve 16/20, that a "no
difference" result was proved to a high degree of confidence.
>The listener should also have trialwise accuracy feedback.
One needs to be very careful about allowing feedback during the test.
See previous threads concerning "Sequential ABX Tests." Basically,
allowing too much feedback can bias the result in the listener's
favor. The safest method is to not allow any feedback concerning the
number of correct trials until the very end.
Darryl Miyaguchi
WWCASE
>In some audiophiles "dreams," there would be manufacturers who
>PURPOSELY dream up cables that sound worse, i.e., deliberately
>alter the sound.
If I were in the audiophile wire business, I'd purposely design
something that sounded different (not worse), invest in packaging
and advertising and balance that against wholesale price and
volume. I'd believe that the combination of packaging,
marketing, "different" and price would sell enough product to make
a profit. My fundamental bet would be that I could convince enough
audiophiles that "Different + $1000 = Better"
>For the rest, the nightmare is that some manufacturers REALLY do
>this.
Seems like a victimless crime. Yeah, something's being
prostituted, but the customer is getting that audiophile rush on his
dollar.
B. Case
dangling entity
specialty cable is no different than conventional zipcord with a $5.00
passive network stuck on the end. What this test *is* attempting to
reveal, however, is the alleged "mystical" sound quality of a certain
cable which seemingly transcends the frequency domain and RLC modeling
altogether. If one takes the test and fails, you logically must "pick
your poison" among the above premises (there is no such thing as
"mystical" cable sound or cable sound is completely frequency
domain-based, not to mention imperfect as a transparent, straight link
medium, and utterly reproducible via more mundane means than high $
cable).
I see there are actually a number of posters in this topic who are in
support of the "cable sound" concept. Question is, are they up to the
challenge for a cash prize?
Loiskelly1
>What I find most interesting in this challenge is not the particulars of test
(level matching, et al) but the selection of the number 16 as a "passing"
grade. I was wondering if you had examined the
significance of this number, or if you just thought it seemed impressively high
enough a success rate to justify awarding the prize.>
Excellent point. I've read what seems to be a million posts on this thread,
and I always just assumed that 16 would be a really high number to "guess".
> Employing the rather cumbersome binomial probability
>theorum-
>p(x)=n!/((n-x)!x!)*p^x*q^n-x
You gots ta be kidding! Can anyone verify??? Elaborate???
>This represents an aproximate odds against of 168 to 1.
All of a sudden I feel that someone may collect on this, given the number of
folks in here.
>Of course, the "winner"
>will have proven nothing, as he will have accomplished no more difficult a
>feat
>than FAILING to properly identify the cable 16 or more times out of 20.
If I read this correctly, then answering correctly 4 or less times should
receive the payoff as well.
> Again, 168 to 1 is no easy task, but people routinely play (and win) the
"number" at
>odds
>of 1000:1.
Sounds as if I'll be getting much better odds on the cable challenge.
>Only 17.6% would
>get 10 right and 10 wrong, which will seem incomprehensible to those not
>familiar with probability.
This does sound amazing. Is this true?
harrison
> differences in sound are NOT explained by differences in level
> and frequency response. In other words, they claim that the RLC
> parameters of the cable are NOT sufficient to account for these
> differences. The idea to equalizing level and frequency response
> differences was to remove those differences due RLC effects,
> leaving only the differences due to these "other" mechanisms.
>
> --
> | Dick Pierce |
> | Professional Audio Development |
> | 1-781/826-4953 Voice and FAX |
> | DPi...@world.std.com |
>
There are electrical parameters that are difficult to express in RLC and
these could be the elusive differences. The first to come to mind is
dissipation factor of the dielectric, which is almost like a resistive lossy
effect, except that it has some time dependence.
My old experiments with capacitors, way back in the old days of Suffolk
Audio, showed some 'extreme' non-linearities that had time constants of
large fractions of a second. Once you have non-linear behavior, the usual
Maxwell equation solutions come apart pretty quickly. As a reference, we
measured dielectric dissipations with time constants on the order of 5
milliseconds. Mylar tended to have about a 10 bit effective accuracy and
the best, Teflon, was equivalent to about 14 bits.
Is an experiment measuring voltage on a capacitor as a function of injected
charge valid for cables? Probably not, as the connecting wires we used to
measure these effects were PVC, a fairly lossy material, and did not seem to
affect the measurements. The entire network needs to be analyzed and the
impact of dielectric non-linearities calculated. I did these calculations a
long time ago when trying to build my own crossover. I could not calculate
any meaningful effects, but used better capacitors as a prophylactic measure
as they were readily available and inexpensive.
Hope this does not muddy the waters.
Dave Harrison
Steve G
(...)
> > The test will comprise 20 trials, the 'pass' criterion being correct
> identification in 16 out of the 20 trials.
>
> 16 as a "passing" grade. I was wondering if you had examined the significance
> of this number, or if you just thought it seemed impressively high enough a
> success rate to justify awarding the prize.
> I've come up with the following-
> the probability of randomly selecting 16 or more correct answers out of 20 is
> 0.0059. This represents an aproximate odds against of 168 to 1. While this
> makes it a reasonably difficult test to pass, I wouldn't plan on opening a
> casino. If enough people took your test (including the deaf), you would have
> to pay off much sooner than you may have expected.
In fact, according to scientific convention this is a very
*conservative* acceptance criterion (p<.05 is the conventional
number)! It is also a very small number of trials, and the test has
relatively low power with this number of trials.
(...)
>
> I would hate to see someone I so fully agree with (at least in the matter of
> cables) lose a wager when the loss would not, in fact, be dictated by any
> fallacy in his basic premise regarding the ability of a listener to identify
> cables.
Indeed - if enough people took the test, eventually there'd likely be
a positive result even with no perceived differences. However, for the
'subjectivist', I think the 'social cost' of failing the test will not
be overcome. Prediction: No-one with a vested interest in the
existence of cable differences will take the test. Also, if an
individual could genuinely successfully discriminate, then this would
be a fully repeatable event. (Though whether the winner would allow
repeat testing is another matter entirely!)
(...)
> of 1000:1. A fifteen dollar entrance fee would, theoretically make your
> scenario fair and equitable. Otherwise, what the hell, put me down for a
> listen!
With no belief system behind you, I suspect the exercise becomes a lot
less interesting for most involved. Let's be honest - a lot of this
malarkey is designed to tempt a cable-activist onto the guillotine,
and get him to pull the lever himself!
Steve.
Richard D Pierce
harrison <harriso...@rcn.com> wrote:
>> But the assertion that some cable proponents is that the
>> differences in sound are NOT explained by differences in level
>> and frequency response. In other words, they claim that the RLC
>> parameters of the cable are NOT sufficient to account for these
>> differences. The idea to equalizing level and frequency response
>> differences was to remove those differences due RLC effects,
>> leaving only the differences due to these "other" mechanisms.
>
>these could be the elusive differences. The first to come to mind is
>dissipation factor of the dielectric, which is almost like a resistive lossy
>effect, except that it has some time dependence.
No doubt. What is the magnitude of these effects that we might
find in speaker cables is the relevant question.
>Is an experiment measuring voltage on a capacitor as a function of injected
>charge valid for cables? Probably not, as the connecting wires we used to
>measure these effects were PVC, a fairly lossy material, and did not seem to
>affect the measurements. The entire network needs to be analyzed and the
>impact of dielectric non-linearities calculated. I did these calculations a
>long time ago when trying to build my own crossover. I could not calculate
>any meaningful effects, but used better capacitors as a prophylactic measure
>as they were readily available and inexpensive.
That's all well and good, and that is PRECISELY the reason why
it's important to eliminate the simple RLC effects if one is
searching for some "deeper reality." We don't want the gross and
easily explanable effects completely swamping these others if we
are attempting to detect something that is on the edge of
detectability, do we?
That being said, no one, not a single person, has ever shown the
effects that you allude to are of ANY relevance, either through
objective measurements or reliable, repeatable listening
evaluation.
If these effects exist, and they are as important as some might
claim, why has not a single person stepped forward with the
proof of their existance that such slaims would suiggest is easy
to obtain?
The "prize" offered is important not so much for the cash value,
but specifies a set of conditions that MAXIMIZE the chances that
these non-RLC induced differences will be heard. The deck is, to
a great extent, stacked in THEIR favor, AND there's that gold
ring waiting as well.
Harry Lavo
news:ap45k...@enews3.newsguy.com...
why would anybody subject himself to a test he believes is flawed and
stacked against him, for any amount of money?
Steve G
>
> How many trials does one have to perform in a 2IFC to get the same
> type 1 error as 16/20 correct in ABX?
Assuming we're still happy with the - arguably slightly crude -
binomial test approach to significance, the answer is 16/20, of
course! 2IFC still leads to a correct or incorrect response on each
trial.
> Raw test sensitivity should be
> traded against the probability of tester fatigue. ABX is already
> pretty inefficient and takes a lot of effort and trials. 20 trials at
> full concentration is very tiring.
Well, in psychophysical studies, many hundreds of trials are the norm.
20 trials is simply inadequate for picking up small (but real)
discriminative ability.
(...)
>
> 100 trials is far too many to be practical, IMO. However, I can see
> the value (ideally) in a large number of trials.
If there exists small but genuine discriminative ability in a
listener, many trials has to be 'practical' - alas.
> The test offer as it
> stands is too short to allow for a small type II error. So it would
> not be correct to say, if one failed to achieve 16/20, that a "no
> difference" result was proved to a high degree of confidence.
Or you could simply say - 'the test lacks power'.
>
> >The listener should also have trialwise accuracy feedback.
>
> One needs to be very careful about allowing feedback during the test.
> See previous threads concerning "Sequential ABX Tests." Basically,
> allowing too much feedback can bias the result in the listener's
> favor. The safest method is to not allow any feedback concerning the
> number of correct trials until the very end.
Absolutely not!
Caveat no. 1 - I am unfamiliar with the sequential test procedures to
which you specifically refer (they are very unusual in behavioural
science). However, it appears that even in a sequential test, accuracy
feedback is independent of the inflation of p values. (The problem
appears basically the same as that of performing multiple significance
tests on the same data set.)
Anyway, with the more usual fixed no. of trials there is no reason to
not give feedback, except in very special cases. In such designs the
test isn't biased in the listeners favour. Feedback simply enables the
perceiver to learn to discriminate any differences which do exist. If
there are no audible differences, feedback ain't going to help the
listener to produce performance out of thin air as 'twere! Feedback
just gives the best chance of finding performance, where it exists.
Steve.
Stewart Pinkerton
>> Employing the rather cumbersome binomial probability
>>theorum-
>>p(x)=n!/((n-x)!x!)*p^x*q^n-x
>
>You gots ta be kidding! Can anyone verify??? Elaborate???
It's basic statistics, you'll find it in any stats primer.
> >This represents an aproximate odds against of 168 to 1.
>
>All of a sudden I feel that someone may collect on this, given the number of
>folks in here.
As has been pointed out to those attempting to 'cherry pick' the
Greenhill panel results, it certainly means that if 170 people try it,
it's even money that someone will succeed through random guessing.
I am happy to run with those odds.
>>Of course, the "winner"
>>will have proven nothing, as he will have accomplished no more difficult a
>>feat
>>than FAILING to properly identify the cable 16 or more times out of 20.
>
>If I read this correctly, then answering correctly 4 or less times should
>receive the payoff as well.
No, since this challenge is specifically about claims made by those
who support the notion of 'cable sound', who should therefore get it
right every time! :-)
>> Again, 168 to 1 is no easy task, but people routinely play (and win) the
>"number" at
>>odds
>>of 1000:1.
>
>Sounds as if I'll be getting much better odds on the cable challenge.
Yes indeed! And yet, there's *still* no one with the courage of their
'subjectivist' convictions..........
What does that tell you?
>>Only 17.6% would
>>get 10 right and 10 wrong, which will seem incomprehensible to those not
>>familiar with probability.
>
>This does sound amazing. Is this true?
Yes, it's true. Again, just your basic 'bell curve' distribution for
20 trials. Chop the curve into 20 (21?) slices, and you'll find that
the central 10/10 choice occupies 17.6% of the area under the curve.
Stewart Pinkerton
wrote:
>With no belief system behind you, I suspect the exercise becomes a lot
>less interesting for most involved. Let's be honest - a lot of this
>malarkey is designed to tempt a cable-activist onto the guillotine,
>and get him to pull the lever himself!
Aw gee, now you've done it! I was hoping no one would notice! :-)
Of course, you and I believe this to be true, because we do *not*
believe in 'cable sound'. Those who *claim* to believe in cable sound
do seem strangely reluctant to test their faith.........
Stewart Pinkerton
>"Steve G" <ne...@stevethepsycho.co.uk> wrote in message
>news:ap45k...@enews3.newsguy.com...
>> Indeed - if enough people took the test, eventually there'd likely be
>> a positive result even with no perceived differences. However, for the
>> 'subjectivist', I think the 'social cost' of failing the test will not
>> be overcome. Prediction: No-one with a vested interest in the
>> existence of cable differences will take the test. Also, if an
>> individual could genuinely successfully discriminate, then this would
>> be a fully repeatable event. (Though whether the winner would allow
>> repeat testing is another matter entirely!)
It did occur to me that it would be only fair to hand over the money
with a smile, and for the subject then to do it again, just to prove
to everyone's satisfaction that it wasn't a statistical fluke. After
all, since these differences are claimed to be 'obvious to any *true*
audiophile', there should be no problem, right?
>why would anybody subject himself to a test he believes is flawed and
>stacked against him, for any amount of money?
It has of course just been shown that the test is in *fact* stacked in
your favour, and yet we still see no sign of any takers. What happened
to all those 'obvious' night and day' differences that you guys keep
harping on about?
You get the choice of your own system, your own listening room, your
own music, your own length of test, an agreed panel of independent (or
at least 50/50 split!) observers, and a reasonable statistical chance
of just guessing it, plus you don't even have to put up any money on
your side. What *is* the problem here?
Stewart Pinkerton
>>Stewart Pinkerton wrote:>
>>Is there *anyone* out there who really *does* believe that one cable
>>sounds better than another?
>>
From looking at the number of successful cable manufacturers, and the various
>types of cables they provide it is apparent a lot of audiophiles (probably
>*most* audiophiles other than the rigid objectivists here) believe one cable
>can sound better than another. Personally, I have heard cable differences and
>have selected my cables on the basis of their audible superiority.
So you claim. As ever, and so typically of the 'subjectivists', we are
expected to take your word for this 'audible superiority', but you
carefully avoid providing *any* evidence that you really can tell the
difference if you don't *know* which cable is connected.
>What I don't believe in is the use of blind A/B tests to *prove* subtle audible
>differences exist between audio components.
Very convenient, Mike. I'll take that as a 'no'. Strange how the good
people at B&W, Revel, Mark Levinson et al, who *design* the 'high end'
gear that the audiophiles buy, seem to find DBTs extremely useful for
uncovering subtle - but *real* - sonic differences.
Alan Murphy
For the guys on RAHE $2500 just ain't worth getting out of bed for.
Up the ante.
Alan
Darryl Miyaguchi
wrote:
>Assuming we're still happy with the - arguably slightly crude -
>binomial test approach to significance, the answer is 16/20, of
>course! 2IFC still leads to a correct or incorrect response on each
>trial.
Ok.
>> Raw test sensitivity should be
>> traded against the probability of tester fatigue. ABX is already
>> pretty inefficient and takes a lot of effort and trials. 20 trials at
>> full concentration is very tiring.
>
>Well, in psychophysical studies, many hundreds of trials are the norm.
>20 trials is simply inadequate for picking up small (but real)
>discriminative ability.
Agreed. But if 20 trials is too little to yield adequate power for
discriminating small differences, hundreds of trials is not the
correct answer either, IMO. It's just too impractical, and there is a
big risk of listener fatigue (which is a real phenomenon, also IMO).
The alternative to lots of trials is lots of (sensitive) listeners.
Or a more efficient test. The triangle test (odd-man out), for
example, is more efficient, but may suffer from additional complexity.
>> One needs to be very careful about allowing feedback during the test.
>> See previous threads concerning "Sequential ABX Tests." Basically,
>> allowing too much feedback can bias the result in the listener's
>> favor. The safest method is to not allow any feedback concerning the
>> number of correct trials until the very end.
>
>Absolutely not!
My mistake. The problem with sequential tests only occurs if the
number of trials is not held fixed. What I caution against is
arbitrarily allowing the listener to stop the test to declare victory.
Darryl Miyaguchi
Nousaine
>On 21 Oct 2002 16:36:58 GMT, ne...@stevethepsycho.co.uk (Steve G)
>wrote:
>
>>pat...@dircon.co.uk (Stewart Pinkerton) wrote in message
>news:<aou0c5$acp$1...@bourbaki.localdomain>...
>>(...)
>>>
>>> The test protocol will be double-blind ABX, implemented in whatever
>>> way is both practical for the circumstances, and acceptable to all
>>> interested parties.
>>
>>Although this is reasonable, the incumbent might be better going for a
>>2-interval forced-choice design, since there's some evidence that this
>>leads to higher estimated sensitivity cf. ABX (e.g., [1], [2]). 2IFC
>>is also a simpler procedure for inexperienced listeners, assuming the
>>task is easily conceptualised as 2IFC.
>
>How many trials does one have to perform in a 2IFC to get the same
>type 1 error as 16/20 correct in ABX? Raw test sensitivity should be
>traded against the probability of tester fatigue. ABX is already
>pretty inefficient and takes a lot of effort and trials. 20 trials at
>full concentration is very tiring.
Only when the differences are subtle. If they're non-extant, 20 trials can be a
killer. But don't forget that you can get 20 trials with 2 sessions or 2
listeners too.
>>> The test will comprise 20 trials, the 'pass'
>>> criterion being correct identification in 16 out of the 20 trials.
>>
>>20 trials provides inadequate power for testing small - but better
>>than chance - 'genuine' performance levels. The hapless listener would
>>be better to push for >100 trials, if they could stand this.
>
>100 trials is far too many to be practical, IMO.
Actually that's not exactly true. In the two dozen DBT on amplifiers published
prior to 1990 the average number of trials was greater than 90.
But, when you have a real difference that is of an audio-nature (occurs all the
time on selected material) 16 trials will certainly suffice. With codecs or
other truly subtle differences sometimes fairly large numbers of trials may be
needed to determine the program material that elicits the expected response.
However, I can see
>the value (ideally) in a large number of trials. The test offer as it
>stands is too short to allow for a small type II error. So it would
>not be correct to say, if one failed to achieve 16/20, that a "no
>difference" result was proved to a high degree of confidence.
Don't forget that the besy way to insure a low Type 2 error is to have a
sensitive experiment; one with a truly audible effect that is contained with
all the program material. IMO an audio experiment that will not return a
positive result with 20 trials simply hasn't been well designed OR there is
nothing to hear.
I've conducted dozens of blind listening tests and have yet to find ONE that
eventually turned up a claimed effect in subsequent experiments. Also the
extra-large experiments conducted by Stereophile showed that a large number of
trials (3000+) doesn't guarantee positive results.
ALL their extra-large trial experiments produced null results, even the one
where a smaller experiment was later claimed to have been positive.
Another aspect that often goes unregonized in these experiments is that these
audio experiments have a built-in anti-type 1 error bias in that they are
specifically designed to provide conditions that highlight possible
differences. IOW they are designed to find conditions where differences will be
MOST likely to occur.
However, because these differences (say castanets on a codec) have been 'sought
out and isolated' for the experiment it might be possible that no user would
ever experience them in actual use, even though the device that failed a DBT
would forever be considered substandard.
Or, more pertinently to this discussion, an amplifier found to sound
'different' into a specific load might never see such a load in normally use
but would forever be considered to have a special (could be bad or good) sound
quality.
My point is that the honing of experimental conditions to highlight sensitivity
tends to ameliorate the statistical considerations when extrapolated to general
use.
OTOH, it has been gratifying to see experimenters moving toward 99% confidence
intervals and 2-tailed analyses.
>>The listener should also have trialwise accuracy feedback.
>
>One needs to be very careful about allowing feedback during the test.
>See previous threads concerning "Sequential ABX Tests." Basically,
>allowing too much feedback can bias the result in the listener's
>favor. The safest method is to not allow any feedback concerning the
>number of correct trials until the very end.
>
>Darryl Miyaguchi
Yes, the tendency for subjects to quit when they are mid-ship either through
discouragement or optimism is a problem. I try to partially deal with that by
paying subjects and asking them to agree to a minimum number of trials in
advance.
Dr Bukkake
>stacked against him, for any amount of money?
You miss the point completely. If anything, the person taking the test has the
odds well in his favor, as he would receive a payout 15 times greater than what
would be considered fair and equitable based on the laws of probability alone.
In this regard, Mr. Pinkerton is being quite generous, whatever you may think
about the rest of the requirements.
Alan Murphy
> On 21 Oct 2002 19:51:37 GMT, "Alan Murphy" <afm...@btinternet.com>
> wrote:
> >Perhaps $2500 isn't enough to tempt anybody.
> >
> >I believe I can hear distinct differences in cables and if you up the
> >ante to say $100,000 I'd be prepared to pursue your challenge.
>
> Excellent! I'll be happy to accommodate you, at even odds.
> Stewart
If you were really sure you'd give me 10 to 1 :-)
Alan
Howard Ferstler
> Indeed - if enough people took the test, eventually there'd likely be
> a positive result even with no perceived differences. However, for the
> 'subjectivist', I think the 'social cost' of failing the test will not
> be overcome. Prediction: No-one with a vested interest in the
> existence of cable differences will take the test. Also, if an
> individual could genuinely successfully discriminate, then this would
> be a fully repeatable event. (Though whether the winner would allow
> repeat testing is another matter entirely!)
It is a fact of life that exotic wires and cables sell precisely
because the information about them is so vague. Many enthusiasts
prefer to have the whole audio situation be vague, because it allows
them to speculate about the performance of new accessories (super
wires, support cones and stands, air bladders, wire wraps, green ink,
disc demagnetizers, disc spinners, disc treatments, etc.) and how the
sound has mysteriously improved because of the incorporation of those
accessories into a system. It is the "mysterious" factor that makes
the hobby so compelling to certain individuals, and which makes it
sometimes so profitable to certain companies.
Regarding the statistical hair splitting that has been going on, I
find it almost comic that people are arguing such small details when
enthusiasts who swear by wires claim that they can immediately hear
improved soundstaging, focus, bloom, detail, depth, etc., etc. when
they replace some old wires with some newer ones and do sighted
comparisons or no comparisons at all. Those guys do not debate
hair-splitting details. That individuals who are dealing with the DBT
approach can worry about nuances when wire believers are raving about
black-and-white improvements shows just how surreal this entire wire
situation has become.
The very fact that hair-splitting details have to be debated at all is
proof positive that the wire-belief system that is under discussion is
pretty bogus. You do not need no stinking statistical hair splitting
to see that.
> (...)
> > of 1000:1. A fifteen dollar entrance fee would, theoretically make your
> > scenario fair and equitable. Otherwise, what the hell, put me down for a
> > listen!
> With no belief system behind you, I suspect the exercise becomes a lot
> less interesting for most involved. Let's be honest - a lot of this
> malarkey is designed to tempt a cable-activist onto the guillotine,
> and get him to pull the lever himself!
There is money to be made in wires. Plenty of money. Any cable
activist, be he a true believer or an opportunist, would be crazy to
chance killing the golden goose. He may talk the talk when it comes to
comparing, but walking the walk is a different matter altogether.
The sad thing is that many reputable and smaller-scale product
manufacturers (speakers, amplifiers, CD players, etc.) are forced into
playing the silly wire game, because to not play would seriously harm
sales. This is because any potential customers do not see it as a
"game" at all. For them it is the essence of the hobby and it is a
very serious essence, indeed.
Howard Ferstler
Bruce Abrams
*snip*
> >In some audiophiles "dreams," there would be manufacturers who
> >PURPOSELY dream up cables that sound worse, i.e., deliberately
> >alter the sound.
>
> If I were in the audiophile wire business, I'd purposely design
> something that sounded different (not worse), invest in packaging
> and advertising and balance that against wholesale price and
> volume. I'd believe that the combination of packaging,
> marketing, "different" and price would sell enough product to make
> a profit. My fundamental bet would be that I could convince enough
> audiophiles that "Different + $1000 = Better"
You would then be no different than any of the current cable companies ;-)
> >For the rest, the nightmare is that some manufacturers REALLY do
> >this.
>
> Seems like a victimless crime. Yeah, something's being
> prostituted, but the customer is getting that audiophile rush on his
> dollar.
It's anything but a victimless crime, and there are three victims:
Victim #1 is the consumer who has his audio budget (which could be invested
in sonically meaningful items like speaker upgrades, room treatments, etc.)
significantly curtailed for having wasted money on mega-buck snake oil
cables.
Victim #2 is the innovative engineering based audio companies (ie. Meridian)
who see potential revenues siphoned off to mega-buck snake oil cable
companies.
Victim #3 is the audiophile consumer who doesn't reap the benefits of the
R&D dollars that the innovative engineering based audio companies would have
spent part of their revenues on.
Dr Bukkake
>
If this is true, we should consider renaming this newsgroup
rec.audio.high-roller
>Up the ante.
>
If a shot at $2500, with winnable odds and no entance fee, coupled with the
immense satisfaction and honor that would come with being pronounced poster
child of the cable-sound set, vanquisher of the infidels, and grand poobah of
RAHE, is insufficient enticement to take this test, then I will personally
throw in a set of beer mugs.
Stewart Pinkerton
wrote:
>There are electrical parameters that are difficult to express in RLC and
>these could be the elusive differences. The first to come to mind is
>dissipation factor of the dielectric, which is almost like a resistive lossy
>effect, except that it has some time dependence.
Sure there are, and to old 'test and measurement' engineers like me,
some of them can be an absolute nightmare! *However*, the proof of the
pudding is in the eating, and since we have *no* observations to
suggest that any such secondary effects are *audible*, then there is
*no* investigation to perform.
>My old experiments with capacitors, way back in the old days of Suffolk
>Audio, showed some 'extreme' non-linearities that had time constants of
>large fractions of a second. Once you have non-linear behavior, the usual
>Maxwell equation solutions come apart pretty quickly.
Observational results, however, do not come apart.
> As a reference, we
>measured dielectric dissipations with time constants on the order of 5
>milliseconds. Mylar tended to have about a 10 bit effective accuracy and
>the best, Teflon, was equivalent to about 14 bits.
Sure, and 'pedestal voltage' problems plagued many early CD players.
This has *never* been shown to have any correlation to 'cable sound',
since 'cable sound' has *never* ben shown to exist - beyond *gross*
LCR differences.
>Is an experiment measuring voltage on a capacitor as a function of injected
>charge valid for cables? Probably not, as the connecting wires we used to
>measure these effects were PVC, a fairly lossy material, and did not seem to
>affect the measurements. The entire network needs to be analyzed and the
>impact of dielectric non-linearities calculated.
Why?
It's trivially easy to *measure* quite substantial differences among
cables, but so far we have *zero* instances of any such non-LCR
differences being shown to be *audible*.
> I did these calculations a
>long time ago when trying to build my own crossover. I could not calculate
>any meaningful effects, but used better capacitors as a prophylactic measure
>as they were readily available and inexpensive.
No harm in playing it safe! I use 23AWG silver-clad OFHC solid-core
twisted pair with Teflon insulation for my interconnects, because it
has many *theoretical* advantages. It's also just standard MIL-spec
hookup wire, at a few cents a foot..............
BTW, a certain regular poster also uses the heavy stranded version of
that MIL-spec hookup wire, made up into bundles to reduce resistance
and inductance in good old EE101 style - but he sells it for *much*
more than a few bucks a foot, as an 'exotic' so-called 'High End'
speaker cable.
>Hope this does not muddy the waters.
It doesn't. Nice try, though! :-)
Mkuller
>>What I don't believe in is the use of blind A/B tests to *prove* subtle
>audible
>>differences exist between audio components.
>
>Very convenient, Mike. I'll take that as a 'no'. Strange how the good
>people at B&W, Revel, Mark Levinson et al, who *design* the 'high end'
>gear that the audiophiles buy, seem to find DBTs extremely useful for
>uncovering subtle - but *real* - sonic differences.
>
sighted listening and *only* DBTs in their design and testing work? Wow, what
a claim! Can you provide proof of that? If you can then it will prove without
a doubt that DBTs are superior.
Regards,
Mike
Mkuller
>For the guys on RAHE $2500 just ain't worth getting out of bed for.
>
He's right, you know. To you, Stewart and most of us, this is chump change.
Howard offered a whopping $1. Let's see the courage of your convictions
(assuming you have some of either). Put up some serious money and leave it in
escrow with a disinterested third party. Then I can assure you there will be
takers. Otherwise, your challenge is a joke.
Regards,
Mike
Harry Lavo
> On 22 Oct 2002 18:26:44 GMT, loisk...@aol.com (Loiskelly1) wrote:
>
> >> Employing the rather cumbersome binomial probability
> >>theorum-
> >>p(x)=n!/((n-x)!x!)*p^x*q^n-x
> >
> >You gots ta be kidding! Can anyone verify??? Elaborate???
>
> It's basic statistics, you'll find it in any stats primer.
>
> > >This represents an aproximate odds against of 168 to 1.
> >
> >All of a sudden I feel that someone may collect on this, given the number
of
> >folks in here.
>
> As has been pointed out to those attempting to 'cherry pick' the
> Greenhill panel results, it certainly means that if 170 people try it,
> it's even money that someone will succeed through random guessing.
>
> I am happy to run with those odds.
Then you'd lose your money because there were only 11 subjects in the
Greenhill tests, not 170. There were between 135 and 165 total trials among
the eleven subjects in each of the tests. There is a big difference.
> snip>
Stewart Pinkerton
Good man! I'll pay for the engraving!
Stewart Pinkerton
I made no such claim, and it's transparently dishonest of you to
suggest that I did. The point is that, for *subtle* differences such
as can occur when tweaking crossovers, they certainly *do* use ABX as
a check method. Like anyone else, they 'voice' the speaker under
sighted conditions, since this is a matter of *preference*, not
*difference*. Of course, you already know this.................
Stewart Pinkerton
wrote:
>"Stewart Pinkerton" <pat...@dircon.co.uk> wrote in message
>news:ap203e$7mc$1...@bourbaki.localdomain...
>> On 21 Oct 2002 19:51:37 GMT, "Alan Murphy" <afm...@btinternet.com>
>> wrote:
>> >Perhaps $2500 isn't enough to tempt anybody.
>> >
>> >I believe I can hear distinct differences in cables and if you up the
>> >ante to say $100,000 I'd be prepared to pursue your challenge.
>>
>> Excellent! I'll be happy to accommodate you, at even odds.
>> Stewart
>
>If you were really sure you'd give me 10 to 1 :-)
Hey, *I* am not the one making the claims about 'cable sound', now am
I? I'm perfectly happy to accept new *evidence*, which can then be
investigated as to cause.
If *you* believe that you really *can* hear 'distinct differences in
cables', then even odds should be a racing certainty for you, yes?
Tim S.
>That may indeed be true, since MIT shotgun is notoriously 'dull',
>perhaps due to extremely high inductance.
If there is no audible difference between cables, then exactly what do you
mean by "notoriously dull"? Seems to me you are describing an audible
difference. If you mean that cable differences cannot be reliably detected
by a DBT, that's another kettle of fish.
Tim S.
WWCASE
>>There are electrical parameters that are difficult to express in RLC and
>>these could be the elusive differences
>Stewart Pinkerton replied:
>*However*, the proof of the
>pudding is in the eating, and since we have *no* observations to
>suggest that any such secondary effects are *audible*, then there is
>*no* investigation to perform.
>
Perhaps not audible, which is what your test is about, but perhaps
elusive differences still exists. For example, two bowls of the pudding,
one laced with nicotine. I doubt one could spot the laced pudding
"in the eating" since it's difficult to sense the elusive nature of
nicotine or connect its effect to conscious recognition and expression,
and yet it is real and addictive. It seems if this test were applied to
the pudding, a "no tasteble difference" could be true, even though a
powerful difference was present. But nevertheless, your test could
debunk those who claimed to taste the difference.
I'm not saying wire does this, but only that music affects the state
of the brain, and as nicotine shows, some of those states are
difficult to capture and express consciously.
B. Case
Nousaine
Sure; but that's what blind testing is all about. To test for differences that
are related to sound quality differences, no matter how introduced, and not any
other causes.
Stewart Pinkerton
>>>"harrison" suggested:
>
>>>There are electrical parameters that are difficult to express in RLC and
>>>these could be the elusive differences
>
>>Stewart Pinkerton replied:
>
>>*However*, the proof of the
>>pudding is in the eating, and since we have *no* observations to
>>suggest that any such secondary effects are *audible*, then there is
>>*no* investigation to perform.
>>
>
>Perhaps not audible, which is what your test is about, but perhaps
>elusive differences still exists.
There many easily *measurable* differences among cables, but since
none of them are *audible*, who cares?
> For example, two bowls of the pudding,
>one laced with nicotine. I doubt one could spot the laced pudding
>"in the eating" since it's difficult to sense the elusive nature of
>nicotine or connect its effect to conscious recognition and expression,
>and yet it is real and addictive. It seems if this test were applied to
>the pudding, a "no tasteble difference" could be true, even though a
>powerful difference was present. But nevertheless, your test could
>debunk those who claimed to taste the difference.
And indeed, the test would be valid, since it is a *taste* test, and
is not concerned with the presence of drugs which do not affect the
taste. You have unfortunately stumbled right into a stunningly poor
analogy, since this is *exactly* what it done in some 'blind' wine
tasting trials, where taste-free colouring is added to white wines, so
that they cannot be distinguished by colour alone.
Similarly, differences which are not *audible* are irrelevant in a
hi-fi system.
>I'm not saying wire does this, but only that music affects the state
>of the brain, and as nicotine shows, some of those states are
>difficult to capture and express consciously.
If you're not saying that wire does this, what is your point?
Stewart Pinkerton
wrote:
>"Stewart Pinkerton" <pat...@dircon.co.uk> wrote in message
No Harry, I'd keep the money, because in the tests where
level-matching was not a *big* issue (Monster vs 16AWG zipcord), there
was only a random chance score. You really *must* learn that you can't
'cherry-pick' only those results which suit your prejudices.
Richard D Pierce
>>Stewart Pinkerton wrote:>
>
>>That may indeed be true, since MIT shotgun is notoriously 'dull',
>>perhaps due to extremely high inductance.
>
>If there is no audible difference between cables, then exactly what do you
>mean by "notoriously dull"?
He never said that there were NO audible differences between
cables, did he? I believe, all along (that is, for years), his
criteria was something along the lines of "competently
designed," which as explained on several occasions as meaning by
the lack of "intentional" or "deliberate" response altering
contrivances, i.e., filters.
>Seems to me you are describing an audible difference.
Uh, yeah: frequency response differences of sufficient magnitude
could certainly be audible.
Note then the requirement of matching level and frequency
response such that the differences remaining are due to the
"deeper" differences alledged by certain propnents and
manufacturers.
--
| Dick Pierce |
| Professional Audio Development |
| 1-781/826-4953 Voice and FAX |
| DPi...@world.std.com |
dangling entity
> For the guys on RAHE $2500 just ain't worth getting out of bed for.
>
> Up the ante.
>
> Alan
Aw crap! That is such an empty cop-out! Can that be any more
pompous? If this was a discussion of a car race between a Mustang
Cobra owner and a Porsche owner, everybody would be on the floor
laughing their collective a$$es off at the Porsche owner for such a
comment. Not enough money??? Where's your sense of sport?! The
money is just a formality to solidify a wager. Picking out a
"night-and-day" difference should be like taking candy from a baby.
At the very least, it would substantiate one's worthitude as a fellow
Golden Ear Audiophile (if they *can* consistently give a positive
test). People should be falling over themselves for a chance like
that.
...But it does give a view into the psyche of the "rich" audiophile.
Evidently, $2500 is piddle whether it is a stack of cash or cash blown
on a set of divined cables. It doesn't matter if it actually works or
not- just that you can spend that amount in a day and not give it a
second thought. That's what it is about, right? ...not the sound at
all.
Bob Blase
news:ap6pj...@enews3.newsguy.com...
>
> ...Put up some serious money and leave it in
> escrow with a disinterested third party. Then I can assure you there will
be
> takers. Otherwise, your challenge is a joke.
> Regards,
> Mike
>
That would be counter-productive. If we put up a LOT of money, all we would
be able to find out is that there are con artists out there who are good
enough to be able to set up a rigged test without the panel of "independent
observers" detecting it. Or that there are independent observers who are
willing to be bought off. We already knew both of those, no need to prove
it. We'd best stick to amounts that actually ARE chump-change to those kind
of people. In other words, not much more than we've already got. A few more
beer mugs, maybe :-)
Bob
Arny Krueger
news:ap45d...@enews3.newsguy.com...
> There are electrical parameters that are difficult to express in
RLC and
> these could be the elusive differences. The first to come to mind
is
> dissipation factor of the dielectric, which is almost like a
resistive lossy
> effect, except that it has some time dependence.
How many RC models of DF have I seen in my life? There is even one in
the Marsh/Jung article that gave rise to decades of misapprehensions.
http://www.capacitors.com/pickcap/pickcap.htm more specifically
http://www.capacitors.com/pickcap/pick3lg.jpg .
Have I seen dozens or R-C models of DF? 100's? I'll point yet another
one out from a highly expert source as a URL, later on in this post.
> My old experiments with capacitors, way back in the old days of
Suffolk
> Audio, showed some 'extreme' non-linearities that had time
constants of
> large fractions of a second. Once you have non-linear behavior,
the usual
> Maxwell equation solutions come apart pretty quickly. As a
reference, we
> measured dielectric dissipations with time constants on the order
of 5
> milliseconds. Mylar tended to have about a 10 bit effective
accuracy and
> the best, Teflon, was equivalent to about 14 bits.
In what context? Sample-and-hold circuits, loudspeakers or line-level
amplification circuits?
The problem with DF as a major determiner of audio performance is the
relatively low impedance of audio circuitry.
> Is an experiment measuring voltage on a capacitor as a function of
injected
> charge valid for cables? Probably not, as the connecting wires we
used to
> measure these effects were PVC, a fairly lossy material, and did
not seem to
> affect the measurements.
It's valid if you keep other circuit parameters in mind. Of course if
you do that, the sub-1-ohm world of speakers is highly antagonistic
to the multimeghom range of impedances associated with DF. After all,
where do the long time constants come if not from but high
impedances?
>The entire network needs to be analyzed and the
> impact of dielectric non-linearities calculated.
It's fair play to exclude nth-order effects like DF once you find out
that they are so unimportant.
> I did these calculations a
> long time ago when trying to build my own crossover. I could not
calculate
> any meaningful effects, but used better capacitors as a
prophylactic measure
> as they were readily available and inexpensive.
For tweeter passive crossovers, yes. For midrange speaker crossovers
maybe. For subwoofer passive crossovers, no way Jose'.
> Hope this does not muddy the waters.
Here's a good reference for un-muddying the waters:
Stewart Pinkerton
wrote:
Well, there's the rub. Being as I actually do know what I'm talking
about, I could win the prize myself, simply by cheating - or perhaps I
should say, by introducing a totaly unrealistic condition not catered
for in the rules.
OTOH, this would have nothing to do with improving the quality of my
audio system...........................
Stewart Pinkerton
wrote:
>In article <ap7gav$ktf$1...@bourbaki.localdomain>, Tim S. <t...@skene.org> wrote:
>>>Stewart Pinkerton wrote:>
>>
>>>That may indeed be true, since MIT shotgun is notoriously 'dull',
>>>perhaps due to extremely high inductance.
>>
>>If there is no audible difference between cables, then exactly what do you
>>mean by "notoriously dull"?
>
>He never said that there were NO audible differences between
>cables, did he? I believe, all along (that is, for years), his
>criteria was something along the lines of "competently
>designed," which as explained on several occasions as meaning by
>the lack of "intentional" or "deliberate" response altering
>contrivances, i.e., filters.
Correct, and readily identified in a test situation by the very basic
criterion of level-matching to +/- 0.1 dB from 2--20kHz at the speaker
terminals.
>>Seems to me you are describing an audible difference.
>
>Uh, yeah: frequency response differences of sufficient magnitude
>could certainly be audible.
>
>Note then the requirement of matching level and frequency
>response such that the differences remaining are due to the
>"deeper" differences alledged by certain propnents and
>manufacturers.
Quite so. And do we see anyone stepping up to claim a reasonably
substantial prize by proving their ability to distinguish $1,000 a
foot Kimber 'Black Pearl' from 50 cents a foot 12 AWG 'zipcord'?
No, we do not. What does that tell you?
dangling entity
gets to pick which sounds better (openess, imaging, transients,
spectral richness, accuracy?) between the choice of a certain exotic
cable and conventional zip cord, except the zip cord has an
intentional bump in signal gain of say .2 or .5 db? This is all DBT,
of course.
Stewart Pinkerton
You don't need DBT protocols once you have established *difference*,
single-blind is adequate for *preference*, indeed you could use
sighted testing for that, since preference covers a lot more than just
sound quality.
What you're describing is just a con game anyway, as practiced by such
as Musical Fidelity with the X-10D.
dangling entity
first, and then the zipcord is bumped up by the prescribed amount.
dangling entity
Well that wasn't what I was shooting for, really. Just bringing to
light an interesting side premise whereby if zip cord can be
demonstrated to sound "better" than exotic cable just with an
incremental difference in gain, that really puts things into an
interesting perspective. What seems more practical to achieve the
same benefit- spend extreme sums of money on exotic cable or turn the
volume up a notch?
TonyP
>
> > >Regarding speakers, the whole point is that since differences are
> > >*not* subtle, DBTs have no function.
> >
> > Nonsense. Tests have shown that sighted biases can have substantial affects on
> > speaker preferences even when large differences between the speakers in
> > question exist. So it is absurd to assert they would have no function. It has
> > been proven otherwise. Objectivists habitually attack subjectivists for sighted
> > tests of components that many objectivists claim make little or no difference.
> > Why? Because they have no way of knowing whether or not their preferences were
> > based on sighted biases or actual sound. Yet most objectivists fail to use DBTs
> > for auditioning the one component that they seem to believe matters most,
> > speakers. Due to the possible effects of sighted biases objectivists are in
> > effect doing the very thing they attack subjectivists of doing. Subjectivists
> > listen to all components sighted (most of the time) and choose their
> > preferences at the risk of sighted biases affecting their preferences for
> > reasons that have nothing to do with what they actually hear. Objectivists do
> > the very same thing (most of the time) with the only component that they
> > believe will make a sonic difference. I find that ironic.
That you for bringing this up and to the forefront. It seems that the DBT people have not answered your
statements and it speaks volumes. I have not performed DBT with equipment to measure exactly to level
match anything. I would sit down and listen with music I was familiar with. Then, I would make a
decision on what I would spend my money on. I believe Stewart Pinkerton, a believer in the DBT said
that the MIT Shotgun cables were known for being "dull" sounding without qualifying the methods that he
used to determined this.
Subject: Re: Does anyone REALLY believe in 'cable sound'?
Date: Tue, 22 Oct 2002 07:25:14 GMT
From: pat...@dircon.co.uk (Stewart Pinkerton)
Newsgroups: rec.audio.high-end
References: 1 , 2 , 3
On 21 Oct 2002 23:01:09 GMT, TonyP <api...@ix.netcom.com> wrote:
>>Although I would agree on the success of a lot of "hi end" cable people (how much
>>CAN wire really cost??), I know that I have heard differences in cables. Some were
>>very subtle. Some, I couldn't really tell the difference.
>>But I do remember comparing at the time, MIT shotgun everything to my Monster
>>Powerline 2 and Interlink Reference A. I had nothing to prove and wanted to know if
>>there was a difference in sound, since the price difference was vast. Well, my
>>buddy who bought the MIT stuff left my home a little sadden by the experience. At
>>least in my system (at the time, Acoustat 1+1's, Counterpoint SA 3 pre amp, SA 2
>>head amp, VPI/MMT/Koetsu analog), the MIT didn't "sound" as good as the Monster
>>stuff I had.
>That may indeed be true, since MIT shotgun is notoriously 'dull', perhaps due to extremely high
>inductance.
Compared to my Monster Interlink Reference and Powerline 2 cables, it just didn't "sound" as good.
Neither I, nor the person who bought the MIT stuff needed to level match anything to hear the
difference. If anything, the person who bought the MIT stuff would want to be able to demonstrate their
sonic "superiority" to my Monster stuff on my equipment in my home. It didn't matter what we listened to,
at any volume level, the MIT stuff just didn't "sound" as good to either of us. And, I was using what is
considered good audio equipment. But then, I don't have a bias. Either I like it or not. It doesn't
matter how much or little it costs. Unlike those that do have a bias... towards DBT as the only reliable
method of testing. I believe that it has a place, but it is not the "end all, be all".
cl
news:apd0vc$br0$1...@bourbaki.localdomain...
> >That may indeed be true, since MIT shotgun is notoriously 'dull', perhaps
due to extremely high
> >inductance.
>
> Compared to my Monster Interlink Reference and Powerline 2 cables, it just
didn't "sound" as good.
> Neither I, nor the person who bought the MIT stuff needed to level match
anything to hear the
> difference. If anything, the person who bought the MIT stuff would want to
be able to demonstrate their
> sonic "superiority" to my Monster stuff on my equipment in my home. It
didn't matter what we listened to,
> at any volume level, the MIT stuff just didn't "sound" as good to either
of us. And, I was using what is
> considered good audio equipment. But then, I don't have a bias. Either I
like it or not. It doesn't
> matter how much or little it costs. Unlike those that do have a bias...
towards DBT as the only reliable
> method of testing. I believe that it has a place, but it is not the "end
all, be all".
The normal qualifier they have used in the past here is that the MITs are a
composite device: cables + passive components.
So they are cables with an extraordinary cost that include a network that
truely only cost pennies on the dollar to implement.
Stewart Pinkerton
>> S888Wheel wrote:
>>
>> > >Regarding speakers, the whole point is that since differences are
>> > >*not* subtle, DBTs have no function.
>> >
>> > Nonsense. Tests have shown that sighted biases can have substantial affects on
>> > speaker preferences even when large differences between the speakers in
>> > question exist. So it is absurd to assert they would have no function. It has
>> > been proven otherwise. Objectivists habitually attack subjectivists for sighted
>> > tests of components that many objectivists claim make little or no difference.
>> > Why? Because they have no way of knowing whether or not their preferences were
>> > based on sighted biases or actual sound. Yet most objectivists fail to use DBTs
>> > for auditioning the one component that they seem to believe matters most,
>> > speakers. Due to the possible effects of sighted biases objectivists are in
>> > effect doing the very thing they attack subjectivists of doing. Subjectivists
>> > listen to all components sighted (most of the time) and choose their
>> > preferences at the risk of sighted biases affecting their preferences for
>> > reasons that have nothing to do with what they actually hear. Objectivists do
>> > the very same thing (most of the time) with the only component that they
>> > believe will make a sonic difference. I find that ironic.
It is of course not 'ironic' at all, since the whole point about DBTs
ais that they are used for revealing *subtle* differences. As such,
they are often used in *development* of an individual speaker, but
differences among finished models are so great that DBTs have no
sensible function. Also, note that a speaker will sound very different
in different rooms, and with different placement in the same room, so
DBT is simply *not* an appropriate technique for this component.
> That you for bringing this up and to the forefront. It seems that the DBT people have not answered your
>statements and it speaks volumes.
Actually, we most certainly *have* endeavoured to engage him in
rational debate, but he seems to have more interest in semantics. I am
cutrrently engaged in trying to get a reply to the above post past the
moderators, such is my concern with his 'debating trade' tactics.
> I have not performed DBT with equipment to measure exactly to level
>match anything.
Usually unnecessary with cables.
> I would sit down and listen with music I was familiar with. Then, I would make a
>decision on what I would spend my money on. I believe Stewart Pinkerton, a believer in the DBT said
>that the MIT Shotgun cables were known for being "dull" sounding without qualifying the methods that he
>used to determined this.
That is untrue, I suggested that excess inductance would cause the
notorious high-end droop of MIT cables. This would of course be easily
measured at the speaker terminals.
You, OTOH, have claimed that you did not need to employ DBT methods
for your MIT/Monster comparison as the differences were 'not subtle'.
Sorry, but that simply does not wash for *any* cable I have ever
encountered.
chung
Perhaps the "DBT people" have not bothered to respond because the answer
is obvious?
In a consumer test scenario (as opposed to R&D scenario) two speakers of
different makes/models are easily detectible in DBT's.
If you are doing a DBT to detect audible difference, it is in general a
big waste of time, especially given the physical difficulties involved,
unless you have two speakers that are extrememly closely matched. In
general the speakers will have noticeable differences to insure positive
DBT results. For example, different bass extensions, different responses
to pink noise, etc. (Remember that level differences of 0.3dB are
detectible via DBT.)
If you are doing a DBT to determine preference, it is the same as a
sighted test because the speakers are easily identifiable. The testee
knows which speaker is which.
Chung.
S888Wheel
>is obvious?
Some have responded. IMO the obvious answer is that the RAHE objectivist camp
is faced with yet another proverbial dead moose in the middle of their living
room.
>
>In a consumer test scenario (as opposed to R&D scenario) two speakers of
>different makes/models are easily detectible in DBT's.
>
Has anybody made any arguement that DBTs should be used to detect *differences*
in different models of speakers? I haven't.
>If you are doing a DBT to detect audible difference, it is in general a
>big waste of time, especially given the physical difficulties involved,
>unless you have two speakers that are extrememly closely matched. In
>general the speakers will have noticeable differences to insure positive
>DBT results. For example, different bass extensions, different responses
>to pink noise, etc. (Remember that level differences of 0.3dB are
>detectible via DBT.)
If this is in any way in refernce to anything I have posted on using DBTs to
eliminate sighted biases in choosing a *preference8 amoung different speakers
you have completely missed the point.
>
>If you are doing a DBT to determine preference, it is the same as a
>sighted test because the speakers are easily identifiable. The testee
>knows which speaker is which.
>
You are plainly wrong and are either unaware or are ignoring tests that have
shown that sighted biases affect prefernces between different models of
loudspeakers.
S888Wheel
I don't know why my quote from a different thread ended up here but some things
need to be addressed in the response by Mr. Pinkerton who seems bent on picking
and choosing the realities that suit his positions in audio.
>
>It is of course not 'ironic' at all,
It absolutely is ironic. Irony: incognuity between the actual results of a
sequence of events and the normal expected result. The sequence of events
being the ongoing insistance in using DBTs to eleiminate sighted biases from
the proccess of auditioning components in audio. The normal expected result
being the use of this favored test to eliminate sighted biases from the
proccess of auditioning the one component that the promoters of DBTs for the
most part agree makes the biggest if not the only sonic difference and should
be given the biggest chunk of the budget. and lastly the actual revents being
that few objectivists actually use their favorite test on that which they
believe matters most. it is truely ironic.
> since the whole point about DBTs
>ais that they are used for revealing *subtle* differences.
This is simply not true. Mr. Pinkerton is trying to personally write the rules
of use for all DBTs in audio. It is utter and complete nonsense. *The point* of
DBT is to eliminate the effects of sighted bias period! That he or many other
objectivists fail to utilize this test in auditioning speakers is not grounds
for trying to misreperesent the reality of the usefulness of DBTs in
eliminating sighted biases, even when it is inconvenient for them.
> As such,
>they are often used in *development* of an individual speaker, but
>differences among finished models are so great that DBTs have no
>sensible function.
Again complete nonsense that flies in the face of good research which has shown
that sighted biases affect *preferences* amoung speakers with *great sonic
differences.* Perhaps Mr. Pinkerton believes he is beyond the effects of
sighted biases when listening to speakers. But isn't this the very critisism
that objectivists including Mr. Pinkerton level against subjectivists and their
methods of auditioning components? Ironic.
>Also, note that a speaker will sound very different
>in different rooms, and with different placement in the same room, so
>DBT is simply *not* an appropriate technique for this component.
It certainly is appropriate if you wish to eliminate sighted biases from the
audition, something objectivists have often seemed obsessed with. It may not be
convenient but when subjectivists make this claim about DBTs in general the
objectivists dismiss this as a weak excuse, lazyness in the face of expensive
purchases. With all the effort put forth by objectivists to promote the use of
DBTs with amps and cables I fail to understand the choice not to make the extra
effort in their own most important purchase in audio
.>
>Actually, we most certainly *have* endeavoured to engage him in
>rational debate, but he seems to have more interest in semantics.
Nonsense. Cite one example of my using semantics as the object of the debate.
you call it semantics I call it clarity in communication. I have been
completely fair and honest in my debate tactics.
> I am
>cutrrently engaged in trying to get a reply to the above post past the
>moderators, such is my concern with his 'debating trade' tactics.
My tactics are fine. I support my claims, and my logic is sound. And unlike
this very post, I don't resort to attacks on tactics even when I think they are
suspect.
chung
>>If you are doing a DBT to determine preference, it is the same as a
>>sighted test because the speakers are easily identifiable. The testee
>>knows which speaker is which.
>>
>
> You are plainly wrong and are either unaware or are ignoring tests that have
> shown that sighted biases affect prefernces between different models of
> loudspeakers.
What is the purpose of the DBT if the speaker identities are known to
the testee during the test?
This is a different issue than sighted bias affecting preference. No one
is arguing against that. The point some of us are making, that you
apparently fail to address, is that if the differences are non-subtle,
DBT's will not remove sighted bias, since the testee will know which
pair of speakers is being listened to. The normal bias will still be in
effect, once the testee knows the identitly of the speakers.
Chung.
chung
>>
>>If you are doing a DBT to determine preference, it is the same as a
>>sighted test because the speakers are easily identifiable. The testee
>>knows which speaker is which.
>>
>
> You are plainly wrong and are either unaware or are ignoring tests that have
> shown that sighted biases affect prefernces between different models of
> loudspeakers.
>
Hmmm, there seems to be a point of confusion here. You are probably
assuming that prior to the DBT, the testee does not know which speakers
are being compared, and he/she has not listened to those speakers before.
What I have been assuming is that the testee knows which speakers are
being compared, and he/she has listened to those already (and may have
already formed a sunconscious bias). In a subsequent DBT (such as ABX),
the testee will have no trouble telling which is which, and therefore
the DBT does not help remove sighted bias.
I don't think people will go into a speaker DBT not knowing which is
being compared against which in a consumer scenario. Is someone going to
bring two sets of speakers to your listening room without not letting
you know what the speakers are, and then go through the difficulties by
conducting a DBT for you? What about long term DBT's? Perhaps that is
why DBT's have not be used to set initial preference?
Chung.
S888Wheel
>>>sighted test because the speakers are easily identifiable. The testee
>>>knows which speaker is which.
>>>
>
>> You are plainly wrong and are either unaware or are ignoring tests that
>have
>> shown that sighted biases affect prefernces between different models of
>> loudspeakers.
>>
>
>
>Hmmm, there seems to be a point of confusion here. You are probably
>assuming that prior to the DBT, the testee does not know which speakers
>are being compared, and he/she has not listened to those speakers before.
The test I did read about that put sighted biases to the test involved a group
of listeners doing sighted tests between two different models of speakers, one
being large and impressive looking and one being small and not so impressive
looking. In sighted tests about 80% of the listeners prefered the larger
speakers. When the *same subjects* repeated the test blind it was almost a
50/50 split. So it does appear that even with some emediate experience in the
same listening envirement (room and equipment) listeners may not find blind ID
of speaker quite so easy. Yes there are circustances inwhich the differences
are so great or the listeners are so familiar with the speakers that they may
be able to ID the different speakers on any muusic material which would render
the test a nonblind test in effect. One can't safely make such assumptions
about comparisons between legitimately competative speakers in light of the
above test even if the testees are familiar with the speakers being tested. As
the objectivists so often say it's easy to be confident about what you are
hearing when you can see what you are listening to.
TonyP
Then why, may I ask, would someone use a process in revealing "subtle" differences when in the end, it
wouldn't have made a difference? Seems like a waste of time to me, and if I were a manufacturer, a waste of
money.
> > That you for bringing this up and to the forefront. It seems that the DBT people have not answered your
> >statements and it speaks volumes.
>
> Actually, we most certainly *have* endeavoured to engage him in
> rational debate, but he seems to have more interest in semantics. I am
> cutrrently engaged in trying to get a reply to the above post past the
> moderators, such is my concern with his 'debating trade' tactics.
I hope to read it soon.
> > I have not performed DBT with equipment to measure exactly to level
> >match anything.
>
> Usually unnecessary with cables.
Accepted.
> > I would sit down and listen with music I was familiar with. Then, I would make a
> >decision on what I would spend my money on. I believe Stewart Pinkerton, a believer in the DBT said
> >that the MIT Shotgun cables were known for being "dull" sounding without qualifying the methods that he
> >used to determined this.
>
> That is untrue, I suggested that excess inductance would cause the
> notorious high-end droop of MIT cables. This would of course be easily
> measured at the speaker terminals.
And... since it is an "excess inductance" that would cause the "notorious high-end droop", then, it should be
easily heard audibly. And, that is, exactly what I said.
> You, OTOH, have claimed that you did not need to employ DBT methods
> for your MIT/Monster comparison as the differences were 'not subtle'.
> Sorry, but that simply does not wash for *any* cable I have ever
> encountered.
I suggest you re-read your statement on the "excess" and the "notorious high-end droop" again. And then try
and tell me how "subtle" the differences are. Either that, or your use of adjectives are not appropriate.
S888Wheel
>>>sighted test because the speakers are easily identifiable. The testee
>>>knows which speaker is which.
>
>
>> You are plainly wrong and are either unaware or are ignoring tests that
>have
>> shown that sighted biases affect prefernces between different models of
>> loudspeakers.
>
>What is the purpose of the DBT if the speaker identities are known to
>the testee during the test?
If the speaker's identity are known then it isn't a DBT is it?
>This is a different issue than sighted bias affecting preference. No one
>is arguing against that. The point some of us are making, that you
>apparently fail to address, is that if the differences are non-subtle,
>DBT's will not remove sighted bias, since the testee will know which
>pair of speakers is being listened to.
>The normal bias will still be in
>effect, once the testee knows the identitly of the speakers.
I have not failed to adress this assertion, I simply have pointed out the fact
that studies have suggested under blind conditions their preference will often
be different than under sighted conditions of the same comparison. This also
shows that blind ID of each speaker being compared might not always be so easy.
Yes if the speakers are both so different and so well known to the auditioner
as to be easily identifiable then you can't remove sighted bias with such a
test. You really don't know that is the case until you do the test though. And
as stated before this ID task under blind conditions might not be as easy as
one would presume.
TonyP
But, they are still cables. And as such, have to be regarded as cables and are
subject to the same examination as cables.
Again, I held no real bias. I did not own the MIT's. And if anything, because of
the rave reviews it had received, I would have thought that there would have
been, if any, a "better" sound from them than the Monsters that I had. Such was
not the case. It was not the case for me nor the person who had spend the
"extraordinary cost" to purchase them.
chung
> The test I did read about that put sighted biases to the test
> involved a group of listeners doing sighted tests between two
> different models of speakers, one being large and impressive looking
> and one being small and not so impressive looking. In sighted tests
> about 80% of the listeners prefered the larger speakers. When the
> *same subjects* repeated the test blind it was almost a 50/50 split.
> So it does appear that even with some emediate experience in the same
> listening envirement (room and equipment) listeners may not find
> blind ID of speaker quite so easy. Yes there are circustances inwhich
> the differences are so great or the listeners are so familiar with
> the speakers that they may be able to ID the different speakers on
> any muusic material which would render the test a nonblind test in
> effect. One can't safely make such assumptions about comparisons
> between legitimately competative speakers in light of the above test
> even if the testees are familiar with the speakers being tested. As
> the objectivists so often say it's easy to be confident about what
> you are hearing when you can see what you are listening to.
I will be very surprised to find an audiophile who cannot tell two
speakers of different models apart in an ABX-style DBT, unless that
person does not have control over what kind of stimulus is to be used.
Perhaps someone who does not listen to hi-fi much may have trouble
telling speakers apart, but in my experience the frequency responses,
dispersion characteristics, imaging, power handling characteristics,
etc. are easy to cue in to distinguish the two speakers, if I have my
choice of music or test signals, and if I have control of how loud I
want to play the music.
Chung.
S888Wheel
> > involved a group of listeners doing sighted tests between two
> > different models of speakers, one being large and impressive looking
> > and one being small and not so impressive looking. In sighted tests
> > about 80% of the listeners
>prefered the larger speakers. When the
> > *same subjects* repeated the test blind it was almost a 50/50 split.
> > So it does appear that even with some emediate experience in the same
> > listening envirement (room and equipment) listeners may not find
> > blind ID of speaker quite so easy. Yes there are circustances inwhich
> > the differences are so great or the listeners are so familiar with
> > the speakers that they may be able to ID the different speakers on
> > any muusic material which would render the test a nonblind test in
> > effect. One can't safely make
>such assumptions about comparisons
> > between legitimately competative speakers in light of the above test
> > even if the testees are familiar with the speakers being tested. As
> > the objectivists so often say it's easy to be confident about what
> > you are hearing when you can see what you are listening to.
>
>I will be very surprised to find an audiophile who cannot tell two
>speakers of different models apart in an ABX-style DBT, unless that
>person does not have control over what kind of stimulus is to be used.
>Perhaps someone who does not listen to hi-fi much may have trouble
Everything you say sounds pretty logical even though it is very much the same
sort of thing subjectivists claim they can do with other components.I hope the
parrallels are not lost on you. You may know how it feels to be on the other
side of the objectivist/subjectivist debate. You don't *know* until youv'e dont
the test do you? The evidence suggests that you *may* not be as good at IDing
speakers blindly as you believe. The 80/20 split reduced to a near 50/50 split
when the same two speakers were compared sighted and then blind by the same
panel strongly suggests that sighted biases profoundly affect preference. I
think it is hard to deny that this panel was at least somewhat familiar with
the speakers since the blind test followed the sighted test. It has been said
numerous times by objectivists that you can't just will away the effects of
sighted biases.
chung
>>I will be very surprised to find an audiophile who cannot tell two
>>speakers of different models apart in an ABX-style DBT, unless that
>>person does not have control over what kind of stimulus is to be used.
>>Perhaps someone who does not listen to hi-fi much may have trouble
>
> Everything you say sounds pretty logical even though it is very much the same
> sort of thing subjectivists claim they can do with other components.I hope the
> parrallels are not lost on you. You may know how it feels to be on the other
> side of the objectivist/subjectivist debate. You don't *know* until youv'e dont
> the test do you? The evidence suggests that you *may* not be as good at IDing
> speakers blindly as you believe. The 80/20 split reduced to a near 50/50 split
> when the same two speakers were compared sighted and then blind by the same
> panel strongly suggests that sighted biases profoundly affect preference. I
> think it is hard to deny that this panel was at least somewhat familiar with
> the speakers since the blind test followed the sighted test. It has been said
> numerous times by objectivists that you can't just will away the effects of
> sighted biases.
>
Here's the difference. You read about many people failing DBT's on
cables. I have yet to see any anudiophile failing DBT's on speakers of
different makes. Differences caused by cables are very subtle, if at all
real. Differences between different makes/models speakers are
measureable and not subtle, especially if one uses the ABX paradigm.
There is really no parallelism here.
What your anecdote fails to point out is whether the listeners have full
control of stimulus used, and whether it is a DBT/ABX.
Another difference: different speakers can be measured to be able to
cause audible differences (like frequency response differences in the dB
range, dispersion in the dB range, dispersion charecteristics at
crossovers, etc). Not true for cables, given our knowledge of
just-noticeable-differences.
Chung.
S888Wheel
And you probably won't even if such DBTs for speaker *differences* are ever
conducted. Here is the all important similarities. They both remove the effects
of sighted biases. this is something objectivists have championed.
>Differences caused by cables are very subtle, if at all
>real. Differences between different makes/models speakers are
>measureable and not subtle, especially if one uses the ABX paradigm.
>There is really no parallelism here.
subtle is a subjective evaluation. If they are the same there is nothing to
talk about, if they are different subtle or gross become a mtter of opinion. By
the way if the differences are audible they are also measurable so this does
not seperate anything from anything. The parrallel still stands. The issue of
the effects of sighted bias is the parrallel and a person's inability to hear
past or will away those biases is there in both cases.
>
>What your anecdote fails to point out is whether the listeners have full
>control of stimulus used, and whether it is a DBT/ABX.
>
First off it is not exactly an anecdote. this was a published test. Second it
doesn't matter that the test was not conducted ABX style. A DBT or SBT does not
have to be an ABX test to remove the effects of sighted biases which to me is
*the* issue.
>Another difference: different speakers can be measured to be able to
>cause audible differences (like frequency response differences in the dB
>range, dispersion in the dB range, dispersion charecteristics at
>crossovers, etc). Not true for cables, given our knowledge of
>just-noticeable-differences.
>
If cables sound different it is easy to show it in a simple null test.
Differences are ultimately measurable if they exist.
Why limit this valuable tool, blind testing, to ABX tests of things you already
suspect sound the same or so close that you wouldn't really care anyways?
L Mirabel
news:apf72s$1os$1...@bourbaki.localdomain...
'subtle" differences but object to testing speakers because the differences
are "gross" and it would be "too easy".
I'd like to get your mind off theory for the moment and put it down on
earth. The question for audiophiles is not- is ABX is a wonderful idea in
all kinds of fields?- but- does it *work* for comparing components?
It happens that so far all the published reports of component comparisons
involving a reasonable crosssection of audiophile-panelists had an identical
outcome in its proctors conclusions: "They all sound the same" (References
given ntimes before but gladly repeated on request).
As you know null ie. negative reports can not prove that the selected
testing method *works*. The failure to differentiate could be due to it. Or
else indeed all the cables, preamps, amps, cxdplayers and dacs are
indistiguishable. This is an "extraordinary claim" that needs strong
validation not in words but by an experiment.
*You* need some positive reports to vindicate ABX. In the current "How to
get a positive ABX test" thread and before I suggested a noncontroversial,
easy component choice for testing. You don't like speakers pick anything
as "gross" as you like. No takers so far- no suggestions either.
So this is where we are. What you call "subtle" can not be differentiated by
ABX. The rest is too "gross" for you to touch. ABX rules.
Incidentally, talking about speakers- and please do not catch onto this to
the exclusion oof the basic question- your "practical" objections to
testing them are piffle. No one proposes doing it in somebody's home. Quite
apart from the fact that were it done it would be only valid for the
individual concerened- we're all different. It has to be a panel.
Harman_Kardon have an elaborate speaker testing set-up. Univ. of Ottawa
extensive facilities are available for legitimate research. I think B&W use
something similar at Cambridge Univ.
You *assume* that speaker differences would be so obvious that everyone
would know them blinded and ABXed. Prove it using high-end speakers!.
Experiments have been known to surprise experimenters.
But let's not stick to speakers. Pick anything electrically comparable and
get an ABX positive
result by a proctored audiophile panel. Till you do or can quote such a
positive outcome test published
anywhere please do not lecture such as myself from a podium. I happen to
believe that the very idea of a "test" for comparing the musical performance
of components, with validity to anyone but the individual doing it, is
ludicrous. But I do not have to prove anything. You do.
Ludovic Mirabel
chung
>>
> S888Wheel asks why the "objectivists" like the difficult comparisons with
> 'subtle" differences but object to testing speakers because the differences
> are "gross" and it would be "too easy".
The so-called objectivists are interested in finding out if there truly
are audible differencs among certain components. Speaker differences are
known, and measureable, to be non-subtle, and therefore they don't waste
time DBT'ing speakers to determine audible differences. Also, from what
I can tell, they are less interested in using DBT's to determine
preference in the event that differences exist. Hence the lack of great
desire to DBT speakers, given the physical difficulties of such exercises.
I will not object to anyone doing a speaker DBT's. My recommendation is
that it is a waste of time if you are trying to determine if there are
audible differences. But don't let me get in the way. Do a few, and you
know what I mean.
> I'd like to get your mind off theory for the moment and put it down on
> earth. The question for audiophiles is not- is ABX is a wonderful idea in
> all kinds of fields?- but- does it *work* for comparing components?
You have your answer for that question firmly entrenched in your mind
already. There is nothing I can do to change your answer, and to save
everyone's time, I will not participate in any such debate with you.
Many have tried, and I have nothing more to add.
Chung.
Stewart Pinkerton
>Stewart Pinkerton wrote:
>> It is of course not 'ironic' at all, since the whole point about DBTs
>> ais that they are used for revealing *subtle* differences. As such,
>> they are often used in *development* of an individual speaker, but
>> differences among finished models are so great that DBTs have no
>> sensible function. Also, note that a speaker will sound very different
>> in different rooms, and with different placement in the same room, so
>> DBT is simply *not* an appropriate technique for this component.
>
>Then why, may I ask, would someone use a process in revealing "subtle" differences when in the end, it
>wouldn't have made a difference? Seems like a waste of time to me, and if I were a manufacturer, a waste of
>money.
You seem to be missing the point. If DBTs reveal that say a Dynaudio
D26 sounds exactly the same as the vastly more expensive Esotar, that
cheap electrolytic capacitors and ferrite cored coils sound just the
same as polyproylene caps and air-cored coils in the crossover, or
that a simple rectangular MDF box sounds the same as a complex
multi-curved and heavily braced enclosure made from laminated
hardwood, then you can save an absolute fortune in production costs,
at no sacrifice in sound quality.
>> You, OTOH, have claimed that you did not need to employ DBT methods
>> for your MIT/Monster comparison as the differences were 'not subtle'.
>> Sorry, but that simply does not wash for *any* cable I have ever
>> encountered.
>
>I suggest you re-read your statement on the "excess" and the "notorious high-end droop" again. And then try
>and tell me how "subtle" the differences are. Either that, or your use of adjectives are not appropriate.
Excess inductance is that which is higher than that of 'zipcord', and
if it causes even as much as a 2dB droop at 20kHz (certainly a 'gross'
difference by any standard), this will be audibly undetectable by many
listeners on many musical passages. That's what I call subtle, and
that's what *demands* DBT protocols if you are to be sure that a
*real* sonic difference exists.
Stewart Pinkerton
<elm...@pacificcoast.net> wrote:
>*You* need some positive reports to vindicate ABX. In the current "How to
>get a positive ABX test" thread and before I suggested a noncontroversial,
>easy component choice for testing. You don't like speakers pick anything
>as "gross" as you like. No takers so far- no suggestions either.
Ludovic, this is disingenuous at best. You have been offered *many*
examples of such positive ABX test outcomes, and you have chosen to
disregard them.
>You *assume* that speaker differences would be so obvious that everyone
>would know them blinded and ABXed. Prove it using high-end speakers!.
>Experiments have been known to surprise experimenters.
No need, since as you admit, B&W, Revel et al already do this, and
obtain positive results all the time.
>But let's not stick to speakers. Pick anything electrically comparable and
>get an ABX positive
>result by a proctored audiophile panel. Till you do or can quote such a
>positive outcome test published
>anywhere please do not lecture such as myself from a podium.
Ludovic, we will continue to 'lecture' you so long as you keep
introducing additional requirements which neatly exclude ABX tests
which don't give the results you want to read. This is not how science
works.
cl
> > The normal qualifier they have used in the past here is that the MITs
are a
> > composite device: cables + passive components.
> > So they are cables with an extraordinary cost that include a network
that
> > truely only cost pennies on the dollar to implement.
>
> But, they are still cables. And as such, have to be regarded as cables and
are
> subject to the same examination as cables.
> Again, I held no real bias. I did not own the MIT's. And if anything,
because of
> the rave reviews it had received, I would have thought that there would
have
> been, if any, a "better" sound from them than the Monsters that I had.
Such was
> not the case. It was not the case for me nor the person who had spend the
> "extraordinary cost" to purchase them.
>
Tony, I had a setup that was tremendously improved by even a cheap set of
MITs speaker cables. Of course a good designer would attribute the reason
to the fact that the Zobel was just accomplishing what the amp/xover
designer should have done in the first place. It isn't the esorteric
materials, however that is to be given the credit for the change.
If equipment matching were not a problem and i/o circuits were truely
designed with a compatibility standard, the cable industry wouldn't have a
footing to stand on. However, there are enough differences in crossbrand
compatibility that cable can introduce a dfference depending on how well the
two items mesh together in the first place. What this means is, you have to
find a setup that needs matching help in the first place before you can jump
thru the hoops to find a cable to fix the problem. Well designed equipment
should never reveal a difference in sound just by swapping cables that are
not 'actively' (thru passive components) altering the transfer
characteristics of the signal.
That said, I do run MITs. The layering was improved to a degree that I
would have had to spend over 3 times the cost of the cable + original
speaker cost to move up to a setup that could mimic the results I got. Now,
I could have spent $20-$50 and put the same network on the xovers myself and
just used zip cord. This is the point that needs to be gotten across, you
don't have to spend much to accomplish the same task. It isn't like the
makers of cables have a deal with the devil that allows them to use
chokes,caps and resistors outside of the bounds of physics like the rest of
us are governed by.
I however like the looks of the firehose sized cables and it adds to the
look of the system I'm still awaiting the day someone thinks they blew out
my speakers and I get to respond, no you just blew out the cables :)
_CAL
Howard Ferstler
I have compared some very good speakers that had *similar*
radiation patterns that sounded very, very similar with
musical source material. These were upscale models that also
generated averaged room-response curves that were also very
similar.
However, if I switched to pink noise the differences were
instantaneously apparent.
Interestingly, if the speakers had notably *different*
radiation patterns (wide dispersion vs controlled narrow
dispersion) the differences were easily audible with musical
source material, even when the systems both had very similar
room-response curves.
Howard Ferstler
chung
>
>>
>>Here's the difference. You read about many people failing DBT's on
>>cables. I have yet to see any anudiophile failing DBT's on speakers of
>>different makes.
>
> And you probably won't even if such DBTs for speaker *differences* are ever
> conducted. Here is the all important similarities. They both remove the effects
> of sighted biases. this is something objectivists have championed.
>
>>Differences caused by cables are very subtle, if at all
>>real. Differences between different makes/models speakers are
>>measureable and not subtle, especially if one uses the ABX paradigm.
>>There is really no parallelism here.
>
> subtle is a subjective evaluation. If they are the same there is nothing to
> talk about, if they are different subtle or gross become a mtter of opinion. By
> the way if the differences are audible they are also measurable so this does
> not seperate anything from anything. The parrallel still stands. The issue of
> the effects of sighted bias is the parrallel and a person's inability to hear
> past or will away those biases is there in both cases.
>
The point is that between speakers of different makes, the audible
difference exceeds JND, and is at least an order of magnitude greater
than between cables, according to measurements. That makes passing DBT
of speakers easy. In the case of cables, no one has given a scientific
theory as to why two similarly constructed cables with similar
measurements will sound different.
There is really no parallel here. I don't even know how you would ensure
level-matching of two speakers with differing frequency responses. Do
you level match with 1KHz, or with poink noise. These are some of the
practical difficulties involved that are not minor. However, no
objectivist is stopping you from doing a DBT of speakers, if you feel
you have to. I believe you said you did some kind of blind test between
two speakers. Were the differences clear to you?
>
> Why limit this valuable tool, blind testing, to ABX tests of things you already
> suspect sound the same or so close that you wouldn't really care anyways?
>
No one is limiting anyone from using DBT in speaker testing. Some of us
have given you examples of people doing exactly such, albeit in a R&D
environment. My point is that for consumers, it is a waste of time, and
that is why very few people need to do it. But no one here is going to
stop anyone else from doing a speaker DBT.
Chung.
Nousaine
This is exactly true. But if you are referencing the Toole/Olive tests I think
you should deliver a better background for those experiments. Those tests were
not "preference" based but asked subjects to rate performance categories on a
1-10 quality scale. When subjects, some of who might have been engineers of the
speakers involved, rated the devices under test under sighted conditions they
gave quite high numerical ratings to the larger tower speaker and lower ratings
to the 3-piece satellite/common woofer system. On the blind retest the ratings
still favored the larger speakers but the ratings difference narrowed
significantly.
The point about sighted and unconscious bias is well-taken. But it doesn't
address the real issue ...... blind tests of electronic and cable products are
easy to do in the present day. Buy a QSC Comparitor. The $600 cost is
chump-change for someone ready to spend $100/ meter on cables, especially
speaker cables.
But. aside from the several hundreds of thousands of dollars that Harman has
spent on a 3-second track speaker switching systems, it's very difficult to do
blind tests on speakers because of the same-location issue.
>>Here's the difference. You read about many people failing DBT's on
>>cables. I have yet to see any anudiophile failing DBT's on speakers of
>>different makes.
Actually you haven't seen any because none have been published.
>And you probably won't even if such DBTs for speaker *differences* are ever
>conducted.
They have and so far all of them have been conducted at Harman.
Here is the all important similarities. They both remove the
>effects
>of sighted biases. this is something objectivists have championed.
Well, all bias. With the ABX test and electronic devices it's easy to 'remove'
sighted bias with a good switching device and allow the subject to "see"
everything.
>>Differences caused by cables are very subtle, if at all
>>real.
Not "if" real. With nominally competent devices they are non-extant.
Differences between different makes/models speakers are
>>measureable and not subtle, especially if one uses the ABX paradigm.
That's generally true.
>>There is really no parallelism here.
>
>subtle is a subjective evaluation. If they are the same there is nothing to
>talk about, if they are different subtle or gross become a mtter of opinion.
>By
>the way if the differences are audible they are also measurable so this does
>not seperate anything from anything. The parrallel still stands. The issue of
>the effects of sighted bias is the parrallel and a person's inability to hear
>past or will away those biases is there in both cases.
This is true.
>>What your anecdote fails to point out is whether the listeners have full
>>control of stimulus used, and whether it is a DBT/ABX.
>>
>
>First off it is not exactly an anecdote. this was a published test.
Reference please.
Second it
>doesn't matter that the test was not conducted ABX style. A DBT or SBT does
>not
>have to be an ABX test to remove the effects of sighted biases which to me is
>*the* issue.
>
>>Another difference: different speakers can be measured to be able to
>>cause audible differences (like frequency response differences in the dB
>>range, dispersion in the dB range, dispersion charecteristics at
>>crossovers, etc). Not true for cables, given our knowledge of
>>just-noticeable-differences.
>>
>
>If cables sound different it is easy to show it in a simple null test.
>Differences are ultimately measurable if they exist.
That's very true. And they are easily 'measurable' with a DBT IF they exist.
>Why limit this valuable tool, blind testing, to ABX tests of things you
>already
>suspect sound the same or so close that you wouldn't really care anyways?
There's no limit. Single or Double Blind testing is the method of choice no
matter what the device being tested. The only limit with speakers is the
simultaneous location problem.
normanstrong
>
> I will be very surprised to find an audiophile who cannot tell two
> speakers of different models apart in an ABX-style DBT, unless that
> person does not have control over what kind of stimulus is to be used.
> Perhaps someone who does not listen to hi-fi much may have trouble
> telling speakers apart, but in my experience the frequency responses,
> dispersion characteristics, imaging, power handling characteristics,
> etc. are easy to cue in to distinguish the two speakers, if I have my
> choice of music or test signals, and if I have control of how loud I
> want to play the music.
I would also be surprised. However, it wouldn't surprise me at all to find
out that said audiophile was unable to characterize the sound quality of an
unknown speaker or to estimate with any degree of accuracy how much it sells
for. Not without seeing and knowing.
I've only conducted one blind speaker test, involving the NHT SuperOne and 2
other speakers of the same general type. The owner of the S1 never did
identify it correctly! And believe me, he had plenty of chances to hear it
during the open portion of the test.
Norm Strong
Steven Sullivan
DBTs to distinguish two makes of speakers will usually fall into the realm
of reinventing the wheel. I imagine it could
still serve some pedagogic purpose -- e.g., it would
be a trivial 'proof' that DBTs can 'detect' differences in components -- and it
might be required for *absolute* rigor in a paper submitted to a scientific
journal.
> If you are doing a DBT to determine preference, it is the same as a
> sighted test because the speakers are easily identifiable. The testee
> knows which speaker is which.
For preference decisions intended to be based *solely* on what's heard,
it could still be useful to 'hide' the speakers, since brand name and
appearance could conceivably affect preference.