Objectivist or Subjectivist? Give Me a Break

[JackD201]

I actually liked Frank too. I also like Tom by the way. The repetition however reminded me of Frankie.

 

[NorthStar]

---Which Tom, Jack?

By the way, are we objective or subjective right now?

 

[JackD201]

Tomelex of course..... and Both!

 

[Phelonious Ponk]

We're sub-optive.

 

[NorthStar]

Tomelex of course..... and Both!

Tomelex is a very nice guy. :b

Both? Tom & Elex? ...Alex?

_______________

You know, we have other Toms here (thumbs).

Seriously though, it's true. For example, Syntax is Tom.
...And there are more.

_______________
_______________

Ethan's latest post is very true; we do have a lot of information from "objective" measurements (lab tests). The biggest challenge, I also believe like him, is how to interpret them accurately.

Subjectively and scientifically and biologically, it is "impossible" to exactly duplicate/replicate a live music event from just a bunch of electronica, mechanical devices (loudspeakers), and a bunch of copper wires, or silver and with gold plated connectors and jacks (Jack, you Jack ), and whatever we got so far.
Besides, many recordings are dubs from several 'threads' recorded here and there, at different places (studios, homes, in different parts of the world). ...Who wants to reproduce that???

Realistically and truthfully we can certainly "enjoy" reproduced music for our sensorial pleasure of the ears, heart, and soul (emotions). ...Even the birds and many other animals of the Earth's kingdom love Classical music (piano, violin, cello, and all that jazz, like harp).

 

[Ethan Winer]

I could be wrong, Ethan, but I don't think a null test would address the issue, because I think some in this discussion are supposing that accurate reproduction of the input (recording) is not necessarily the goal.

I was addressing only the notion that "current measurements" are inadequate. If a null test shows all differences at 80+ dB down, then those difference are irrelevant whether people understand it or not. That's half the battle, just getting people to understand the unreliability of their own hearing. And if the null residual is louder than -80, then its audibility can be assessed based on understanding masking and Fletcher-Munson. It's not as complex as some would have us believe.

--Ethan

 

[Ethan Winer]

are null tests revealing differences between amplifiers that the stock measurements aren't?

Either method works fine. But with a null test, you get directly to whatever is different, versus having to measure a bunch of different things at 20 different volume levels.

Or are we back to the same ol', same ol' refrain - all (reasonably built amplifiers) sound the same & measure the same?

Yes, exactly. If you have any hard evidence to the contrary - real evidence, not anecdotes - now's the time to present it!

--Ethan

 

[DonH50]

Ethan, probably not here, but I am interested in how you achieve 80 dB nulls. That is way down there so I'd think you'd need careful phase as well as amplitude matching? All-pass?

 

[Groucho]

If a null test shows all differences at 80+ dB down, then those difference are irrelevant whether people understand it or not.

Hello Ethan

I am convinced. What do you have to do in order to get such a good null? I'm thinking along the lines of having to match the gain exactly, obviously, but are there other things such as frequency-dependent phase shifts that have to be compensated for? (that most people might think of as 'harmless' but would prevent you from getting a good null). And what interesting things have you found in such tests? Is feedback the terrible thing it is made out to be, for instance? Do amplifiers fall apart when driving real speakers as opposed to dummy loads? That sort of thing...

Edit: Apologies to DonH50! I refreshed the previous page and didn't see yours (on the next page) until I posted mine.

 

[jkeny]

Indeed, Ethan, as DonH & others have said here, your nulls are unbelievable if you are testing the full replay chain i.e you are testing the output from the speakers.

Of course, if you are just testing the signal that comes from the outputs of an amplifier then you haven't being reading the last number of pages about the interaction between amplifier & speaker.

 

[Groucho]

Indeed, Ethan, as DonH & others have said here, your nulls are unbelievable if you are testing the full replay chain i.e you are testing the output from the speakers.

Of course, if you are just testing the signal that comes from the outputs of an amplifier then you haven't being reading the last number of pages about the interaction between amplifier & speaker.

I did sort of describe how I would try to do this sort of thing in post #1175 - without realising that people had actually been able to do it for real. If we don't believe it's possible then we're presumably holding our hands up and admitting that there may be all sorts of mysterious gaps in our knowledge of what amplifiers do - which I'm not prepared to believe! I can believe that getting a null might be difficult, but I would be going for a software post-processing null so could apply phase shifts etc. as necessary (but surely having to apply something as simple as a phase shift is still potentially controversial with regards to its effect on the sound).

 

[jkeny]

I did sort of describe how I would try to do this sort of thing in post #1175 - without realising that people had actually been able to do it for real. If we don't believe it's possible then we're presumably holding our hands up and admitting that there may be all sorts of mysterious gaps in our knowledge of what amplifiers do - which I'm not prepared to believe! I can believe that getting a null might be difficult, but I would be going for a software post-processing null so could apply phase shifts etc. as necessary (but surely having to apply something as simple as a phase shift is still potentially controversial with regards to its effect on the sound).

We have to be careful in assuming that "people had actually been able to do it for real." without examining the test procedure & evidence of the results!
It's also interesting to see what a professional who uses null testing, Bob Carver, has to say about it http://www.stereophile.com/content/bob-carver-carving-name-himself-page-3

Atkinson: Because one of the things which came out of that whole business for me was how unstable the null was. If you breathed near the amplifiers, the depth of the null would change and the position of the null would change.

Carver: If you let the sunlight shine on an amplifier that's sitting there with a 70dB null, the null will go higher. Or deeper. Less null.

Atkinson: This is the crux of the matter...

Carver: A 70dB null is a very steep null. It's really down to the roots of the universe and things like that. 70dB nulls aren't possible to achieve in production.

Atkinson: What is your target null between the Silver Seven-t and the original Silver Seven?

Carver: About 36dB. When you play music, the null will typically hover around the 36dB area. So it's not a perfect null. No question about it.

Atkinson: It's 98.5% the same...

Carver: It's not a bad null.

So if Carver can't achieve -70dB " the roots of the universe" & "70dB nulls aren't possible to achieve in production." how does Ethan do it?

 

[Groucho]

We have to be careful in assuming that "people had actually been able to do it for real." without examining the test procedure & evidence of the results!
It's also interesting to see what a professional who uses null testing, Bob Carver, has to say about it http://www.stereophile.com/content/bob-carver-carving-name-himself-page-3



So if Carver can't achieve -70dB " the roots of the universe" & "70dB nulls aren't possible to achieve in production." how does Ethan do it?

Fascinating stuff, and I can't say I'm surprised at how difficult it is. But... it is sounding as though there's a huge gap that subjectivists can walk through. If the 'straight wire with gain' changes if the sun shines on it, then it's also going to change while it's playing music. And if it doesn't show up in the conventional tests, then we'll never know what it was doing to the music.

My suggestion would be to do it all in software and 'auto-null' i.e. allow the software to track the null in non-real time, telling us what it has to do to achieve it. If it's a tiny gain change as the temperature changes then maybe that's not too controversial. If it's a significant wobble after a sharp transient then that's more interesting.

 

[jkeny]

Fascinating stuff, and I can't say I'm surprised at how difficult it is. But... it is sounding as though there's a huge gap that subjectivists can walk through. If the 'straight wire with gain' changes if the sun shines on it, then it's also going to change while it's playing music. And if it doesn't show up in the conventional tests, then we'll never know what it was doing to the music.

My suggestion would be to do it all in software and 'auto-null' i.e. allow the software to track the null in non-real time, telling us what it has to do to achieve it. If it's a tiny gain change as the temperature changes then maybe that's not too controversial. If it's a significant wobble after a sharp transient then that's more interesting.

What you are talking about is already available in software product called AudioDiffmaker.
A series of datapoints using this software for null testing is also interesting http://www.gearslutz.com/board/gear-shoot-outs-sound-file-comparisons-audio-tests/607481-evaluating-ad-da-loops-means-audio-diffmaker.html

by means of Audio DiffMaker and tested also on the same audio sample three audio interfaces : RME Fireface 400, RME Babyface and Steinberg MR816X. Here below the 'correlated null depth' figures (both last ones) from Audio DiifMaker, the higher the figures in dB, the closer to the original the AD-DA loop:

Mytek 8 x 192
-1,002sec, 0,450dB (L), 0,553dB (R)..Corr Depth: 35,3 dB (L), 36,9 dB (R)

Orpheus
-1sec, 0,075dB (L), 0,064dB (R)..Corr Depth: 30,8 dB (L), 32,4 dB (R)

Fireface 400
-36,94usec, 0,298dB (L), 0,311dB (R)..Corr Depth: 30,2 dB (L), 31,8 dB (R)

MR816X
-2,101msec, 1,104dB (L), 0,605dB (R)..Corr Depth: 19,1 dB (L), 21,0 dB (R)

Babyface
1,555usec, 0,352dB (L), 0,414dB (R)..Corr Depth: 9,2 dB (L), 10,8 dB (R)

There are further null tests on this site which are also of interest. As you can see, again nowhere near 80dB nulls reported by Ethan. Why?

 

[rbbert]

Are we sure Ethan said he has actually measured 80 dB nulls, or just that if they were achieved it would mean that the amplifiers were sonically identical (and from what Bob Carver said many years ago, maybe only -70 dB nulls would be adequate for that conclusion)?? Furthermore, from what Carver says it sounds pretty unlikely that any two current production amps will null even to -40 dB under real world conditions.

 

[Groucho]

What you are talking about is already available in software product called AudioDiffmaker.

What a great discovery!

There are further null tests on this site which are also of interest. As you can see, again nowhere near 80dB nulls reported by Ethan. Why?

Indeed, you might have expected better. First thing I'm going to try is applying some synthetic differences between identical files and see if they're compensated for/ picked up.

Edit: and there's nothing to prevent averaging over many tests, I suppose, to eliminate random variations e.g. noise - while also noting that it was necessary, of course.

 

[jkeny]

Are we sure Ethan said he has actually measured 80 dB nulls, or just that if they were achieved it would mean that the amplifiers were sonically identical (and from what Bob Carver said many years ago, maybe only -70 dB nulls would be adequate for that conclusion)?? Furthermore, from what Carver says it sounds pretty unlikely that any two current production amps will null even to -40 dB under real world conditions.

Are you suggesting that Ethan is stating something which is unachievable in real-world testing? Are you suggesting that he is then using this hypothetical as a strawman argument "If a null test shows all differences at 80+ dB down, then those difference are irrelevant whether people understand it or not. That's half the battle, just getting people to understand the unreliability of their own hearing." ?

 

[DonH50]

Re-reading Ethan's post, he did not say he had measured 80 dB nulls. In fact, it could be read to imply nobody has, thus the contention that 80 dB null would be more than enough. The measuring gear may be able to do it, the components cannot. Ethan is quite capable of defending himself soI shan't go further except to add a couple of points of my own:

1. Achieving >80 dB nulls in the real world can be done but is a nightmare for both measurement instrumentation and components. I have never achieved that at the speaker, nor anywhere close. In fact, those nulls were in non-audio systems.

2. Two production amps of the same model compared should easily achieve 40 dB nulls (much higher, in fact) at their outputs. However, given the fact that most speakers exhibit >1% distortion, you wouldn't expect more than a 40 dB null at the speaker's output. The interesting null would be at the amplifiers' outputs for different amps driving the same speakers and same signal. I tried that ages ago but technical difficulties made it pretty impractical. Now, we could sample one, then the other, in separate test runs and generate the null from the sampled data. Would be much easier.

Onwards - Don

 

[Phelonious Ponk]

Are you suggesting that Ethan is stating something which is unachievable in real-world testing? Are you suggesting that he is then using this hypothetical as a strawman argument "If a null test shows all differences at 80+ dB down, then those difference are irrelevant whether people understand it or not. That's half the battle, just getting people to understand the unreliability of their own hearing." ?

Not to put too fine a point on it John, but while that would probably be a logical fallacy of some sort, it would not be a strawman argument....

A straw man, known in the UK as an Aunt Sally, is a type of argument and is an informal fallacy based on misrepresentation of an opponent's position.[1] To "attack a straw man" is to create the illusion of having refuted a proposition by replacing it with a superficially similar yet unequivalent proposition (the "straw man"), and refuting it, without ever having actually refuted the original position.[1][2]

...as he hasn't misrepresented anyone's position except, perhaps, his own. I can't wait to hear from Ethan on this one.

Tim

 

[jkeny]

Don, I think it is pretty clear what Ethan is claiming here & has claimed before in other threads http://www.whatsbestforum.com/showt...rom-each-other&p=101164&viewfull=1#post101164

If there really were some as-yet unknown aspect of audio fidelity, it would have been revealed 50+ years ago in a null test. The original Hewlett-Packard distortion analyzers used nulling, and more modern methods can null complete music tracks and determine what remains. So the notion that some audiophiles can hear aspects of fidelity that can't be measured is easily disproved. I've explained this so many times in the past, I don't understand why we're still discussing it.

http://www.whatsbestforum.com/showt...ive-Me-a-Break&p=138068&viewfull=1#post138068

Or are we back to the same ol', same ol' refrain - all (reasonably built amplifiers) sound the same & measure the same?

Yes, exactly.