General debate thread about audio measurements

[Thomas savage]

Member @Beherit has been issued with a reply ban, primarily for persisting to present a non constructive argument , foul language and lowering the tone.
I’m all for us being challenged but it must be more than just empty opinion being thrown about devoid of any actual merit.

 

[March Audio]

I'm done with this argument. Amir and company are just retired tech brahs with money to spend and seemingly have an agenda to push against Schiit when most of this stuff is just as bad or worse in real world use. Most of the major shit talkers here are posers with crap gear and rooms. Have fun with your measurement gear Amir and friends. Let me make it clear that is plainly obvious that hardly any of you have experience with great gear in great environments. You're just a bunch of poser internet blow hards. Soundcloud EDM DJs have more real world experience than you.

!

Good. Please retire to another forum that suits your own dogma.

 

[Thomas savage]

That is pretty mean Ray. After all it was your GayBeeEls that got to him. I have the GayBeeEl 305, but he specifically mentioned the 308s.

Seems if anyone has a good case of hearing good stuff next to the GayBeeEls it would be you with Krell and M-L in the same room. Of course I have the Soundlab ESLs. What is it about ESL lovers who also think the GayBeeEls are pretty okay? Neither of us have the Topping or the Schiit. What does it all mean?

Is bees being gay the reason there’s less of them about ...

 

[Wombat]

Is bees being gay the reason there’s less of them about ...

You know a drought is bad when the bees are being hand-fed.

We need to stop hand-feeding some Bbbbbbbbs. There is no audio drought, here.

 

[Thomas savage]

Reading through this thread I’m moved to apologise to our members for the amount of word salad that’s here.

 

[Palladium]

Is bees being gay the reason there’s less of them about ...

Ok that was pretty funny.

 

[garbulky]

I am firmly of the opinion that something sounds different because of the measured deviations. The idea that these distortions/additions are actually inaudible but the device sounds better for some magic as-yet-to-be-discovered reason is totally cockeyed IMO.
Audio is a mature science and there is not much genuinely new being brought out, either in the equipment or the understanding of the engineering. DSP for speakers and rooms is about it.
I was involved in R&D into the design of record players in the 1970s. I have seen nothing around today about record players that wasn't fully understood back then, and a lot of total BS/misunderstanding from enthusiasts which would exasperate those who were active back then since so much has been forgotten or not learned by a lot of present day practitioners.

We still can't accurately relate perception of audio to measurements in enough detail though we can give a general stab at it. So I venture to say that there's still a way to go in terms of it being a mature science.
For instance you say
"I am firmly of the opinion that something sounds different because of the measured deviations. "
I would say using these measurement deviation - which usually is that of a sine wave -, exactly what does this difference sound like to that reviewer when he is listening to a certain song. If one can't answer that in great detail, then I question the value of the statement.

 

[Purité Audio]

It has already been said Gar, if it sounds different, first you must ascertain that there is a difference and not imagination, then that difference can be measured and the difference explained.
Keith

 

[garbulky]

It has already been said Gar, if it sounds different, first you must ascertain that there is a difference and not imagination, then that difference can be measured and the difference explained.
Keith

But if the measurements are already different, can you say how this difference would manifest in a song you listen to?

 

[Frank Dernie]

Why would that matter?
if a system is non-linear it is difficult to predict how it will behave with a complex signal, if it is linear it is not. The non-linear system is defective, even if one likes the sound it adds.
I don't have a problem with people enjoying this sort of thing, after all I have 4 record players, my problem is with the idea that the distortion must be inaudible because I like the sound, so there must be something else we haven't discovered that explains my enjoyment. That is an emotional rather than logical outlook.

 

[mindbomb]

Let's be real here, when companies are designing dacs, their engineers are looking at a lot of the same measurements that we later see in reviews. They are using them to judge the performance of their products. Of course, in interviews and such there will be the obligatory stories about how they went through a bunch of iterations until they found one that sounded just right on their favorite song, but that's just a myth to appeal to the people who believe that designing these products is more an art than a science.

 

[Blumlein 88]

We still can't accurately relate perception of audio to measurements in enough detail though we can give a general stab at it. So I venture to say that there's still a way to go in terms of it being a mature science.
For instance you say
"I am firmly of the opinion that something sounds different because of the measured deviations. "
I would say using these measurement deviation - which usually is that of a sine wave -, exactly what does this difference sound like to that reviewer when he is listening to a certain song. If one can't answer that in great detail, then I question the value of the statement.

Suppose I am building a bridge. The most important thing is the bridge not break and fall down. Maybe the science about such things will allow an engineer to be very careful and build a bridge that just barely handles the max loads on the bridge. If we are skirting on the limits of it nearly breaking we might have need of analyzing how it will break. Some manners of breaking being more of a problem than others. Yes one might also consider the beauty of the bridge, but that isn't about design capabilities.

On the other hand, if materials technology and analyzing loads allows us to build the bridge so it is 10 times stronger than anything it will need to be we don't much care about the particulars of how it would break. We know it wouldn't break. Building in a 10 times safety factor is expensive and usually we'll have safety factors of 1.5 or 2x.

In much of audio we are at the point where building in 10x more accuracy than would ever be potentially audible is trivial in design and expense. Often more is spent on the fancy box the gear is housed in. So while some gear is also built that is rather audible there is little impetus to determine just how audible various inaccuracies may be. If you like it with a unique sound buy it and enjoy it. Just don't confuse the inaccuracy for superiority. Realize when knowledgeable people tell you the inaccuracy is what you are hearing and realize we don't have to have that if it isn't enjoyed by anyone.

The areas where we aren't at that point would mostly be transducers. Speakers being the most obvious thing.

 

[garbulky]

Suppose I am building a bridge. The most important thing is the bridge not break and fall down. Maybe the science about such things will allow an engineer to be very careful and build a bridge that just barely handles the max loads on the bridge. If we are skirting on the limits of it nearly breaking we might have need of analyzing how it will break. Some manners of breaking being more of a problem than others. Yes one might also consider the beauty of the bridge, but that isn't about design capabilities.

On the other hand, if materials technology and analyzing loads allows us to build the bridge so it is 10 times stronger than anything it will need to be we don't much care about the particulars of how it would break. We know it wouldn't break. Building in a 10 times safety factor is expensive and usually we'll have safety factors of 1.5 or 2x.

In much of audio we are at the point where building in 10x more accuracy than would ever be potentially audible is trivial in design and expense. Often more is spent on the fancy box the gear is housed in. So while some gear is also built that is rather audible there is little impetus to determine just how audible various inaccuracies may be. If you like it with a unique sound buy it and enjoy it. Just don't confuse the inaccuracy for superiority. Realize when knowledgeable people tell you the inaccuracy is what you are hearing and realize we don't have to have that if it isn't enjoyed by anyone.

The areas where we aren't at that point would mostly be transducers. Speakers being the most obvious thing.

But the innacuracies we are talking about is already buried way down where we can't really here it. I think this whole audibility of certain specs is completely exagerrated.

So if things do sound notably different in music*, then I doubt it's these things making these differences:
- A -90db snr vs a 120 db snr
- better accuracies of a -90db sine wave (seriously?!)
- 0.4 db channel imbalance of the left and right channel
- a roll off of 1-2 db at 18-20 khz
- THD of 0.3% vs a THD of 0.0001%
- A harmonic at -90 db vs a harmonic at -120 db
- 50 harmonics at -80 db vs one harmonic at -120 db

In some of these specs, it is true that some small degree of audibility may be there. But it's nowhere near the importance attached to them.
*Now how we would determine IF there actually is a (positive) difference is up for debate. I acknowledge that one can't prove to others there is a positive difference by performing non-level matched subjective listening. I'm not sure there is an accurate way to go about proving that yet... I know the forum loves their level matched dbt's, but I'm not sold on it.

 

[Sal1950]

I know the forum loves their level matched dbt's, but I'm not sold on it.

That's why your arguments hold no value here. You insist there are things being heard that can't be measured, while providing no verifiable evidence that these magic things are being heard or are just the delusions of believers in Santa Claus.
Provide such evidence and the associated reason will be revealed.

 

[garbulky]

That's why your arguments hold no value here. You insist there are things being heard that can't be measured, while providing no verifiable evidence that these magic things are being heard or are just the delusions of believers in Santa Claus.
Provide such evidence and the associated reason will be revealed.

You also can't prove that such positive differences don't exist...or prove that they are delusions.
I'm not trying to prove they exist. I'm trying to say the minute differences we are talking about in measurements can't describe how a unit sounds.
Either way, let's stop with the derisive tone. You want to have a conversation, don't be salty, please.

 

[Sal1950]

You also can't prove that such positive differences don't exist...or prove that they are delusions.
I'm not trying to prove they exist. I'm trying to say the minute differences we are talking about in measurements can't describe how a unit sounds.
Either way, let's stop with the derisive tone. You want to have a conversation, don't be salty. Just because you "know" you are right and I'm wrong doesn't give you the right to an attitude.

You pursue a circular argument that's becomes nauseating after about 20 years of this BS
If you can't provide blind evidence of things heard sighted, then it's just BS.
A falsehood repeated a million times will never become the truth no matter how hard you try.

 

[garbulky]

You pursue a circular argument that's becomes nauseating after about 20 years of this BS.

Sounds like a problem, but please leave me out of it. Take it up with all those other guys. You are not trying to have a conversation. You are just taking pot shots with a smug attitude.

If you can't provide blind evidence of things heard sighted, then it's just BS.

You think I'm arguing for sighted listening being scientifically valid? I think you aren't reading my posts and lumping me with a caricature you've created....

 

[Blumlein 88]

But the innacuracies we are talking about is already buried way down where we can't really here it. I think this whole audibility of certain specs is completely exagerrated.

So if things do sound notably different in music*, then I doubt it's these things making these differences:
- A -90db snr vs a 120 db snr
- better accuracies of a -90db sine wave (seriously?!)
- 0.4 db channel imbalance of the left and right channel
- a roll off of 1-2 db at 18-20 khz
- THD of 0.3% vs a THD of 0.0001%
- A harmonic at -90 db vs a harmonic at -120 db
- 50 harmonics at -80 db vs one harmonic at -120 db

In some of these specs, it is true that some small degree of audibility may be there. But it's nowhere near the importance attached to them.
*Now how we would determine IF there actually is a (positive) difference is up for debate. I acknowledge that one can't prove to others there is a positive difference by performing non-level matched subjective listening. I'm not sure there is an accurate way to go about proving that yet... I know the forum loves their level matched dbt's, but I'm not sold on it.

So okay, I am foggy on the point you wish to pursue or make.

I agree many small measurable differences have nothing to do with how two devices sound different. The fact plenty still say they sound different may or may not mean anything. Item one is good procedure to be able to tell if there is a difference. The gold standard is blind testing. There are other steps to take short of the gold standard.

So is your point that doing blind testing can miss perceptibly important differences? Is it that perceptibly important differences are not being measured? Again I am foggy about your position. I agree with some of it, but not conclusions maybe.

Is your point that some inaccuracies may enhance how good something sounds to people listening to music? That is possible and we know a few that some people prefer to accuracy.

How to determine if there is a difference comes before deciding if it is a positive difference.

One method is using the extensive knowledge about what the human hearing system is able to hear. What its limits are. Then measuring the resulting output of devices to see if they corrupt the input in ways that can be heard. A given speaker interacts with amplifier A to roll off the treble by a few decibels and interacts with amplifier B so that treble response is flat, we know why it would sound different. Both amplifiers interact with a speaker to have the same measured result or very nearly so and the results are no levels of distortion that should be audible and someone says they sound different? We need to determine if they really do. Blind level matched testing will do that unless you can give us examples of why it wouldn't. Sighted level matched listening is a good first step in cleaning up comparative listening. But we know the human being is very easily swayed by all kinds of information that will effect how they think something sounds.

 

[garbulky]

So okay, I am foggy on the point you wish to pursue or make.

I agree many small measurable differences have nothing to do with how two devices sound different. The fact plenty still say they sound different may or may not mean anything. Item one is good procedure to be able to tell if there is a difference. The gold standard is blind testing. There are other steps to take short of the gold standard.

So is your point that doing blind testing can miss perceptibly important differences? Is it that perceptibly important differences are not being measured? Again I am foggy about your position. I agree with some of it, but not conclusions maybe.

Is your point that some inaccuracies may enhance how good something sounds to people listening to music? That is possible and we know a few that some people prefer to accuracy.

How to determine if there is a difference comes before deciding if it is a positive difference.

One method is using the extensive knowledge about what the human hearing system is able to hear. What its limits are. Then measuring the resulting output of devices to see if they corrupt the input in ways that can be heard. A given speaker interacts with amplifier A to roll off the treble by a few decibels and interacts with amplifier B so that treble response is flat, we know why it would sound different. Both amplifiers interact with a speaker to have the same measured result or very nearly so and the results are no levels of distortion that should be audible and someone says they sound different? We need to determine if they really do. Blind level matched testing will do that unless you can give us examples of why it wouldn't. Sighted level matched listening is a good first step in cleaning up comparative listening. But we know the human being is very easily swayed by all kinds of information that will effect how they think something sounds.

My point didn't have much to do with blind testing. It's about using measurements as some sort of declarative tool for making large pronouncements when the small (measured) differences noted don't really have much to do with predicting how a unit would sound during music listening. But yet people trash gear on here and chortle at manufacturers based on these small differences. Some demand to know why anybody would choose any gear that produces even slightly worse than their favorite measurements even though they have little reflection on the sound.

 

[Blumlein 88]

My point didn't have much to do with blind testing. It's about using measurements as some sort of declarative tool for making large pronouncements when the small (measured) differences noted don't really have much to do with predicting how a unit would sound during music listening. But yet people trash gear on here and chortle at manufacturers based on these small differences. Some demand to know why anybody would choose any gear that produces even slightly worse than their favorite measurements even though they have little reflection on the sound.

Okay, I think I understand you better now.

I largely agree. As long as that isn't an excuse for over-priced sub-standard gear. Still room for appearance, or UI differences, or feature differences or plenty of things to differentiate beyond sound and make different prices for different gear make sense.

So where do we draw the line? Is that a question you have or a suggestion you'd like to make or just wish it were discussed more than the fact Dac A has .0002% distortion and Dac B is substandard because it has .03% distortion when in fact both will sound the same.