Food for Thought

heiney9

http://www.herronaudio.com/images/Measurements.pdf

Please discuss your thoughts on the article

H9

intangible

Moderately interesting. It would appear that tools and techniques to measure sound waves could stand to be improved, but I would doubt that there is any money there, or someone would have already done it.

Insofar as this article would apply to esoteric cables and power cords and such, which I assume is your point, we already have a device to determine whether they live up to the claims of their creators - double blind ABX testing.

heiney9

Nope and

Nope

What does money have anything to do with it?

Double blind ABX testing is too flawed.

analog97

Most interesting to me was Herron's claim of the documented ability to hear RIAA deviations as low as .03 db. This is the smallest difference I've ever "heard" of.....I can't dispute it, but must admit I am from Missouri on that claim....talk about "Golden Ears".

doctor r

The flaws in the blind test are inherent and we will never be able to overcome them. On the other side of that argument I think all the designers interviewed agreed that "listening" is the ultimate test mechanism. Each of us experience this within our own systems and label it synergy when it all works as our minds and ears expect that it should. We each carry an expectation as to how a system should sound, as does each of the designers who indirectly refered to their final design step as "listening" evaluations after the numbers have run their course. Perhaps what we really need to figure out is how to measure what those expectations are that our ears and minds are looking for and then designing to meet them would be a simpler challenge.
Rick

allstock

Some pretty heavy hitters in that article, who i'm sure have forgotton more than most of us will ever "know" about audio gear design. I agree that measurments don't tell the whole story, perhaps not even close. I trust my ears, even though some "experts" say our ears are the least reliable measurement of how something sounds, a theory I could never understand. I can not understand how when someone states a ic or pc or some other tweak changed the sound of thier system, how someone else can refute it...I KNOW, FIRST HAND, that ic's, pc's, synergy between componants, etc, make a difference. I have been into audio since I was 10,(32 years), and have never been afraid to experiment with gear to see if certain things make a difference or not. Some forums, who I don't need to name,( think "scientists"), are so against the possibility of things just sounding good, whether the measurements do or not, that discussions about it come down to "virtual" fist fights. It doesn't just apply to audio equipment though. I build hi-performance two-stroke motors(think snowmobile, outboard marine motors). Back in the day, when we raced mod-vp boats, a machinest we dealt with designed a "top hat" style cylinder head for a Mercury 2.4l. It did not make ONE MORE HORSEPOWER on the dyno, but it accelerated better out of the corners. In other words, we could not "prove" that the cylinder head design was superior, but the driver was sure it just "felt better" with the top hat design. It must have done something, we were very successful in the early nineties with that set-up.
Sorry to rant.:biggrin:

allstock

And another thing, double blind abx testing of audio componants is flawed on so many levels, I don't even consider it a "test".

On3s&Z3r0s

What specifically are the flaws inherent in double blind ABX tests?

allstock

On3s&Z3r0s wrote: »

What specifically are the flaws inherent in double blind ABX tests?

Sorry, I not taking that bait. Lots of discussion on this forum and others about it. Try the search function.

quadzilla

Can't speak for allstock, but for me, the primary flaws are the extremely short listening times on systems that the listener is unfamiliar with, and potentially, tracks they are also unfamiliar with. I don't see how someone, short of there being extremely dramatic differences, be expected to accurately a/b two or more systems when they can't know what they're listening for.

allstock

quadzilla wrote: »

Can't speak for allstock, but for me, the primary flaws are the extremely short listening times on systems that the listener is unfamiliar with, and potentially, tracks they are also unfamiliar with. I don't see how someone, short of there being extremely dramatic differences, be expected to accurately a/b two or more systems when they can't know what they're listening for.

Yup, you can speak for me on that, pretty much sums it up.

intangible

Money has everything to do with every persistent consumer product. If someone could make money by producing better measurement systems for sound waves, then it would be done. This article shows that the market has been there for at least a decade.

Non-double blind listening tests change multiple variables at once and are therefore scientifically inconclusive. End of story. If the engineers can manage to use them to design a better amp, then good for them. ABX tests can be done on familiar equipment with familiar tracks over long listening periods; the protocol itself does not require anything you listed. You can do one single blind in your living room if you can find someone to switch the cables for you. I don't understand the issue with the subject not knowing what to listen for; they listen for differences between the two samples.

As always, if buying expensive esoteric cables makes you happy, then, by all means, indulge yourself.

Tour2ma

allstock wrote: »

Sorry, I not taking that bait.

Coward... :biggrin:

... and neither should the pro-ABX crowd take h9's bait, i.e., this thread.

I think this sums the article and the discussion up...

Nelson Pass wrote:

I have seen pieces in Stereo Review and elsewhere stating outright that measurements have already adequately defined performance and that the subjectivists are fooling themselves. The opposite viewpoint is stated just as dogmatically. Both sides are emotional, and neither side is particularly reasonable.

quadzilla

I don't know... there's someone on this board, or maybe another, that had a sig for a long time that said something to the effect of "If it measures good but sounds bad, it is bad. If it measures bad but sounds good, you measured the wrong thing." This kind of sums it up for me. While I am an admitted subjectivist at heart, I do not discount measurements, either, and use them extensively when setting up gear. I just don't let something that measures well persuade me to believe it sounds good when my ears are hating it.

Non-double blind listening tests change multiple variables at once and are therefore scientifically inconclusive. End of story. If the engineers can manage to use them to design a better amp, then good for them. ABX tests can be done on familiar equipment with familiar tracks over long listening periods; the protocol itself does not require anything you listed. You can do one single blind in your living room if you can find someone to switch the cables for you. I don't understand the issue with the subject not knowing what to listen for; they listen for differences between the two samples.

While this is true as far as it goes, this is not how A/B/X testing is done in stereo evaluations by those that trumpet A/B/X as being the ultimate evaluation method. The tests are typically done in 30 second to 1 minute sessions, on unfamiliar eqiupment. This is the reason I stated that I don't consider the tests valid. Generally speaking, if I swap something out in my rig, I can tell immediately if there's a difference by listening to a few of my reference tracks, and swapping things back and forth myself. After than, the only question is whether it's a change I like and want to keep.

Further, it would seem that perhaps you didn't read the article, since one section must be quoted:

There is a very interesting AES paper presented by Jon Risch at the latest AES convention, in which he explains a system for measuring cable distortion using a TEF-20 and a PHI series of tones (AES Preprint #4803, “A New Class of In-Band Multitone Test Signals,” AES website, http://www.aes.org). It clearly shows differences between cable and cabling systems, such as bi-wiring. His system plots harmonic distortion in cable current transfer.

That would be George Cardas. But hey, he must be making it all up since he sells them fancy esoteric cables that A/B/X testing invariably says make no difference, right?

heiney9

I didn't post this to necessarily discuss A/B/X testing. Darqueknight has covered that topic already and it's very extensive. I thought the article was interesting to give one some perspective about measurements from some of the industry leaders.

I do know NP does uses measurements to get in the "ball park" but most of his stuff has to pass the listening test before it gets released for public consumption. I thought George Cardas's comments were telling as well.

H9s

analog97

The statement made by Nelson Pass hits home the most with me and my experience. It is simple and elegant:

My experience is that there is a reasonable correlation between sound and measurement for simple Class A circuitry with minimal or no
feedback. This relationship seems to disappear when the circuit becomes complex or has a lot of non-linearity corrected by feedback

heiney9

One of my favorite quotes:

We build amplifiers that sound good and measure reasonably well and don’t break. If you want to get a machine to listen to them for you, be my guest!

H9

heiney9

Another favorite from the article.

The personal computer has allowed great improvements in speaker measurements during the last 20 years. Although we now have much more information easily available to us, the improvements have not facilitated prediction of sonic character.

Extrapolating this trend, I don’t predict that measurements will replace listening
as an evaluative tool in the foreseeable future.

intangible

quadzilla wrote: »

While this is true as far as it goes, this is not how A/B/X testing is done in stereo evaluations by those that trumpet A/B/X as being the ultimate evaluation method. The tests are typically done in 30 second to 1 minute sessions, on unfamiliar eqiupment. This is the reason I stated that I don't consider the tests valid. Generally speaking, if I swap something out in my rig, I can tell immediately if there's a difference by listening to a few of my reference tracks, and swapping things back and forth myself. After than, the only question is whether it's a change I like and want to keep.

Then your issue is with the particular experiments you have seen using DB ABX tests, not with the method itself. If you want to do unscientific tests and take it on faith that the difference you hear is due to just one of them (the cable), then go for it. As I said before, make yourself happy; it doesn't matter to me how you spend your money.

That would be George Cardas. But hey, he must be making it all up since he sells them fancy esoteric cables that A/B/X testing invariably says make no difference, right?

I think it is telling that the Cardas site is full of scientific looking figures and numbers but lacks any sort of data reflecting the actual difference in sound the cables produce. I am also skeptical of anything he or another salesperson has to say on the matter, since they have a direct financial interest in you believing that their products work as advertised.

Of course, I might be wrong. I haven't done any sort of blind testing myself nor seen any peer reviewed experiments on the subject, so my skepticism comes entirely from the lack of data.


Anyway, I have given my thoughts on the article and will not clutter the topic further.

analog97

Extrapolating this trend, I don?t predict that measurements will replace listening as an evaluative tool in the foreseeable future.

Here is a quote that reaches the sub-basement of bottom lines!! I have always shed equipment that did not live up to my listening expectations. I could never opt for measurements as the sole criterion for equipment evaluation. I am my own final arbiter. I am confident that 95% of the Forum would agree. This must be a fundamental thread in the Polk Forum fabric.

Tour2ma

analog97 wrote: »

I have always shed equipment that did not live up to my listening expectations... I am confident that 95% of the Forum would agree. This must be a fundamental thread in the Polk Forum fabric.

Actually I think 100% would agree that we buy/ keep what we like to listen to... :rolleyes:

As for the fundamental thread bit... nah. This thread has the depth of a wading pool compared to many others here.

But it's early yet... :biggrin:

analog97

[QUOTEThis thread has the depth of a wading pool compared to many others here. But it's early yet... ][/QUOTE]


I was referring to the other kind of "thread", i.e the kind that binds together. :rolleyes:

soundfreak1

Mental masterbation--- sounds good-- or it don't! Each person has two ears and they all hear differtly!

thsmith

heiney9 wrote: »

Another favorite from the article.

The personal computer has allowed great improvements in speaker measurements during the last 20 years. Although we now have much more information easily available to us, the improvements have not facilitated prediction of sonic character.

Extrapolating this trend, I don?t predict that measurements will replace listening
as an evaluative tool in the foreseeable future.

What more needs to be said !

I know what sounds good to me and what does'nt.